Google Git
Sign in
ara-mdk / platform / external / llvm / b16f55f3c82fe72e2ca73f1071db5daaa2124e51 / . / docs / ProgrammersManual.html
blob: 66bca9b7757dc8dc122be4a12aa86ab434a098ec [file] [log] [blame]
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Programmer's Manual</title>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]6 <link rel="stylesheet" href="llvm.css" type="text/css">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]7</head>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]8<body>
9
10<div class="doc_title">
11 LLVM Programmer's Manual
12</div>
13
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]14<ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]15 <li><a href="#introduction">Introduction</a></li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]16 <li><a href="#general">General Information</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]17 <ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]18 <li><a href="#stl">The C++ Standard Template Library</a></li>
19<!--
20 <li>The <tt>-time-passes</tt> option</li>
21 <li>How to use the LLVM Makefile system</li>
22 <li>How to write a regression test</li>
Chris Lattner61db4652004-12-08 19:05:44 +0000[diff] [blame]23
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]24-->
Chris Lattner84b7f8d2003-08-01 22:20:59 +0000[diff] [blame]25 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]26 </li>
27 <li><a href="#apis">Important and useful LLVM APIs</a>
28 <ul>
29 <li><a href="#isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt>
30and <tt>dyn_cast&lt;&gt;</tt> templates</a> </li>
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]31 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]32option</a>
33 <ul>
34 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
35and the <tt>-debug-only</tt> option</a> </li>
36 </ul>
37 </li>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]38 <li><a href="#Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]39option</a></li>
40<!--
41 <li>The <tt>InstVisitor</tt> template
42 <li>The general graph API
43-->
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]44 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]45 </ul>
46 </li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]47 <li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
48 <ul>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]49 <li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
50 <ul>
51 <li><a href="#dss_fixedarrays">Fixed Size Arrays</a></li>
52 <li><a href="#dss_heaparrays">Heap Allocated Arrays</a></li>
53 <li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
54 <li><a href="#dss_vector">&lt;vector&gt;</a></li>
55 <li><a href="#dss_deque">&lt;deque&gt;</a></li>
56 <li><a href="#dss_list">&lt;list&gt;</a></li>
57 <li><a href="#dss_ilist">llvm/ADT/ilist</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]58 <li><a href="#dss_other">Other Sequential Container Options</a></li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]59 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]60 <li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
61 <ul>
62 <li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
63 <li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
64 <li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]65 <li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]66 <li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
67 <li><a href="#dss_set">&lt;set&gt;</a></li>
68 <li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]69 <li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
70 <li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]71 </ul></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]72 <li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
73 <ul>
74 <li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]75 <li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]76 <li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
77 <li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
78 <li><a href="#dss_map">&lt;map&gt;</a></li>
79 <li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
80 </ul></li>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]81 <li><a href="#ds_bit">BitVector-like containers</a>
82 <ul>
83 <li><a href="#dss_bitvector">A dense bitvector</a></li>
84 <li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
85 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]86 </ul>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]87 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]88 <li><a href="#common">Helpful Hints for Common Operations</a>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]89 <ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]90 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
91 <ul>
92 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
93in a <tt>Function</tt></a> </li>
94 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
95in a <tt>BasicBlock</tt></a> </li>
96 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
97in a <tt>Function</tt></a> </li>
98 <li><a href="#iterate_convert">Turning an iterator into a
99class pointer</a> </li>
100 <li><a href="#iterate_complex">Finding call sites: a more
101complex example</a> </li>
102 <li><a href="#calls_and_invokes">Treating calls and invokes
103the same way</a> </li>
104 <li><a href="#iterate_chains">Iterating over def-use &amp;
105use-def chains</a> </li>
106 </ul>
107 </li>
108 <li><a href="#simplechanges">Making simple changes</a>
109 <ul>
110 <li><a href="#schanges_creating">Creating and inserting new
111 <tt>Instruction</tt>s</a> </li>
112 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
113 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
114with another <tt>Value</tt></a> </li>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]115 <li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]116 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]117 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]118<!--
119 <li>Working with the Control Flow Graph
120 <ul>
121 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
122 <li>
123 <li>
124 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]125-->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]126 </ul>
127 </li>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]128
129 <li><a href="#advanced">Advanced Topics</a>
130 <ul>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]131 <li><a href="#TypeResolve">LLVM Type Resolution</a>
132 <ul>
133 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
134 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
135 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
136 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
137 </ul></li>
138
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]139 <li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> and <tt>TypeSymbolTable</tt> classes </a></li>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]140 </ul></li>
141
Joel Stanley9b96c442002-09-06 21:55:13 +0000[diff] [blame]142 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]143 <ul>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]144 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]145 <li><a href="#Module">The <tt>Module</tt> class</a></li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]146 <li><a href="#Value">The <tt>Value</tt> class</a>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]147 <ul>
148 <li><a href="#User">The <tt>User</tt> class</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]149 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]150 <li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
151 <li><a href="#Constant">The <tt>Constant</tt> class</a>
152 <ul>
153 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]154 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]155 <li><a href="#Function">The <tt>Function</tt> class</a></li>
156 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
157 </ul>
158 </li>
159 </ul>
160 </li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]161 </ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]162 </li>
163 <li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
164 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
165 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]166 </li>
167 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]168 </li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]169</ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]170
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]171<div class="doc_author">
172 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
Chris Lattner94c43592004-05-26 16:52:55 +0000[diff] [blame]173 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
174 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>, and
175 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]176</div>
177
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]178<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]179<div class="doc_section">
180 <a name="introduction">Introduction </a>
181</div>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]182<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]183
184<div class="doc_text">
185
186<p>This document is meant to highlight some of the important classes and
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]187interfaces available in the LLVM source-base. This manual is not
188intended to explain what LLVM is, how it works, and what LLVM code looks
189like. It assumes that you know the basics of LLVM and are interested
190in writing transformations or otherwise analyzing or manipulating the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]191code.</p>
192
193<p>This document should get you oriented so that you can find your
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]194way in the continuously growing source code that makes up the LLVM
195infrastructure. Note that this manual is not intended to serve as a
196replacement for reading the source code, so if you think there should be
197a method in one of these classes to do something, but it's not listed,
198check the source. Links to the <a href="/doxygen/">doxygen</a> sources
199are provided to make this as easy as possible.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]200
201<p>The first section of this document describes general information that is
202useful to know when working in the LLVM infrastructure, and the second describes
203the Core LLVM classes. In the future this manual will be extended with
204information describing how to use extension libraries, such as dominator
205information, CFG traversal routines, and useful utilities like the <tt><a
206href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
207
208</div>
209
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]210<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]211<div class="doc_section">
212 <a name="general">General Information</a>
213</div>
214<!-- *********************************************************************** -->
215
216<div class="doc_text">
217
218<p>This section contains general information that is useful if you are working
219in the LLVM source-base, but that isn't specific to any particular API.</p>
220
221</div>
222
223<!-- ======================================================================= -->
224<div class="doc_subsection">
225 <a name="stl">The C++ Standard Template Library</a>
226</div>
227
228<div class="doc_text">
229
230<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]231perhaps much more than you are used to, or have seen before. Because of
232this, you might want to do a little background reading in the
233techniques used and capabilities of the library. There are many good
234pages that discuss the STL, and several books on the subject that you
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]235can get, so it will not be discussed in this document.</p>
236
237<p>Here are some useful links:</p>
238
239<ol>
240
241<li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
242reference</a> - an excellent reference for the STL and other parts of the
243standard C++ library.</li>
244
245<li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
Tanya Lattner09cf73c2004-06-22 04:24:55 +0000[diff] [blame]246O'Reilly book in the making. It has a decent
247Standard Library
248Reference that rivals Dinkumware's, and is unfortunately no longer free since the book has been
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]249published.</li>
250
251<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
252Questions</a></li>
253
254<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
255Contains a useful <a
256href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
257STL</a>.</li>
258
259<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
260Page</a></li>
261
Tanya Lattner79445ba2004-12-08 18:34:56 +0000[diff] [blame]262<li><a href="http://64.78.49.204/">
Reid Spencer096603a2004-05-26 08:41:35 +0000[diff] [blame]263Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
264the book).</a></li>
265
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]266</ol>
267
268<p>You are also encouraged to take a look at the <a
269href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
270to write maintainable code more than where to put your curly braces.</p>
271
272</div>
273
274<!-- ======================================================================= -->
275<div class="doc_subsection">
276 <a name="stl">Other useful references</a>
277</div>
278
279<div class="doc_text">
280
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]281<ol>
282<li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]283Branch and Tag Primer</a></li>
Misha Brukmana0f71e42004-06-18 18:39:00 +0000[diff] [blame]284<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
285static and shared libraries across platforms</a></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]286</ol>
287
288</div>
289
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]290<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]291<div class="doc_section">
292 <a name="apis">Important and useful LLVM APIs</a>
293</div>
294<!-- *********************************************************************** -->
295
296<div class="doc_text">
297
298<p>Here we highlight some LLVM APIs that are generally useful and good to
299know about when writing transformations.</p>
300
301</div>
302
303<!-- ======================================================================= -->
304<div class="doc_subsection">
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]305 <a name="isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt> and
306 <tt>dyn_cast&lt;&gt;</tt> templates</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]307</div>
308
309<div class="doc_text">
310
311<p>The LLVM source-base makes extensive use of a custom form of RTTI.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]312These templates have many similarities to the C++ <tt>dynamic_cast&lt;&gt;</tt>
313operator, but they don't have some drawbacks (primarily stemming from
314the fact that <tt>dynamic_cast&lt;&gt;</tt> only works on classes that
315have a v-table). Because they are used so often, you must know what they
316do and how they work. All of these templates are defined in the <a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]317 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]318file (note that you very rarely have to include this file directly).</p>
319
320<dl>
321 <dt><tt>isa&lt;&gt;</tt>: </dt>
322
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]323 <dd><p>The <tt>isa&lt;&gt;</tt> operator works exactly like the Java
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]324 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
325 a reference or pointer points to an instance of the specified class. This can
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]326 be very useful for constraint checking of various sorts (example below).</p>
327 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]328
329 <dt><tt>cast&lt;&gt;</tt>: </dt>
330
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]331 <dd><p>The <tt>cast&lt;&gt;</tt> operator is a "checked cast" operation. It
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]332 converts a pointer or reference from a base class to a derived cast, causing
333 an assertion failure if it is not really an instance of the right type. This
334 should be used in cases where you have some information that makes you believe
335 that something is of the right type. An example of the <tt>isa&lt;&gt;</tt>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]336 and <tt>cast&lt;&gt;</tt> template is:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]337
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]338<div class="doc_code">
339<pre>
340static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
341 if (isa&lt;<a href="#Constant">Constant</a>&gt;(V) || isa&lt;<a href="#Argument">Argument</a>&gt;(V) || isa&lt;<a href="#GlobalValue">GlobalValue</a>&gt;(V))
342 return true;
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]343
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]344 // <i>Otherwise, it must be an instruction...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]345 return !L-&gt;contains(cast&lt;<a href="#Instruction">Instruction</a>&gt;(V)-&gt;getParent());
346}
347</pre>
348</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]349
350 <p>Note that you should <b>not</b> use an <tt>isa&lt;&gt;</tt> test followed
351 by a <tt>cast&lt;&gt;</tt>, for that use the <tt>dyn_cast&lt;&gt;</tt>
352 operator.</p>
353
354 </dd>
355
356 <dt><tt>dyn_cast&lt;&gt;</tt>:</dt>
357
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]358 <dd><p>The <tt>dyn_cast&lt;&gt;</tt> operator is a "checking cast" operation.
359 It checks to see if the operand is of the specified type, and if so, returns a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]360 pointer to it (this operator does not work with references). If the operand is
361 not of the correct type, a null pointer is returned. Thus, this works very
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]362 much like the <tt>dynamic_cast&lt;&gt;</tt> operator in C++, and should be
363 used in the same circumstances. Typically, the <tt>dyn_cast&lt;&gt;</tt>
364 operator is used in an <tt>if</tt> statement or some other flow control
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]365 statement like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]366
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]367<div class="doc_code">
368<pre>
369if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast&lt;<a href="#AllocationInst">AllocationInst</a>&gt;(Val)) {
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]370 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]371}
372</pre>
373</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]374
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]375 <p>This form of the <tt>if</tt> statement effectively combines together a call
376 to <tt>isa&lt;&gt;</tt> and a call to <tt>cast&lt;&gt;</tt> into one
377 statement, which is very convenient.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]378
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]379 <p>Note that the <tt>dyn_cast&lt;&gt;</tt> operator, like C++'s
380 <tt>dynamic_cast&lt;&gt;</tt> or Java's <tt>instanceof</tt> operator, can be
381 abused. In particular, you should not use big chained <tt>if/then/else</tt>
382 blocks to check for lots of different variants of classes. If you find
383 yourself wanting to do this, it is much cleaner and more efficient to use the
384 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]385
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]386 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]387
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]388 <dt><tt>cast_or_null&lt;&gt;</tt>: </dt>
389
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]390 <dd><p>The <tt>cast_or_null&lt;&gt;</tt> operator works just like the
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]391 <tt>cast&lt;&gt;</tt> operator, except that it allows for a null pointer as an
392 argument (which it then propagates). This can sometimes be useful, allowing
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]393 you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]394
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]395 <dt><tt>dyn_cast_or_null&lt;&gt;</tt>: </dt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]396
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]397 <dd><p>The <tt>dyn_cast_or_null&lt;&gt;</tt> operator works just like the
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]398 <tt>dyn_cast&lt;&gt;</tt> operator, except that it allows for a null pointer
399 as an argument (which it then propagates). This can sometimes be useful,
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]400 allowing you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]401
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]402</dl>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]403
404<p>These five templates can be used with any classes, whether they have a
405v-table or not. To add support for these templates, you simply need to add
406<tt>classof</tt> static methods to the class you are interested casting
407to. Describing this is currently outside the scope of this document, but there
408are lots of examples in the LLVM source base.</p>
409
410</div>
411
412<!-- ======================================================================= -->
413<div class="doc_subsection">
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]414 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]415</div>
416
417<div class="doc_text">
418
419<p>Often when working on your pass you will put a bunch of debugging printouts
420and other code into your pass. After you get it working, you want to remove
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]421it, but you may need it again in the future (to work out new bugs that you run
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]422across).</p>
423
424<p> Naturally, because of this, you don't want to delete the debug printouts,
425but you don't want them to always be noisy. A standard compromise is to comment
426them out, allowing you to enable them if you need them in the future.</p>
427
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]428<p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]429file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
430this problem. Basically, you can put arbitrary code into the argument of the
431<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
432tool) is run with the '<tt>-debug</tt>' command line argument:</p>
433
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]434<div class="doc_code">
435<pre>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]436DOUT &lt;&lt; "I am here!\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]437</pre>
438</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]439
440<p>Then you can run your pass like this:</p>
441
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]442<div class="doc_code">
443<pre>
444$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]445<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]446$ opt &lt; a.bc &gt; /dev/null -mypass -debug
447I am here!
448</pre>
449</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]450
451<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
452to not have to create "yet another" command line option for the debug output for
453your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
454so they do not cause a performance impact at all (for the same reason, they
455should also not contain side-effects!).</p>
456
457<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
458enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
459"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
460program hasn't been started yet, you can always just run it with
461<tt>-debug</tt>.</p>
462
463</div>
464
465<!-- _______________________________________________________________________ -->
466<div class="doc_subsubsection">
Chris Lattnerc9151082005-04-26 22:57:07 +0000[diff] [blame]467 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]468 the <tt>-debug-only</tt> option</a>
469</div>
470
471<div class="doc_text">
472
473<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
474just turns on <b>too much</b> information (such as when working on the code
475generator). If you want to enable debug information with more fine-grained
476control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
477option as follows:</p>
478
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]479<div class="doc_code">
480<pre>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]481DOUT &lt;&lt; "No debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]482#undef DEBUG_TYPE
483#define DEBUG_TYPE "foo"
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]484DOUT &lt;&lt; "'foo' debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]485#undef DEBUG_TYPE
486#define DEBUG_TYPE "bar"
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]487DOUT &lt;&lt; "'bar' debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]488#undef DEBUG_TYPE
489#define DEBUG_TYPE ""
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]490DOUT &lt;&lt; "No debug type (2)\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]491</pre>
492</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]493
494<p>Then you can run your pass like this:</p>
495
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]496<div class="doc_code">
497<pre>
498$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]499<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]500$ opt &lt; a.bc &gt; /dev/null -mypass -debug
501No debug type
502'foo' debug type
503'bar' debug type
504No debug type (2)
505$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=foo
506'foo' debug type
507$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=bar
508'bar' debug type
509</pre>
510</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]511
512<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
513a file, to specify the debug type for the entire module (if you do this before
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]514you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]515<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
516"bar", because there is no system in place to ensure that names do not
517conflict. If two different modules use the same string, they will all be turned
518on when the name is specified. This allows, for example, all debug information
519for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]520even if the source lives in multiple files.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]521
522</div>
523
524<!-- ======================================================================= -->
525<div class="doc_subsection">
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]526 <a name="Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]527 option</a>
528</div>
529
530<div class="doc_text">
531
532<p>The "<tt><a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]533href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]534provides a class named <tt>Statistic</tt> that is used as a unified way to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]535keep track of what the LLVM compiler is doing and how effective various
536optimizations are. It is useful to see what optimizations are contributing to
537making a particular program run faster.</p>
538
539<p>Often you may run your pass on some big program, and you're interested to see
540how many times it makes a certain transformation. Although you can do this with
541hand inspection, or some ad-hoc method, this is a real pain and not very useful
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]542for big programs. Using the <tt>Statistic</tt> class makes it very easy to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]543keep track of this information, and the calculated information is presented in a
544uniform manner with the rest of the passes being executed.</p>
545
546<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
547it are as follows:</p>
548
549<ol>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]550 <li><p>Define your statistic like this:</p>
551
552<div class="doc_code">
553<pre>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]554#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
555STATISTIC(NumXForms, "The # of times I did stuff");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]556</pre>
557</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]558
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]559 <p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
560 specified by the first argument. The pass name is taken from the DEBUG_TYPE
561 macro, and the description is taken from the second argument. The variable
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]562 defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]563
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]564 <li><p>Whenever you make a transformation, bump the counter:</p>
565
566<div class="doc_code">
567<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]568++NumXForms; // <i>I did stuff!</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]569</pre>
570</div>
571
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]572 </li>
573 </ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]574
575 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
576 statistics gathered, use the '<tt>-stats</tt>' option:</p>
577
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]578<div class="doc_code">
579<pre>
580$ opt -stats -mypassname &lt; program.bc &gt; /dev/null
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]581<i>... statistics output ...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]582</pre>
583</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]584
Reid Spencer6b6c73e2007-02-09 16:00:28 +0000[diff] [blame]585 <p> When running <tt>opt</tt> on a C file from the SPEC benchmark
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]586suite, it gives a report that looks like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]587
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]588<div class="doc_code">
589<pre>
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]590 7646 bitcodewriter - Number of normal instructions
591 725 bitcodewriter - Number of oversized instructions
592 129996 bitcodewriter - Number of bitcode bytes written
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]593 2817 raise - Number of insts DCEd or constprop'd
594 3213 raise - Number of cast-of-self removed
595 5046 raise - Number of expression trees converted
596 75 raise - Number of other getelementptr's formed
597 138 raise - Number of load/store peepholes
598 42 deadtypeelim - Number of unused typenames removed from symtab
599 392 funcresolve - Number of varargs functions resolved
600 27 globaldce - Number of global variables removed
601 2 adce - Number of basic blocks removed
602 134 cee - Number of branches revectored
603 49 cee - Number of setcc instruction eliminated
604 532 gcse - Number of loads removed
605 2919 gcse - Number of instructions removed
606 86 indvars - Number of canonical indvars added
607 87 indvars - Number of aux indvars removed
608 25 instcombine - Number of dead inst eliminate
609 434 instcombine - Number of insts combined
610 248 licm - Number of load insts hoisted
611 1298 licm - Number of insts hoisted to a loop pre-header
612 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
613 75 mem2reg - Number of alloca's promoted
614 1444 cfgsimplify - Number of blocks simplified
615</pre>
616</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]617
618<p>Obviously, with so many optimizations, having a unified framework for this
619stuff is very nice. Making your pass fit well into the framework makes it more
620maintainable and useful.</p>
621
622</div>
623
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]624<!-- ======================================================================= -->
625<div class="doc_subsection">
626 <a name="ViewGraph">Viewing graphs while debugging code</a>
627</div>
628
629<div class="doc_text">
630
631<p>Several of the important data structures in LLVM are graphs: for example
632CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
633LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
634<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
635DAGs</a>. In many cases, while debugging various parts of the compiler, it is
636nice to instantly visualize these graphs.</p>
637
638<p>LLVM provides several callbacks that are available in a debug build to do
639exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
640the current LLVM tool will pop up a window containing the CFG for the function
641where each basic block is a node in the graph, and each node contains the
642instructions in the block. Similarly, there also exists
643<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
644<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
645and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]646you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]647up a window. Alternatively, you can sprinkle calls to these functions in your
648code in places you want to debug.</p>
649
650<p>Getting this to work requires a small amount of configuration. On Unix
651systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
652toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
653Mac OS/X, download and install the Mac OS/X <a
654href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]655<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]656it) to your path. Once in your system and path are set up, rerun the LLVM
657configure script and rebuild LLVM to enable this functionality.</p>
658
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]659<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
660<i>interesting</i> nodes in large complex graphs. From gdb, if you
661<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]662next <tt>call DAG.viewGraph()</tt> would highlight the node in the
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]663specified color (choices of colors can be found at <a
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]664href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]665complex node attributes can be provided with <tt>call
666DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
667found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
668Attributes</a>.) If you want to restart and clear all the current graph
669attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
670
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]671</div>
672
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]673<!-- *********************************************************************** -->
674<div class="doc_section">
675 <a name="datastructure">Picking the Right Data Structure for a Task</a>
676</div>
677<!-- *********************************************************************** -->
678
679<div class="doc_text">
680
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]681<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
682 and we commonly use STL data structures. This section describes the trade-offs
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]683 you should consider when you pick one.</p>
684
685<p>
686The first step is a choose your own adventure: do you want a sequential
687container, a set-like container, or a map-like container? The most important
688thing when choosing a container is the algorithmic properties of how you plan to
689access the container. Based on that, you should use:</p>
690
691<ul>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]692<li>a <a href="#ds_map">map-like</a> container if you need efficient look-up
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]693 of an value based on another value. Map-like containers also support
694 efficient queries for containment (whether a key is in the map). Map-like
695 containers generally do not support efficient reverse mapping (values to
696 keys). If you need that, use two maps. Some map-like containers also
697 support efficient iteration through the keys in sorted order. Map-like
698 containers are the most expensive sort, only use them if you need one of
699 these capabilities.</li>
700
701<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
702 stuff into a container that automatically eliminates duplicates. Some
703 set-like containers support efficient iteration through the elements in
704 sorted order. Set-like containers are more expensive than sequential
705 containers.
706</li>
707
708<li>a <a href="#ds_sequential">sequential</a> container provides
709 the most efficient way to add elements and keeps track of the order they are
710 added to the collection. They permit duplicates and support efficient
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]711 iteration, but do not support efficient look-up based on a key.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]712</li>
713
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]714<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
715 perform set operations on sets of numeric id's, while automatically
716 eliminating duplicates. Bit containers require a maximum of 1 bit for each
717 identifier you want to store.
718</li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]719</ul>
720
721<p>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]722Once the proper category of container is determined, you can fine tune the
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]723memory use, constant factors, and cache behaviors of access by intelligently
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]724picking a member of the category. Note that constant factors and cache behavior
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]725can be a big deal. If you have a vector that usually only contains a few
726elements (but could contain many), for example, it's much better to use
727<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
728. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
729cost of adding the elements to the container. </p>
730
731</div>
732
733<!-- ======================================================================= -->
734<div class="doc_subsection">
735 <a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
736</div>
737
738<div class="doc_text">
739There are a variety of sequential containers available for you, based on your
740needs. Pick the first in this section that will do what you want.
741</div>
742
743<!-- _______________________________________________________________________ -->
744<div class="doc_subsubsection">
745 <a name="dss_fixedarrays">Fixed Size Arrays</a>
746</div>
747
748<div class="doc_text">
749<p>Fixed size arrays are very simple and very fast. They are good if you know
750exactly how many elements you have, or you have a (low) upper bound on how many
751you have.</p>
752</div>
753
754<!-- _______________________________________________________________________ -->
755<div class="doc_subsubsection">
756 <a name="dss_heaparrays">Heap Allocated Arrays</a>
757</div>
758
759<div class="doc_text">
760<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
761the number of elements is variable, if you know how many elements you will need
762before the array is allocated, and if the array is usually large (if not,
763consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
764allocated array is the cost of the new/delete (aka malloc/free). Also note that
765if you are allocating an array of a type with a constructor, the constructor and
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]766destructors will be run for every element in the array (re-sizable vectors only
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]767construct those elements actually used).</p>
768</div>
769
770<!-- _______________________________________________________________________ -->
771<div class="doc_subsubsection">
772 <a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
773</div>
774
775<div class="doc_text">
776<p><tt>SmallVector&lt;Type, N&gt;</tt> is a simple class that looks and smells
777just like <tt>vector&lt;Type&gt;</tt>:
778it supports efficient iteration, lays out elements in memory order (so you can
779do pointer arithmetic between elements), supports efficient push_back/pop_back
780operations, supports efficient random access to its elements, etc.</p>
781
782<p>The advantage of SmallVector is that it allocates space for
783some number of elements (N) <b>in the object itself</b>. Because of this, if
784the SmallVector is dynamically smaller than N, no malloc is performed. This can
785be a big win in cases where the malloc/free call is far more expensive than the
786code that fiddles around with the elements.</p>
787
788<p>This is good for vectors that are "usually small" (e.g. the number of
789predecessors/successors of a block is usually less than 8). On the other hand,
790this makes the size of the SmallVector itself large, so you don't want to
791allocate lots of them (doing so will waste a lot of space). As such,
792SmallVectors are most useful when on the stack.</p>
793
794<p>SmallVector also provides a nice portable and efficient replacement for
795<tt>alloca</tt>.</p>
796
797</div>
798
799<!-- _______________________________________________________________________ -->
800<div class="doc_subsubsection">
801 <a name="dss_vector">&lt;vector&gt;</a>
802</div>
803
804<div class="doc_text">
805<p>
806std::vector is well loved and respected. It is useful when SmallVector isn't:
807when the size of the vector is often large (thus the small optimization will
808rarely be a benefit) or if you will be allocating many instances of the vector
809itself (which would waste space for elements that aren't in the container).
810vector is also useful when interfacing with code that expects vectors :).
811</p>
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]812
813<p>One worthwhile note about std::vector: avoid code like this:</p>
814
815<div class="doc_code">
816<pre>
817for ( ... ) {
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]818 std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]819 use V;
820}
821</pre>
822</div>
823
824<p>Instead, write this as:</p>
825
826<div class="doc_code">
827<pre>
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]828std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]829for ( ... ) {
830 use V;
831 V.clear();
832}
833</pre>
834</div>
835
836<p>Doing so will save (at least) one heap allocation and free per iteration of
837the loop.</p>
838
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]839</div>
840
841<!-- _______________________________________________________________________ -->
842<div class="doc_subsubsection">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]843 <a name="dss_deque">&lt;deque&gt;</a>
844</div>
845
846<div class="doc_text">
847<p>std::deque is, in some senses, a generalized version of std::vector. Like
848std::vector, it provides constant time random access and other similar
849properties, but it also provides efficient access to the front of the list. It
850does not guarantee continuity of elements within memory.</p>
851
852<p>In exchange for this extra flexibility, std::deque has significantly higher
853constant factor costs than std::vector. If possible, use std::vector or
854something cheaper.</p>
855</div>
856
857<!-- _______________________________________________________________________ -->
858<div class="doc_subsubsection">
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]859 <a name="dss_list">&lt;list&gt;</a>
860</div>
861
862<div class="doc_text">
863<p>std::list is an extremely inefficient class that is rarely useful.
864It performs a heap allocation for every element inserted into it, thus having an
865extremely high constant factor, particularly for small data types. std::list
866also only supports bidirectional iteration, not random access iteration.</p>
867
868<p>In exchange for this high cost, std::list supports efficient access to both
869ends of the list (like std::deque, but unlike std::vector or SmallVector). In
870addition, the iterator invalidation characteristics of std::list are stronger
871than that of a vector class: inserting or removing an element into the list does
872not invalidate iterator or pointers to other elements in the list.</p>
873</div>
874
875<!-- _______________________________________________________________________ -->
876<div class="doc_subsubsection">
877 <a name="dss_ilist">llvm/ADT/ilist</a>
878</div>
879
880<div class="doc_text">
881<p><tt>ilist&lt;T&gt;</tt> implements an 'intrusive' doubly-linked list. It is
882intrusive, because it requires the element to store and provide access to the
883prev/next pointers for the list.</p>
884
885<p>ilist has the same drawbacks as std::list, and additionally requires an
886ilist_traits implementation for the element type, but it provides some novel
887characteristics. In particular, it can efficiently store polymorphic objects,
888the traits class is informed when an element is inserted or removed from the
889list, and ilists are guaranteed to support a constant-time splice operation.
890</p>
891
892<p>These properties are exactly what we want for things like Instructions and
893basic blocks, which is why these are implemented with ilists.</p>
894</div>
895
896<!-- _______________________________________________________________________ -->
897<div class="doc_subsubsection">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]898 <a name="dss_other">Other Sequential Container options</a>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]899</div>
900
901<div class="doc_text">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]902<p>Other STL containers are available, such as std::string.</p>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]903
904<p>There are also various STL adapter classes such as std::queue,
905std::priority_queue, std::stack, etc. These provide simplified access to an
906underlying container but don't affect the cost of the container itself.</p>
907
908</div>
909
910
911<!-- ======================================================================= -->
912<div class="doc_subsection">
913 <a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
914</div>
915
916<div class="doc_text">
917
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]918<p>Set-like containers are useful when you need to canonicalize multiple values
919into a single representation. There are several different choices for how to do
920this, providing various trade-offs.</p>
921
922</div>
923
924
925<!-- _______________________________________________________________________ -->
926<div class="doc_subsubsection">
927 <a name="dss_sortedvectorset">A sorted 'vector'</a>
928</div>
929
930<div class="doc_text">
931
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]932<p>If you intend to insert a lot of elements, then do a lot of queries, a
933great approach is to use a vector (or other sequential container) with
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]934std::sort+std::unique to remove duplicates. This approach works really well if
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]935your usage pattern has these two distinct phases (insert then query), and can be
936coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
937</p>
938
939<p>
940This combination provides the several nice properties: the result data is
941contiguous in memory (good for cache locality), has few allocations, is easy to
942address (iterators in the final vector are just indices or pointers), and can be
943efficiently queried with a standard binary or radix search.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]944
945</div>
946
947<!-- _______________________________________________________________________ -->
948<div class="doc_subsubsection">
949 <a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
950</div>
951
952<div class="doc_text">
953
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]954<p>If you have a set-like data structure that is usually small and whose elements
Chris Lattner4ddfac12007-02-03 07:59:51 +0000[diff] [blame]955are reasonably small, a <tt>SmallSet&lt;Type, N&gt;</tt> is a good choice. This set
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]956has space for N elements in place (thus, if the set is dynamically smaller than
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]957N, no malloc traffic is required) and accesses them with a simple linear search.
958When the set grows beyond 'N' elements, it allocates a more expensive representation that
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]959guarantees efficient access (for most types, it falls back to std::set, but for
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]960pointers it uses something far better, <a
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]961href="#dss_smallptrset">SmallPtrSet</a>).</p>
962
963<p>The magic of this class is that it handles small sets extremely efficiently,
964but gracefully handles extremely large sets without loss of efficiency. The
965drawback is that the interface is quite small: it supports insertion, queries
966and erasing, but does not support iteration.</p>
967
968</div>
969
970<!-- _______________________________________________________________________ -->
971<div class="doc_subsubsection">
972 <a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
973</div>
974
975<div class="doc_text">
976
977<p>SmallPtrSet has all the advantages of SmallSet (and a SmallSet of pointers is
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]978transparently implemented with a SmallPtrSet), but also supports iterators. If
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]979more than 'N' insertions are performed, a single quadratically
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]980probed hash table is allocated and grows as needed, providing extremely
981efficient access (constant time insertion/deleting/queries with low constant
982factors) and is very stingy with malloc traffic.</p>
983
984<p>Note that, unlike std::set, the iterators of SmallPtrSet are invalidated
985whenever an insertion occurs. Also, the values visited by the iterators are not
986visited in sorted order.</p>
987
988</div>
989
990<!-- _______________________________________________________________________ -->
991<div class="doc_subsubsection">
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]992 <a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
993</div>
994
995<div class="doc_text">
996
997<p>
998DenseSet is a simple quadratically probed hash table. It excels at supporting
999small values: it uses a single allocation to hold all of the pairs that
1000are currently inserted in the set. DenseSet is a great way to unique small
1001values that are not simple pointers (use <a
1002href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
1003the same requirements for the value type that <a
1004href="#dss_densemap">DenseMap</a> has.
1005</p>
1006
1007</div>
1008
1009<!-- _______________________________________________________________________ -->
1010<div class="doc_subsubsection">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1011 <a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
1012</div>
1013
1014<div class="doc_text">
1015
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1016<p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1017FoldingSet is an aggregate class that is really good at uniquing
1018expensive-to-create or polymorphic objects. It is a combination of a chained
1019hash table with intrusive links (uniqued objects are required to inherit from
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1020FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
1021its ID process.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1022
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1023<p>Consider a case where you want to implement a "getOrCreateFoo" method for
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1024a complex object (for example, a node in the code generator). The client has a
1025description of *what* it wants to generate (it knows the opcode and all the
1026operands), but we don't want to 'new' a node, then try inserting it into a set
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1027only to find out it already exists, at which point we would have to delete it
1028and return the node that already exists.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1029</p>
1030
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1031<p>To support this style of client, FoldingSet perform a query with a
1032FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1033element that we want to query for. The query either returns the element
1034matching the ID or it returns an opaque ID that indicates where insertion should
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1035take place. Construction of the ID usually does not require heap traffic.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1036
1037<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
1038in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1039Because the elements are individually allocated, pointers to the elements are
1040stable: inserting or removing elements does not invalidate any pointers to other
1041elements.
1042</p>
1043
1044</div>
1045
1046<!-- _______________________________________________________________________ -->
1047<div class="doc_subsubsection">
1048 <a name="dss_set">&lt;set&gt;</a>
1049</div>
1050
1051<div class="doc_text">
1052
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1053<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
1054many things but great at nothing. std::set allocates memory for each element
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1055inserted (thus it is very malloc intensive) and typically stores three pointers
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1056per element in the set (thus adding a large amount of per-element space
1057overhead). It offers guaranteed log(n) performance, which is not particularly
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1058fast from a complexity standpoint (particularly if the elements of the set are
1059expensive to compare, like strings), and has extremely high constant factors for
1060lookup, insertion and removal.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1061
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1062<p>The advantages of std::set are that its iterators are stable (deleting or
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1063inserting an element from the set does not affect iterators or pointers to other
1064elements) and that iteration over the set is guaranteed to be in sorted order.
1065If the elements in the set are large, then the relative overhead of the pointers
1066and malloc traffic is not a big deal, but if the elements of the set are small,
1067std::set is almost never a good choice.</p>
1068
1069</div>
1070
1071<!-- _______________________________________________________________________ -->
1072<div class="doc_subsubsection">
1073 <a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
1074</div>
1075
1076<div class="doc_text">
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1077<p>LLVM's SetVector&lt;Type&gt; is an adapter class that combines your choice of
1078a set-like container along with a <a href="#ds_sequential">Sequential
1079Container</a>. The important property
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1080that this provides is efficient insertion with uniquing (duplicate elements are
1081ignored) with iteration support. It implements this by inserting elements into
1082both a set-like container and the sequential container, using the set-like
1083container for uniquing and the sequential container for iteration.
1084</p>
1085
1086<p>The difference between SetVector and other sets is that the order of
1087iteration is guaranteed to match the order of insertion into the SetVector.
1088This property is really important for things like sets of pointers. Because
1089pointer values are non-deterministic (e.g. vary across runs of the program on
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1090different machines), iterating over the pointers in the set will
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1091not be in a well-defined order.</p>
1092
1093<p>
1094The drawback of SetVector is that it requires twice as much space as a normal
1095set and has the sum of constant factors from the set-like container and the
1096sequential container that it uses. Use it *only* if you need to iterate over
1097the elements in a deterministic order. SetVector is also expensive to delete
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1098elements out of (linear time), unless you use it's "pop_back" method, which is
1099faster.
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1100</p>
1101
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1102<p>SetVector is an adapter class that defaults to using std::vector and std::set
1103for the underlying containers, so it is quite expensive. However,
1104<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
1105defaults to using a SmallVector and SmallSet of a specified size. If you use
1106this, and if your sets are dynamically smaller than N, you will save a lot of
1107heap traffic.</p>
1108
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1109</div>
1110
1111<!-- _______________________________________________________________________ -->
1112<div class="doc_subsubsection">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1113 <a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
1114</div>
1115
1116<div class="doc_text">
1117
1118<p>
1119UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
1120retains a unique ID for each element inserted into the set. It internally
1121contains a map and a vector, and it assigns a unique ID for each value inserted
1122into the set.</p>
1123
1124<p>UniqueVector is very expensive: its cost is the sum of the cost of
1125maintaining both the map and vector, it has high complexity, high constant
1126factors, and produces a lot of malloc traffic. It should be avoided.</p>
1127
1128</div>
1129
1130
1131<!-- _______________________________________________________________________ -->
1132<div class="doc_subsubsection">
1133 <a name="dss_otherset">Other Set-Like Container Options</a>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1134</div>
1135
1136<div class="doc_text">
1137
1138<p>
1139The STL provides several other options, such as std::multiset and the various
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1140"hash_set" like containers (whether from C++ TR1 or from the SGI library).</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1141
1142<p>std::multiset is useful if you're not interested in elimination of
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1143duplicates, but has all the drawbacks of std::set. A sorted vector (where you
1144don't delete duplicate entries) or some other approach is almost always
1145better.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1146
1147<p>The various hash_set implementations (exposed portably by
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1148"llvm/ADT/hash_set") is a simple chained hashtable. This algorithm is as malloc
1149intensive as std::set (performing an allocation for each element inserted,
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1150thus having really high constant factors) but (usually) provides O(1)
1151insertion/deletion of elements. This can be useful if your elements are large
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1152(thus making the constant-factor cost relatively low) or if comparisons are
1153expensive. Element iteration does not visit elements in a useful order.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1154
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1155</div>
1156
1157<!-- ======================================================================= -->
1158<div class="doc_subsection">
1159 <a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
1160</div>
1161
1162<div class="doc_text">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1163Map-like containers are useful when you want to associate data to a key. As
1164usual, there are a lot of different ways to do this. :)
1165</div>
1166
1167<!-- _______________________________________________________________________ -->
1168<div class="doc_subsubsection">
1169 <a name="dss_sortedvectormap">A sorted 'vector'</a>
1170</div>
1171
1172<div class="doc_text">
1173
1174<p>
1175If your usage pattern follows a strict insert-then-query approach, you can
1176trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
1177for set-like containers</a>. The only difference is that your query function
1178(which uses std::lower_bound to get efficient log(n) lookup) should only compare
1179the key, not both the key and value. This yields the same advantages as sorted
1180vectors for sets.
1181</p>
1182</div>
1183
1184<!-- _______________________________________________________________________ -->
1185<div class="doc_subsubsection">
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1186 <a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1187</div>
1188
1189<div class="doc_text">
1190
1191<p>
1192Strings are commonly used as keys in maps, and they are difficult to support
1193efficiently: they are variable length, inefficient to hash and compare when
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1194long, expensive to copy, etc. StringMap is a specialized container designed to
1195cope with these issues. It supports mapping an arbitrary range of bytes to an
1196arbitrary other object.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1197
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1198<p>The StringMap implementation uses a quadratically-probed hash table, where
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1199the buckets store a pointer to the heap allocated entries (and some other
1200stuff). The entries in the map must be heap allocated because the strings are
1201variable length. The string data (key) and the element object (value) are
1202stored in the same allocation with the string data immediately after the element
1203object. This container guarantees the "<tt>(char*)(&amp;Value+1)</tt>" points
1204to the key string for a value.</p>
1205
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1206<p>The StringMap is very fast for several reasons: quadratic probing is very
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1207cache efficient for lookups, the hash value of strings in buckets is not
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1208recomputed when lookup up an element, StringMap rarely has to touch the
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1209memory for unrelated objects when looking up a value (even when hash collisions
1210happen), hash table growth does not recompute the hash values for strings
1211already in the table, and each pair in the map is store in a single allocation
1212(the string data is stored in the same allocation as the Value of a pair).</p>
1213
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1214<p>StringMap also provides query methods that take byte ranges, so it only ever
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1215copies a string if a value is inserted into the table.</p>
1216</div>
1217
1218<!-- _______________________________________________________________________ -->
1219<div class="doc_subsubsection">
1220 <a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
1221</div>
1222
1223<div class="doc_text">
1224<p>
1225IndexedMap is a specialized container for mapping small dense integers (or
1226values that can be mapped to small dense integers) to some other type. It is
1227internally implemented as a vector with a mapping function that maps the keys to
1228the dense integer range.
1229</p>
1230
1231<p>
1232This is useful for cases like virtual registers in the LLVM code generator: they
1233have a dense mapping that is offset by a compile-time constant (the first
1234virtual register ID).</p>
1235
1236</div>
1237
1238<!-- _______________________________________________________________________ -->
1239<div class="doc_subsubsection">
1240 <a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
1241</div>
1242
1243<div class="doc_text">
1244
1245<p>
1246DenseMap is a simple quadratically probed hash table. It excels at supporting
1247small keys and values: it uses a single allocation to hold all of the pairs that
1248are currently inserted in the map. DenseMap is a great way to map pointers to
1249pointers, or map other small types to each other.
1250</p>
1251
1252<p>
1253There are several aspects of DenseMap that you should be aware of, however. The
1254iterators in a densemap are invalidated whenever an insertion occurs, unlike
1255map. Also, because DenseMap allocates space for a large number of key/value
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1256pairs (it starts with 64 by default), it will waste a lot of space if your keys
1257or values are large. Finally, you must implement a partial specialization of
Chris Lattner76c1b972007-09-17 18:34:04 +0000[diff] [blame]1258DenseMapInfo for the key that you want, if it isn't already supported. This
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1259is required to tell DenseMap about two special marker values (which can never be
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1260inserted into the map) that it needs internally.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1261
1262</div>
1263
1264<!-- _______________________________________________________________________ -->
1265<div class="doc_subsubsection">
1266 <a name="dss_map">&lt;map&gt;</a>
1267</div>
1268
1269<div class="doc_text">
1270
1271<p>
1272std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
1273a single allocation per pair inserted into the map, it offers log(n) lookup with
1274an extremely large constant factor, imposes a space penalty of 3 pointers per
1275pair in the map, etc.</p>
1276
1277<p>std::map is most useful when your keys or values are very large, if you need
1278to iterate over the collection in sorted order, or if you need stable iterators
1279into the map (i.e. they don't get invalidated if an insertion or deletion of
1280another element takes place).</p>
1281
1282</div>
1283
1284<!-- _______________________________________________________________________ -->
1285<div class="doc_subsubsection">
1286 <a name="dss_othermap">Other Map-Like Container Options</a>
1287</div>
1288
1289<div class="doc_text">
1290
1291<p>
1292The STL provides several other options, such as std::multimap and the various
1293"hash_map" like containers (whether from C++ TR1 or from the SGI library).</p>
1294
1295<p>std::multimap is useful if you want to map a key to multiple values, but has
1296all the drawbacks of std::map. A sorted vector or some other approach is almost
1297always better.</p>
1298
1299<p>The various hash_map implementations (exposed portably by
1300"llvm/ADT/hash_map") are simple chained hash tables. This algorithm is as
1301malloc intensive as std::map (performing an allocation for each element
1302inserted, thus having really high constant factors) but (usually) provides O(1)
1303insertion/deletion of elements. This can be useful if your elements are large
1304(thus making the constant-factor cost relatively low) or if comparisons are
1305expensive. Element iteration does not visit elements in a useful order.</p>
1306
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1307</div>
1308
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1309<!-- ======================================================================= -->
1310<div class="doc_subsection">
1311 <a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
1312</div>
1313
1314<div class="doc_text">
Chris Lattner7086ce72007-09-25 22:37:50 +0000[diff] [blame]1315<p>Unlike the other containers, there are only two bit storage containers, and
1316choosing when to use each is relatively straightforward.</p>
1317
1318<p>One additional option is
1319<tt>std::vector&lt;bool&gt;</tt>: we discourage its use for two reasons 1) the
1320implementation in many common compilers (e.g. commonly available versions of
1321GCC) is extremely inefficient and 2) the C++ standards committee is likely to
1322deprecate this container and/or change it significantly somehow. In any case,
1323please don't use it.</p>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1324</div>
1325
1326<!-- _______________________________________________________________________ -->
1327<div class="doc_subsubsection">
1328 <a name="dss_bitvector">BitVector</a>
1329</div>
1330
1331<div class="doc_text">
1332<p> The BitVector container provides a fixed size set of bits for manipulation.
1333It supports individual bit setting/testing, as well as set operations. The set
1334operations take time O(size of bitvector), but operations are performed one word
1335at a time, instead of one bit at a time. This makes the BitVector very fast for
1336set operations compared to other containers. Use the BitVector when you expect
1337the number of set bits to be high (IE a dense set).
1338</p>
1339</div>
1340
1341<!-- _______________________________________________________________________ -->
1342<div class="doc_subsubsection">
1343 <a name="dss_sparsebitvector">SparseBitVector</a>
1344</div>
1345
1346<div class="doc_text">
1347<p> The SparseBitVector container is much like BitVector, with one major
1348difference: Only the bits that are set, are stored. This makes the
1349SparseBitVector much more space efficient than BitVector when the set is sparse,
1350as well as making set operations O(number of set bits) instead of O(size of
1351universe). The downside to the SparseBitVector is that setting and testing of random bits is O(N), and on large SparseBitVectors, this can be slower than BitVector. In our implementation, setting or testing bits in sorted order
1352(either forwards or reverse) is O(1) worst case. Testing and setting bits within 128 bits (depends on size) of the current bit is also O(1). As a general statement, testing/setting bits in a SparseBitVector is O(distance away from last set bit).
1353</p>
1354</div>
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]1355
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1356<!-- *********************************************************************** -->
1357<div class="doc_section">
1358 <a name="common">Helpful Hints for Common Operations</a>
1359</div>
1360<!-- *********************************************************************** -->
1361
1362<div class="doc_text">
1363
1364<p>This section describes how to perform some very simple transformations of
1365LLVM code. This is meant to give examples of common idioms used, showing the
1366practical side of LLVM transformations. <p> Because this is a "how-to" section,
1367you should also read about the main classes that you will be working with. The
1368<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
1369and descriptions of the main classes that you should know about.</p>
1370
1371</div>
1372
1373<!-- NOTE: this section should be heavy on example code -->
1374<!-- ======================================================================= -->
1375<div class="doc_subsection">
1376 <a name="inspection">Basic Inspection and Traversal Routines</a>
1377</div>
1378
1379<div class="doc_text">
1380
1381<p>The LLVM compiler infrastructure have many different data structures that may
1382be traversed. Following the example of the C++ standard template library, the
1383techniques used to traverse these various data structures are all basically the
1384same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
1385method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
1386function returns an iterator pointing to one past the last valid element of the
1387sequence, and there is some <tt>XXXiterator</tt> data type that is common
1388between the two operations.</p>
1389
1390<p>Because the pattern for iteration is common across many different aspects of
1391the program representation, the standard template library algorithms may be used
1392on them, and it is easier to remember how to iterate. First we show a few common
1393examples of the data structures that need to be traversed. Other data
1394structures are traversed in very similar ways.</p>
1395
1396</div>
1397
1398<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1399<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1400 <a name="iterate_function">Iterating over the </a><a
1401 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
1402 href="#Function"><tt>Function</tt></a>
1403</div>
1404
1405<div class="doc_text">
1406
1407<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
1408transform in some way; in particular, you'd like to manipulate its
1409<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
1410the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
1411an example that prints the name of a <tt>BasicBlock</tt> and the number of
1412<tt>Instruction</tt>s it contains:</p>
1413
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1414<div class="doc_code">
1415<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1416// <i>func is a pointer to a Function instance</i>
1417for (Function::iterator i = func-&gt;begin(), e = func-&gt;end(); i != e; ++i)
1418 // <i>Print out the name of the basic block if it has one, and then the</i>
1419 // <i>number of instructions that it contains</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1420 llvm::cerr &lt;&lt; "Basic block (name=" &lt;&lt; i-&gt;getName() &lt;&lt; ") has "
1421 &lt;&lt; i-&gt;size() &lt;&lt; " instructions.\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1422</pre>
1423</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1424
1425<p>Note that i can be used as if it were a pointer for the purposes of
Joel Stanley9b96c442002-09-06 21:55:13 +0000[diff] [blame]1426invoking member functions of the <tt>Instruction</tt> class. This is
1427because the indirection operator is overloaded for the iterator
Chris Lattner7496ec52003-08-05 22:54:23 +0000[diff] [blame]1428classes. In the above code, the expression <tt>i-&gt;size()</tt> is
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1429exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
1430
1431</div>
1432
1433<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1434<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1435 <a name="iterate_basicblock">Iterating over the </a><a
1436 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1437 href="#BasicBlock"><tt>BasicBlock</tt></a>
1438</div>
1439
1440<div class="doc_text">
1441
1442<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
1443easy to iterate over the individual instructions that make up
1444<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
1445a <tt>BasicBlock</tt>:</p>
1446
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1447<div class="doc_code">
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]1448<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1449// <i>blk is a pointer to a BasicBlock instance</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1450for (BasicBlock::iterator i = blk-&gt;begin(), e = blk-&gt;end(); i != e; ++i)
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1451 // <i>The next statement works since operator&lt;&lt;(ostream&amp;,...)</i>
1452 // <i>is overloaded for Instruction&amp;</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1453 llvm::cerr &lt;&lt; *i &lt;&lt; "\n";
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]1454</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1455</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1456
1457<p>However, this isn't really the best way to print out the contents of a
1458<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
1459anything you'll care about, you could have just invoked the print routine on the
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1460basic block itself: <tt>llvm::cerr &lt;&lt; *blk &lt;&lt; "\n";</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1461
1462</div>
1463
1464<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1465<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1466 <a name="iterate_institer">Iterating over the </a><a
1467 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1468 href="#Function"><tt>Function</tt></a>
1469</div>
1470
1471<div class="doc_text">
1472
1473<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
1474<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
1475<tt>InstIterator</tt> should be used instead. You'll need to include <a
1476href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
1477and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1478small example that shows how to dump all instructions in a function to the standard error stream:<p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1479
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1480<div class="doc_code">
1481<pre>
1482#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
1483
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1484// <i>F is a pointer to a Function instance</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1485for (inst_iterator i = inst_begin(F), e = inst_end(F); i != e; ++i)
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1486 llvm::cerr &lt;&lt; *i &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1487</pre>
1488</div>
1489
1490<p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1491work list with its initial contents. For example, if you wanted to
1492initialize a work list to contain all instructions in a <tt>Function</tt>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1493F, all you would need to do is something like:</p>
1494
1495<div class="doc_code">
1496<pre>
1497std::set&lt;Instruction*&gt; worklist;
1498worklist.insert(inst_begin(F), inst_end(F));
1499</pre>
1500</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1501
1502<p>The STL set <tt>worklist</tt> would now contain all instructions in the
1503<tt>Function</tt> pointed to by F.</p>
1504
1505</div>
1506
1507<!-- _______________________________________________________________________ -->
1508<div class="doc_subsubsection">
1509 <a name="iterate_convert">Turning an iterator into a class pointer (and
1510 vice-versa)</a>
1511</div>
1512
1513<div class="doc_text">
1514
1515<p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
Joel Stanley9b96c442002-09-06 21:55:13 +0000[diff] [blame]1516instance when all you've got at hand is an iterator. Well, extracting
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1517a reference or a pointer from an iterator is very straight-forward.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1518Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1519is a <tt>BasicBlock::const_iterator</tt>:</p>
1520
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1521<div class="doc_code">
1522<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1523Instruction&amp; inst = *i; // <i>Grab reference to instruction reference</i>
1524Instruction* pinst = &amp;*i; // <i>Grab pointer to instruction reference</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1525const Instruction&amp; inst = *j;
1526</pre>
1527</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1528
1529<p>However, the iterators you'll be working with in the LLVM framework are
1530special: they will automatically convert to a ptr-to-instance type whenever they
1531need to. Instead of dereferencing the iterator and then taking the address of
1532the result, you can simply assign the iterator to the proper pointer type and
1533you get the dereference and address-of operation as a result of the assignment
1534(behind the scenes, this is a result of overloading casting mechanisms). Thus
1535the last line of the last example,</p>
1536
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1537<div class="doc_code">
1538<pre>
1539Instruction* pinst = &amp;*i;
1540</pre>
1541</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1542
1543<p>is semantically equivalent to</p>
1544
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1545<div class="doc_code">
1546<pre>
1547Instruction* pinst = i;
1548</pre>
1549</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1550
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1551<p>It's also possible to turn a class pointer into the corresponding iterator,
1552and this is a constant time operation (very efficient). The following code
1553snippet illustrates use of the conversion constructors provided by LLVM
1554iterators. By using these, you can explicitly grab the iterator of something
1555without actually obtaining it via iteration over some structure:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1556
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1557<div class="doc_code">
1558<pre>
1559void printNextInstruction(Instruction* inst) {
1560 BasicBlock::iterator it(inst);
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1561 ++it; // <i>After this line, it refers to the instruction after *inst</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1562 if (it != inst-&gt;getParent()-&gt;end()) llvm::cerr &lt;&lt; *it &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1563}
1564</pre>
1565</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1566
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1567</div>
1568
1569<!--_______________________________________________________________________-->
1570<div class="doc_subsubsection">
1571 <a name="iterate_complex">Finding call sites: a slightly more complex
1572 example</a>
1573</div>
1574
1575<div class="doc_text">
1576
1577<p>Say that you're writing a FunctionPass and would like to count all the
1578locations in the entire module (that is, across every <tt>Function</tt>) where a
1579certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
1580learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1581much more straight-forward manner, but this example will allow us to explore how
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1582you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudo-code, this
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1583is what we want to do:</p>
1584
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1585<div class="doc_code">
1586<pre>
1587initialize callCounter to zero
1588for each Function f in the Module
1589 for each BasicBlock b in f
1590 for each Instruction i in b
1591 if (i is a CallInst and calls the given function)
1592 increment callCounter
1593</pre>
1594</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1595
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1596<p>And the actual code is (remember, because we're writing a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1597<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1598override the <tt>runOnFunction</tt> method):</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1599
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1600<div class="doc_code">
1601<pre>
1602Function* targetFunc = ...;
1603
1604class OurFunctionPass : public FunctionPass {
1605 public:
1606 OurFunctionPass(): callCounter(0) { }
1607
1608 virtual runOnFunction(Function&amp; F) {
1609 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
1610 for (BasicBlock::iterator i = b-&gt;begin(); ie = b-&gt;end(); i != ie; ++i) {
1611 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a>&lt;<a
1612 href="#CallInst">CallInst</a>&gt;(&amp;*i)) {
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1613 // <i>We know we've encountered a call instruction, so we</i>
1614 // <i>need to determine if it's a call to the</i>
1615 // <i>function pointed to by m_func or not</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1616
1617 if (callInst-&gt;getCalledFunction() == targetFunc)
1618 ++callCounter;
1619 }
1620 }
1621 }
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1622 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1623
1624 private:
1625 unsigned callCounter;
1626};
1627</pre>
1628</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1629
1630</div>
1631
Brian Gaekef1972c62003-11-07 19:25:45 +0000[diff] [blame]1632<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1633<div class="doc_subsubsection">
1634 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
1635</div>
1636
1637<div class="doc_text">
1638
1639<p>You may have noticed that the previous example was a bit oversimplified in
1640that it did not deal with call sites generated by 'invoke' instructions. In
1641this, and in other situations, you may find that you want to treat
1642<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
1643most-specific common base class is <tt>Instruction</tt>, which includes lots of
1644less closely-related things. For these cases, LLVM provides a handy wrapper
1645class called <a
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]1646href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1647It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
1648methods that provide functionality common to <tt>CallInst</tt>s and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1649<tt>InvokeInst</tt>s.</p>
1650
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1651<p>This class has "value semantics": it should be passed by value, not by
1652reference and it should not be dynamically allocated or deallocated using
1653<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
1654assignable and constructable, with costs equivalents to that of a bare pointer.
1655If you look at its definition, it has only a single pointer member.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1656
1657</div>
1658
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]1659<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1660<div class="doc_subsubsection">
1661 <a name="iterate_chains">Iterating over def-use &amp; use-def chains</a>
1662</div>
1663
1664<div class="doc_text">
1665
1666<p>Frequently, we might have an instance of the <a
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]1667href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1668determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
1669<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
1670For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
1671particular function <tt>foo</tt>. Finding all of the instructions that
1672<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
1673of <tt>F</tt>:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1674
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1675<div class="doc_code">
1676<pre>
1677Function* F = ...;
1678
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1679for (Value::use_iterator i = F-&gt;use_begin(), e = F-&gt;use_end(); i != e; ++i)
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1680 if (Instruction *Inst = dyn_cast&lt;Instruction&gt;(*i)) {
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1681 llvm::cerr &lt;&lt; "F is used in instruction:\n";
1682 llvm::cerr &lt;&lt; *Inst &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1683 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1684</pre>
1685</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1686
1687<p>Alternately, it's common to have an instance of the <a
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1688href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1689<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
1690<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
1691<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
1692all of the values that a particular instruction uses (that is, the operands of
1693the particular <tt>Instruction</tt>):</p>
1694
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1695<div class="doc_code">
1696<pre>
1697Instruction* pi = ...;
1698
1699for (User::op_iterator i = pi-&gt;op_begin(), e = pi-&gt;op_end(); i != e; ++i) {
1700 Value* v = *i;
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1701 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1702}
1703</pre>
1704</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1705
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]1706<!--
1707 def-use chains ("finding all users of"): Value::use_begin/use_end
1708 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1709-->
1710
1711</div>
1712
1713<!-- ======================================================================= -->
1714<div class="doc_subsection">
1715 <a name="simplechanges">Making simple changes</a>
1716</div>
1717
1718<div class="doc_text">
1719
1720<p>There are some primitive transformation operations present in the LLVM
Joel Stanley753eb712002-09-11 22:32:24 +0000[diff] [blame]1721infrastructure that are worth knowing about. When performing
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1722transformations, it's fairly common to manipulate the contents of basic
1723blocks. This section describes some of the common methods for doing so
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1724and gives example code.</p>
1725
1726</div>
1727
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1728<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1729<div class="doc_subsubsection">
1730 <a name="schanges_creating">Creating and inserting new
1731 <tt>Instruction</tt>s</a>
1732</div>
1733
1734<div class="doc_text">
1735
1736<p><i>Instantiating Instructions</i></p>
1737
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1738<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1739constructor for the kind of instruction to instantiate and provide the necessary
1740parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
1741(const-ptr-to) <tt>Type</tt>. Thus:</p>
1742
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1743<div class="doc_code">
1744<pre>
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1745AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1746</pre>
1747</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1748
1749<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1750one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1751subclass is likely to have varying default parameters which change the semantics
1752of the instruction, so refer to the <a
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]1753href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1754Instruction</a> that you're interested in instantiating.</p>
1755
1756<p><i>Naming values</i></p>
1757
1758<p>It is very useful to name the values of instructions when you're able to, as
1759this facilitates the debugging of your transformations. If you end up looking
1760at generated LLVM machine code, you definitely want to have logical names
1761associated with the results of instructions! By supplying a value for the
1762<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
1763associate a logical name with the result of the instruction's execution at
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1764run time. For example, say that I'm writing a transformation that dynamically
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1765allocates space for an integer on the stack, and that integer is going to be
1766used as some kind of index by some other code. To accomplish this, I place an
1767<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
1768<tt>Function</tt>, and I'm intending to use it within the same
1769<tt>Function</tt>. I might do:</p>
1770
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1771<div class="doc_code">
1772<pre>
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1773AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1774</pre>
1775</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1776
1777<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1778execution value, which is a pointer to an integer on the run time stack.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1779
1780<p><i>Inserting instructions</i></p>
1781
1782<p>There are essentially two ways to insert an <tt>Instruction</tt>
1783into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
1784
Joel Stanley9dd1ad62002-09-18 03:17:23 +0000[diff] [blame]1785<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1786 <li>Insertion into an explicit instruction list
1787
1788 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
1789 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
1790 before <tt>*pi</tt>, we do the following: </p>
1791
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1792<div class="doc_code">
1793<pre>
1794BasicBlock *pb = ...;
1795Instruction *pi = ...;
1796Instruction *newInst = new Instruction(...);
1797
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1798pb-&gt;getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1799</pre>
1800</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1801
1802 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
1803 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
1804 classes provide constructors which take a pointer to a
1805 <tt>BasicBlock</tt> to be appended to. For example code that
1806 looked like: </p>
1807
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1808<div class="doc_code">
1809<pre>
1810BasicBlock *pb = ...;
1811Instruction *newInst = new Instruction(...);
1812
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1813pb-&gt;getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1814</pre>
1815</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1816
1817 <p>becomes: </p>
1818
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1819<div class="doc_code">
1820<pre>
1821BasicBlock *pb = ...;
1822Instruction *newInst = new Instruction(..., pb);
1823</pre>
1824</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1825
1826 <p>which is much cleaner, especially if you are creating
1827 long instruction streams.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1828
1829 <li>Insertion into an implicit instruction list
1830
1831 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
1832 are implicitly associated with an existing instruction list: the instruction
1833 list of the enclosing basic block. Thus, we could have accomplished the same
1834 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
1835 </p>
1836
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1837<div class="doc_code">
1838<pre>
1839Instruction *pi = ...;
1840Instruction *newInst = new Instruction(...);
1841
1842pi-&gt;getParent()-&gt;getInstList().insert(pi, newInst);
1843</pre>
1844</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1845
1846 <p>In fact, this sequence of steps occurs so frequently that the
1847 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
1848 constructors which take (as a default parameter) a pointer to an
1849 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
1850 precede. That is, <tt>Instruction</tt> constructors are capable of
1851 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
1852 provided instruction, immediately before that instruction. Using an
1853 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
1854 parameter, the above code becomes:</p>
1855
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1856<div class="doc_code">
1857<pre>
1858Instruction* pi = ...;
1859Instruction* newInst = new Instruction(..., pi);
1860</pre>
1861</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1862
1863 <p>which is much cleaner, especially if you're creating a lot of
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1864 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1865</ul>
1866
1867</div>
1868
1869<!--_______________________________________________________________________-->
1870<div class="doc_subsubsection">
1871 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
1872</div>
1873
1874<div class="doc_text">
1875
1876<p>Deleting an instruction from an existing sequence of instructions that form a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1877<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1878you must have a pointer to the instruction that you wish to delete. Second, you
1879need to obtain the pointer to that instruction's basic block. You use the
1880pointer to the basic block to get its list of instructions and then use the
1881erase function to remove your instruction. For example:</p>
1882
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1883<div class="doc_code">
1884<pre>
1885<a href="#Instruction">Instruction</a> *I = .. ;
1886<a href="#BasicBlock">BasicBlock</a> *BB = I-&gt;getParent();
1887
1888BB-&gt;getInstList().erase(I);
1889</pre>
1890</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1891
1892</div>
1893
1894<!--_______________________________________________________________________-->
1895<div class="doc_subsubsection">
1896 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
1897 <tt>Value</tt></a>
1898</div>
1899
1900<div class="doc_text">
1901
1902<p><i>Replacing individual instructions</i></p>
1903
1904<p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1905permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1906and <tt>ReplaceInstWithInst</tt>.</p>
1907
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1908<h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1909
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1910<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1911 <li><tt>ReplaceInstWithValue</tt>
1912
1913 <p>This function replaces all uses (within a basic block) of a given
1914 instruction with a value, and then removes the original instruction. The
1915 following example illustrates the replacement of the result of a particular
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]1916 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1917 pointer to an integer.</p>
1918
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1919<div class="doc_code">
1920<pre>
1921AllocaInst* instToReplace = ...;
1922BasicBlock::iterator ii(instToReplace);
1923
1924ReplaceInstWithValue(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1925 Constant::getNullValue(PointerType::get(Type::Int32Ty)));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1926</pre></div></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1927
1928 <li><tt>ReplaceInstWithInst</tt>
1929
1930 <p>This function replaces a particular instruction with another
1931 instruction. The following example illustrates the replacement of one
1932 <tt>AllocaInst</tt> with another.</p>
1933
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1934<div class="doc_code">
1935<pre>
1936AllocaInst* instToReplace = ...;
1937BasicBlock::iterator ii(instToReplace);
1938
1939ReplaceInstWithInst(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1940 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1941</pre></div></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1942</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1943
1944<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
1945
1946<p>You can use <tt>Value::replaceAllUsesWith</tt> and
1947<tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]1948doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1949and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1950information.</p>
1951
1952<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
1953include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
1954ReplaceInstWithValue, ReplaceInstWithInst -->
1955
1956</div>
1957
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]1958<!--_______________________________________________________________________-->
1959<div class="doc_subsubsection">
1960 <a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
1961</div>
1962
1963<div class="doc_text">
1964
Tanya Lattnerc5dfcdb2007-06-20 20:46:37 +0000[diff] [blame]1965<p>Deleting a global variable from a module is just as easy as deleting an
1966Instruction. First, you must have a pointer to the global variable that you wish
1967 to delete. You use this pointer to erase it from its parent, the module.
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]1968 For example:</p>
1969
1970<div class="doc_code">
1971<pre>
1972<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]1973
Tanya Lattnerc5dfcdb2007-06-20 20:46:37 +0000[diff] [blame]1974GV-&gt;eraseFromParent();
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]1975</pre>
1976</div>
1977
1978</div>
1979
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]1980<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1981<div class="doc_section">
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]1982 <a name="advanced">Advanced Topics</a>
1983</div>
1984<!-- *********************************************************************** -->
1985
1986<div class="doc_text">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1987<p>
1988This section describes some of the advanced or obscure API's that most clients
1989do not need to be aware of. These API's tend manage the inner workings of the
1990LLVM system, and only need to be accessed in unusual circumstances.
1991</p>
1992</div>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]1993
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1994<!-- ======================================================================= -->
1995<div class="doc_subsection">
1996 <a name="TypeResolve">LLVM Type Resolution</a>
1997</div>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]1998
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1999<div class="doc_text">
2000
2001<p>
2002The LLVM type system has a very simple goal: allow clients to compare types for
2003structural equality with a simple pointer comparison (aka a shallow compare).
2004This goal makes clients much simpler and faster, and is used throughout the LLVM
2005system.
2006</p>
2007
2008<p>
2009Unfortunately achieving this goal is not a simple matter. In particular,
2010recursive types and late resolution of opaque types makes the situation very
2011difficult to handle. Fortunately, for the most part, our implementation makes
2012most clients able to be completely unaware of the nasty internal details. The
2013primary case where clients are exposed to the inner workings of it are when
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]2014building a recursive type. In addition to this case, the LLVM bitcode reader,
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2015assembly parser, and linker also have to be aware of the inner workings of this
2016system.
2017</p>
2018
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]2019<p>
2020For our purposes below, we need three concepts. First, an "Opaque Type" is
2021exactly as defined in the <a href="LangRef.html#t_opaque">language
2022reference</a>. Second an "Abstract Type" is any type which includes an
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2023opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
2024Third, a concrete type is a type that is not an abstract type (e.g. "<tt>{ i32,
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]2025float }</tt>").
2026</p>
2027
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2028</div>
2029
2030<!-- ______________________________________________________________________ -->
2031<div class="doc_subsubsection">
2032 <a name="BuildRecType">Basic Recursive Type Construction</a>
2033</div>
2034
2035<div class="doc_text">
2036
2037<p>
2038Because the most common question is "how do I build a recursive type with LLVM",
2039we answer it now and explain it as we go. Here we include enough to cause this
2040to be emitted to an output .ll file:
2041</p>
2042
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2043<div class="doc_code">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2044<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2045%mylist = type { %mylist*, i32 }
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2046</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2047</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2048
2049<p>
2050To build this, use the following LLVM APIs:
2051</p>
2052
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2053<div class="doc_code">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2054<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2055// <i>Create the initial outer struct</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2056<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
2057std::vector&lt;const Type*&gt; Elts;
2058Elts.push_back(PointerType::get(StructTy));
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]2059Elts.push_back(Type::Int32Ty);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2060StructType *NewSTy = StructType::get(Elts);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2061
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2062// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2063// <i>the struct and the opaque type are actually the same.</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2064cast&lt;OpaqueType&gt;(StructTy.get())-&gt;<a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2065
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2066// <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2067// <i>kept up-to-date</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2068NewSTy = cast&lt;StructType&gt;(StructTy.get());
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2069
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2070// <i>Add a name for the type to the module symbol table (optional)</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2071MyModule-&gt;addTypeName("mylist", NewSTy);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2072</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2073</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2074
2075<p>
2076This code shows the basic approach used to build recursive types: build a
2077non-recursive type using 'opaque', then use type unification to close the cycle.
2078The type unification step is performed by the <tt><a
Chris Lattneraff26d12007-02-03 03:06:52 +0000[diff] [blame]2079href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2080described next. After that, we describe the <a
2081href="#PATypeHolder">PATypeHolder class</a>.
2082</p>
2083
2084</div>
2085
2086<!-- ______________________________________________________________________ -->
2087<div class="doc_subsubsection">
2088 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
2089</div>
2090
2091<div class="doc_text">
2092<p>
2093The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
2094While this method is actually a member of the DerivedType class, it is most
2095often used on OpaqueType instances. Type unification is actually a recursive
2096process. After unification, types can become structurally isomorphic to
2097existing types, and all duplicates are deleted (to preserve pointer equality).
2098</p>
2099
2100<p>
2101In the example above, the OpaqueType object is definitely deleted.
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2102Additionally, if there is an "{ \2*, i32}" type already created in the system,
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2103the pointer and struct type created are <b>also</b> deleted. Obviously whenever
2104a type is deleted, any "Type*" pointers in the program are invalidated. As
2105such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
2106live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
2107types can never move or be deleted). To deal with this, the <a
2108href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
2109reference to a possibly refined type, and the <a
2110href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
2111complex datastructures.
2112</p>
2113
2114</div>
2115
2116<!-- ______________________________________________________________________ -->
2117<div class="doc_subsubsection">
2118 <a name="PATypeHolder">The PATypeHolder Class</a>
2119</div>
2120
2121<div class="doc_text">
2122<p>
2123PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
2124happily goes about nuking types that become isomorphic to existing types, it
2125automatically updates all PATypeHolder objects to point to the new type. In the
2126example above, this allows the code to maintain a pointer to the resultant
2127resolved recursive type, even though the Type*'s are potentially invalidated.
2128</p>
2129
2130<p>
2131PATypeHolder is an extremely light-weight object that uses a lazy union-find
2132implementation to update pointers. For example the pointer from a Value to its
2133Type is maintained by PATypeHolder objects.
2134</p>
2135
2136</div>
2137
2138<!-- ______________________________________________________________________ -->
2139<div class="doc_subsubsection">
2140 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
2141</div>
2142
2143<div class="doc_text">
2144
2145<p>
2146Some data structures need more to perform more complex updates when types get
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2147resolved. To support this, a class can derive from the AbstractTypeUser class.
2148This class
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2149allows it to get callbacks when certain types are resolved. To register to get
2150callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]2151methods can be called on a type. Note that these methods only work for <i>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2152 abstract</i> types. Concrete types (those that do not include any opaque
2153objects) can never be refined.
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2154</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2155</div>
2156
2157
2158<!-- ======================================================================= -->
2159<div class="doc_subsection">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2160 <a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
2161 <tt>TypeSymbolTable</tt> classes</a>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2162</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2163
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2164<div class="doc_text">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2165<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
2166ValueSymbolTable</a></tt> class provides a symbol table that the <a
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2167href="#Function"><tt>Function</tt></a> and <a href="#Module">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2168<tt>Module</tt></a> classes use for naming value definitions. The symbol table
2169can provide a name for any <a href="#Value"><tt>Value</tt></a>.
2170The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
2171TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
2172names for types.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2173
Reid Spencera6362242007-01-07 00:41:39 +0000[diff] [blame]2174<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
2175by most clients. It should only be used when iteration over the symbol table
2176names themselves are required, which is very special purpose. Note that not
2177all LLVM
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2178<a href="#Value">Value</a>s have names, and those without names (i.e. they have
2179an empty name) do not exist in the symbol table.
2180</p>
2181
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2182<p>These symbol tables support iteration over the values/types in the symbol
2183table with <tt>begin/end/iterator</tt> and supports querying to see if a
2184specific name is in the symbol table (with <tt>lookup</tt>). The
2185<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
2186simply call <tt>setName</tt> on a value, which will autoinsert it into the
2187appropriate symbol table. For types, use the Module::addTypeName method to
2188insert entries into the symbol table.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2189
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2190</div>
2191
2192
2193
2194<!-- *********************************************************************** -->
2195<div class="doc_section">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2196 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
2197</div>
2198<!-- *********************************************************************** -->
2199
2200<div class="doc_text">
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2201<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
2202<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2203
2204<p>The Core LLVM classes are the primary means of representing the program
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2205being inspected or transformed. The core LLVM classes are defined in
2206header files in the <tt>include/llvm/</tt> directory, and implemented in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2207the <tt>lib/VMCore</tt> directory.</p>
2208
2209</div>
2210
2211<!-- ======================================================================= -->
2212<div class="doc_subsection">
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2213 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
2214</div>
2215
2216<div class="doc_text">
2217
2218 <p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
2219 a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
2220 through its subclasses. Certain primitive types (<tt>VoidType</tt>,
2221 <tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
2222 subclasses. They are hidden because they offer no useful functionality beyond
2223 what the <tt>Type</tt> class offers except to distinguish themselves from
2224 other subclasses of <tt>Type</tt>.</p>
2225 <p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
2226 named, but this is not a requirement. There exists exactly
2227 one instance of a given shape at any one time. This allows type equality to
2228 be performed with address equality of the Type Instance. That is, given two
2229 <tt>Type*</tt> values, the types are identical if the pointers are identical.
2230 </p>
2231</div>
2232
2233<!-- _______________________________________________________________________ -->
2234<div class="doc_subsubsection">
2235 <a name="m_Value">Important Public Methods</a>
2236</div>
2237
2238<div class="doc_text">
2239
2240<ul>
Chris Lattner8f79df32007-01-15 01:55:32 +0000[diff] [blame]2241 <li><tt>bool isInteger() const</tt>: Returns true for any integer type.</li>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2242
2243 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
2244 floating point types.</li>
2245
2246 <li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
2247 an OpaqueType anywhere in its definition).</li>
2248
2249 <li><tt>bool isSized()</tt>: Return true if the type has known size. Things
2250 that don't have a size are abstract types, labels and void.</li>
2251
2252</ul>
2253</div>
2254
2255<!-- _______________________________________________________________________ -->
2256<div class="doc_subsubsection">
2257 <a name="m_Value">Important Derived Types</a>
2258</div>
2259<div class="doc_text">
2260<dl>
2261 <dt><tt>IntegerType</tt></dt>
2262 <dd>Subclass of DerivedType that represents integer types of any bit width.
2263 Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
2264 <tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
2265 <ul>
2266 <li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
2267 type of a specific bit width.</li>
2268 <li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
2269 type.</li>
2270 </ul>
2271 </dd>
2272 <dt><tt>SequentialType</tt></dt>
2273 <dd>This is subclassed by ArrayType and PointerType
2274 <ul>
2275 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
2276 of the elements in the sequential type. </li>
2277 </ul>
2278 </dd>
2279 <dt><tt>ArrayType</tt></dt>
2280 <dd>This is a subclass of SequentialType and defines the interface for array
2281 types.
2282 <ul>
2283 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
2284 elements in the array. </li>
2285 </ul>
2286 </dd>
2287 <dt><tt>PointerType</tt></dt>
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]2288 <dd>Subclass of SequentialType for pointer types.</dd>
Reid Spencer9d6565a2007-02-15 02:26:10 +0000[diff] [blame]2289 <dt><tt>VectorType</tt></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2290 <dd>Subclass of SequentialType for vector types. A
2291 vector type is similar to an ArrayType but is distinguished because it is
2292 a first class type wherease ArrayType is not. Vector types are used for
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2293 vector operations and are usually small vectors of of an integer or floating
2294 point type.</dd>
2295 <dt><tt>StructType</tt></dt>
2296 <dd>Subclass of DerivedTypes for struct types.</dd>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2297 <dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2298 <dd>Subclass of DerivedTypes for function types.
2299 <ul>
2300 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
2301 function</li>
2302 <li><tt> const Type * getReturnType() const</tt>: Returns the
2303 return type of the function.</li>
2304 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
2305 the type of the ith parameter.</li>
2306 <li><tt> const unsigned getNumParams() const</tt>: Returns the
2307 number of formal parameters.</li>
2308 </ul>
2309 </dd>
2310 <dt><tt>OpaqueType</tt></dt>
2311 <dd>Sublcass of DerivedType for abstract types. This class
2312 defines no content and is used as a placeholder for some other type. Note
2313 that OpaqueType is used (temporarily) during type resolution for forward
2314 references of types. Once the referenced type is resolved, the OpaqueType
2315 is replaced with the actual type. OpaqueType can also be used for data
2316 abstraction. At link time opaque types can be resolved to actual types
2317 of the same name.</dd>
2318</dl>
2319</div>
2320
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2321
2322
2323<!-- ======================================================================= -->
2324<div class="doc_subsection">
2325 <a name="Module">The <tt>Module</tt> class</a>
2326</div>
2327
2328<div class="doc_text">
2329
2330<p><tt>#include "<a
2331href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
2332<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
2333
2334<p>The <tt>Module</tt> class represents the top level structure present in LLVM
2335programs. An LLVM module is effectively either a translation unit of the
2336original program or a combination of several translation units merged by the
2337linker. The <tt>Module</tt> class keeps track of a list of <a
2338href="#Function"><tt>Function</tt></a>s, a list of <a
2339href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
2340href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
2341helpful member functions that try to make common operations easy.</p>
2342
2343</div>
2344
2345<!-- _______________________________________________________________________ -->
2346<div class="doc_subsubsection">
2347 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
2348</div>
2349
2350<div class="doc_text">
2351
2352<ul>
2353 <li><tt>Module::Module(std::string name = "")</tt></li>
2354</ul>
2355
2356<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
2357provide a name for it (probably based on the name of the translation unit).</p>
2358
2359<ul>
2360 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
2361 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
2362
2363 <tt>begin()</tt>, <tt>end()</tt>
2364 <tt>size()</tt>, <tt>empty()</tt>
2365
2366 <p>These are forwarding methods that make it easy to access the contents of
2367 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
2368 list.</p></li>
2369
2370 <li><tt>Module::FunctionListType &amp;getFunctionList()</tt>
2371
2372 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
2373 necessary to use when you need to update the list or perform a complex
2374 action that doesn't have a forwarding method.</p>
2375
2376 <p><!-- Global Variable --></p></li>
2377</ul>
2378
2379<hr>
2380
2381<ul>
2382 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
2383
2384 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
2385
2386 <tt>global_begin()</tt>, <tt>global_end()</tt>
2387 <tt>global_size()</tt>, <tt>global_empty()</tt>
2388
2389 <p> These are forwarding methods that make it easy to access the contents of
2390 a <tt>Module</tt> object's <a
2391 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
2392
2393 <li><tt>Module::GlobalListType &amp;getGlobalList()</tt>
2394
2395 <p>Returns the list of <a
2396 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
2397 use when you need to update the list or perform a complex action that
2398 doesn't have a forwarding method.</p>
2399
2400 <p><!-- Symbol table stuff --> </p></li>
2401</ul>
2402
2403<hr>
2404
2405<ul>
2406 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
2407
2408 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2409 for this <tt>Module</tt>.</p>
2410
2411 <p><!-- Convenience methods --></p></li>
2412</ul>
2413
2414<hr>
2415
2416<ul>
2417 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
2418 &amp;Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
2419
2420 <p>Look up the specified function in the <tt>Module</tt> <a
2421 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
2422 <tt>null</tt>.</p></li>
2423
2424 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
2425 std::string &amp;Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
2426
2427 <p>Look up the specified function in the <tt>Module</tt> <a
2428 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
2429 external declaration for the function and return it.</p></li>
2430
2431 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
2432
2433 <p>If there is at least one entry in the <a
2434 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
2435 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
2436 string.</p></li>
2437
2438 <li><tt>bool addTypeName(const std::string &amp;Name, const <a
2439 href="#Type">Type</a> *Ty)</tt>
2440
2441 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2442 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
2443 name, true is returned and the <a
2444 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
2445</ul>
2446
2447</div>
2448
2449
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2450<!-- ======================================================================= -->
2451<div class="doc_subsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2452 <a name="Value">The <tt>Value</tt> class</a>
2453</div>
2454
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2455<div class="doc_text">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2456
2457<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
2458<br>
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]2459doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2460
2461<p>The <tt>Value</tt> class is the most important class in the LLVM Source
2462base. It represents a typed value that may be used (among other things) as an
2463operand to an instruction. There are many different types of <tt>Value</tt>s,
2464such as <a href="#Constant"><tt>Constant</tt></a>s,<a
2465href="#Argument"><tt>Argument</tt></a>s. Even <a
2466href="#Instruction"><tt>Instruction</tt></a>s and <a
2467href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
2468
2469<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
2470for a program. For example, an incoming argument to a function (represented
2471with an instance of the <a href="#Argument">Argument</a> class) is "used" by
2472every instruction in the function that references the argument. To keep track
2473of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
2474href="#User"><tt>User</tt></a>s that is using it (the <a
2475href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
2476graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
2477def-use information in the program, and is accessible through the <tt>use_</tt>*
2478methods, shown below.</p>
2479
2480<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
2481and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
2482method. In addition, all LLVM values can be named. The "name" of the
2483<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
2484
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2485<div class="doc_code">
2486<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2487%<b>foo</b> = add i32 1, 2
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2488</pre>
2489</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2490
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2491<p><a name="nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2492that the name of any value may be missing (an empty string), so names should
2493<b>ONLY</b> be used for debugging (making the source code easier to read,
2494debugging printouts), they should not be used to keep track of values or map
2495between them. For this purpose, use a <tt>std::map</tt> of pointers to the
2496<tt>Value</tt> itself instead.</p>
2497
2498<p>One important aspect of LLVM is that there is no distinction between an SSA
2499variable and the operation that produces it. Because of this, any reference to
2500the value produced by an instruction (or the value available as an incoming
Chris Lattnerd5fc4fc2004-03-18 14:58:55 +0000[diff] [blame]2501argument, for example) is represented as a direct pointer to the instance of
2502the class that
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2503represents this value. Although this may take some getting used to, it
2504simplifies the representation and makes it easier to manipulate.</p>
2505
2506</div>
2507
2508<!-- _______________________________________________________________________ -->
2509<div class="doc_subsubsection">
2510 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
2511</div>
2512
2513<div class="doc_text">
2514
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2515<ul>
2516 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
2517use-list<br>
2518 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
2519the use-list<br>
2520 <tt>unsigned use_size()</tt> - Returns the number of users of the
2521value.<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2522 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2523 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
2524the use-list.<br>
2525 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
2526use-list.<br>
2527 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
2528element in the list.
2529 <p> These methods are the interface to access the def-use
2530information in LLVM. As with all other iterators in LLVM, the naming
2531conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2532 </li>
2533 <li><tt><a href="#Type">Type</a> *getType() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2534 <p>This method returns the Type of the Value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2535 </li>
2536 <li><tt>bool hasName() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2537 <tt>std::string getName() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2538 <tt>void setName(const std::string &amp;Name)</tt>
2539 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
2540be aware of the <a href="#nameWarning">precaution above</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2541 </li>
2542 <li><tt>void replaceAllUsesWith(Value *V)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2543
2544 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
2545 href="#User"><tt>User</tt>s</a> of the current value to refer to
2546 "<tt>V</tt>" instead. For example, if you detect that an instruction always
2547 produces a constant value (for example through constant folding), you can
2548 replace all uses of the instruction with the constant like this:</p>
2549
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2550<div class="doc_code">
2551<pre>
2552Inst-&gt;replaceAllUsesWith(ConstVal);
2553</pre>
2554</div>
2555
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2556</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2557
2558</div>
2559
2560<!-- ======================================================================= -->
2561<div class="doc_subsection">
2562 <a name="User">The <tt>User</tt> class</a>
2563</div>
2564
2565<div class="doc_text">
2566
2567<p>
2568<tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]2569doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2570Superclass: <a href="#Value"><tt>Value</tt></a></p>
2571
2572<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
2573refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
2574that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
2575referring to. The <tt>User</tt> class itself is a subclass of
2576<tt>Value</tt>.</p>
2577
2578<p>The operands of a <tt>User</tt> point directly to the LLVM <a
2579href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
2580Single Assignment (SSA) form, there can only be one definition referred to,
2581allowing this direct connection. This connection provides the use-def
2582information in LLVM.</p>
2583
2584</div>
2585
2586<!-- _______________________________________________________________________ -->
2587<div class="doc_subsubsection">
2588 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
2589</div>
2590
2591<div class="doc_text">
2592
2593<p>The <tt>User</tt> class exposes the operand list in two ways: through
2594an index access interface and through an iterator based interface.</p>
2595
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2596<ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2597 <li><tt>Value *getOperand(unsigned i)</tt><br>
2598 <tt>unsigned getNumOperands()</tt>
2599 <p> These two methods expose the operands of the <tt>User</tt> in a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2600convenient form for direct access.</p></li>
2601
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2602 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
2603list<br>
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]2604 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
2605the operand list.<br>
2606 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2607operand list.
2608 <p> Together, these methods make up the iterator based interface to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2609the operands of a <tt>User</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2610</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2611
2612</div>
2613
2614<!-- ======================================================================= -->
2615<div class="doc_subsection">
2616 <a name="Instruction">The <tt>Instruction</tt> class</a>
2617</div>
2618
2619<div class="doc_text">
2620
2621<p><tt>#include "</tt><tt><a
2622href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2623doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2624Superclasses: <a href="#User"><tt>User</tt></a>, <a
2625href="#Value"><tt>Value</tt></a></p>
2626
2627<p>The <tt>Instruction</tt> class is the common base class for all LLVM
2628instructions. It provides only a few methods, but is a very commonly used
2629class. The primary data tracked by the <tt>Instruction</tt> class itself is the
2630opcode (instruction type) and the parent <a
2631href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
2632into. To represent a specific type of instruction, one of many subclasses of
2633<tt>Instruction</tt> are used.</p>
2634
2635<p> Because the <tt>Instruction</tt> class subclasses the <a
2636href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
2637way as for other <a href="#User"><tt>User</tt></a>s (with the
2638<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
2639<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
2640the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
2641file contains some meta-data about the various different types of instructions
2642in LLVM. It describes the enum values that are used as opcodes (for example
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2643<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2644concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
2645example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2646href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2647this file confuses doxygen, so these enum values don't show up correctly in the
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2648<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2649
2650</div>
2651
2652<!-- _______________________________________________________________________ -->
2653<div class="doc_subsubsection">
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2654 <a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
2655 class</a>
2656</div>
2657<div class="doc_text">
2658 <ul>
2659 <li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
2660 <p>This subclasses represents all two operand instructions whose operands
2661 must be the same type, except for the comparison instructions.</p></li>
2662 <li><tt><a name="CastInst">CastInst</a></tt>
2663 <p>This subclass is the parent of the 12 casting instructions. It provides
2664 common operations on cast instructions.</p>
2665 <li><tt><a name="CmpInst">CmpInst</a></tt>
2666 <p>This subclass respresents the two comparison instructions,
2667 <a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
2668 <a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
2669 <li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
2670 <p>This subclass is the parent of all terminator instructions (those which
2671 can terminate a block).</p>
2672 </ul>
2673 </div>
2674
2675<!-- _______________________________________________________________________ -->
2676<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2677 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
2678 class</a>
2679</div>
2680
2681<div class="doc_text">
2682
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2683<ul>
2684 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2685 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
2686this <tt>Instruction</tt> is embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2687 <li><tt>bool mayWriteToMemory()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2688 <p>Returns true if the instruction writes to memory, i.e. it is a
2689 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2690 <li><tt>unsigned getOpcode()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2691 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2692 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2693 <p>Returns another instance of the specified instruction, identical
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2694in all ways to the original except that the instruction has no parent
2695(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2696and it has no name</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2697</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2698
2699</div>
2700
2701<!-- ======================================================================= -->
2702<div class="doc_subsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2703 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2704</div>
2705
2706<div class="doc_text">
2707
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2708<p>Constant represents a base class for different types of constants. It
2709is subclassed by ConstantInt, ConstantArray, etc. for representing
2710the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
2711a subclass, which represents the address of a global variable or function.
2712</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2713
2714</div>
2715
2716<!-- _______________________________________________________________________ -->
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2717<div class="doc_subsubsection">Important Subclasses of Constant </div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2718<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2719<ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2720 <li>ConstantInt : This subclass of Constant represents an integer constant of
2721 any width.
2722 <ul>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]2723 <li><tt>const APInt&amp; getValue() const</tt>: Returns the underlying
2724 value of this constant, an APInt value.</li>
2725 <li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
2726 value to an int64_t via sign extension. If the value (not the bit width)
2727 of the APInt is too large to fit in an int64_t, an assertion will result.
2728 For this reason, use of this method is discouraged.</li>
2729 <li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
2730 value to a uint64_t via zero extension. IF the value (not the bit width)
2731 of the APInt is too large to fit in a uint64_t, an assertion will result.
Reid Spencer4474d872007-03-02 01:31:31 +0000[diff] [blame]2732 For this reason, use of this method is discouraged.</li>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]2733 <li><tt>static ConstantInt* get(const APInt&amp; Val)</tt>: Returns the
2734 ConstantInt object that represents the value provided by <tt>Val</tt>.
2735 The type is implied as the IntegerType that corresponds to the bit width
2736 of <tt>Val</tt>.</li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2737 <li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
2738 Returns the ConstantInt object that represents the value provided by
2739 <tt>Val</tt> for integer type <tt>Ty</tt>.</li>
2740 </ul>
2741 </li>
2742 <li>ConstantFP : This class represents a floating point constant.
2743 <ul>
2744 <li><tt>double getValue() const</tt>: Returns the underlying value of
2745 this constant. </li>
2746 </ul>
2747 </li>
2748 <li>ConstantArray : This represents a constant array.
2749 <ul>
2750 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
2751 a vector of component constants that makeup this array. </li>
2752 </ul>
2753 </li>
2754 <li>ConstantStruct : This represents a constant struct.
2755 <ul>
2756 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
2757 a vector of component constants that makeup this array. </li>
2758 </ul>
2759 </li>
2760 <li>GlobalValue : This represents either a global variable or a function. In
2761 either case, the value is a constant fixed address (after linking).
2762 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2763</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2764</div>
2765
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2766
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2767<!-- ======================================================================= -->
2768<div class="doc_subsection">
2769 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
2770</div>
2771
2772<div class="doc_text">
2773
2774<p><tt>#include "<a
2775href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]2776doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
2777Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2778Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
2779<a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2780
2781<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
2782href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
2783visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
2784Because they are visible at global scope, they are also subject to linking with
2785other globals defined in different translation units. To control the linking
2786process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
2787<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]2788defined by the <tt>LinkageTypes</tt> enumeration.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2789
2790<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
2791<tt>static</tt> in C), it is not visible to code outside the current translation
2792unit, and does not participate in linking. If it has external linkage, it is
2793visible to external code, and does participate in linking. In addition to
2794linkage information, <tt>GlobalValue</tt>s keep track of which <a
2795href="#Module"><tt>Module</tt></a> they are currently part of.</p>
2796
2797<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
2798by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
2799global is always a pointer to its contents. It is important to remember this
2800when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
2801be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
2802subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2803i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2804the address of the first element of this array and the value of the
2805<tt>GlobalVariable</tt> are the same, they have different types. The
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2806<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
2807is <tt>i32.</tt> Because of this, accessing a global value requires you to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2808dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
2809can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
2810Language Reference Manual</a>.</p>
2811
2812</div>
2813
2814<!-- _______________________________________________________________________ -->
2815<div class="doc_subsubsection">
2816 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
2817 class</a>
2818</div>
2819
2820<div class="doc_text">
2821
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2822<ul>
2823 <li><tt>bool hasInternalLinkage() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2824 <tt>bool hasExternalLinkage() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2825 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
2826 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
2827 <p> </p>
2828 </li>
2829 <li><tt><a href="#Module">Module</a> *getParent()</tt>
2830 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2831GlobalValue is currently embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2832</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2833
2834</div>
2835
2836<!-- ======================================================================= -->
2837<div class="doc_subsection">
2838 <a name="Function">The <tt>Function</tt> class</a>
2839</div>
2840
2841<div class="doc_text">
2842
2843<p><tt>#include "<a
2844href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2845info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2846Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
2847<a href="#Constant"><tt>Constant</tt></a>,
2848<a href="#User"><tt>User</tt></a>,
2849<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2850
2851<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
2852actually one of the more complex classes in the LLVM heirarchy because it must
2853keep track of a large amount of data. The <tt>Function</tt> class keeps track
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2854of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
2855<a href="#Argument"><tt>Argument</tt></a>s, and a
2856<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2857
2858<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
2859commonly used part of <tt>Function</tt> objects. The list imposes an implicit
2860ordering of the blocks in the function, which indicate how the code will be
2861layed out by the backend. Additionally, the first <a
2862href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
2863<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
2864block. There are no implicit exit nodes, and in fact there may be multiple exit
2865nodes from a single <tt>Function</tt>. If the <a
2866href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
2867the <tt>Function</tt> is actually a function declaration: the actual body of the
2868function hasn't been linked in yet.</p>
2869
2870<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
2871<tt>Function</tt> class also keeps track of the list of formal <a
2872href="#Argument"><tt>Argument</tt></a>s that the function receives. This
2873container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
2874nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
2875the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
2876
2877<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
2878LLVM feature that is only used when you have to look up a value by name. Aside
2879from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
2880internally to make sure that there are not conflicts between the names of <a
2881href="#Instruction"><tt>Instruction</tt></a>s, <a
2882href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
2883href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
2884
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]2885<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
2886and therefore also a <a href="#Constant">Constant</a>. The value of the function
2887is its address (after linking) which is guaranteed to be constant.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2888</div>
2889
2890<!-- _______________________________________________________________________ -->
2891<div class="doc_subsubsection">
2892 <a name="m_Function">Important Public Members of the <tt>Function</tt>
2893 class</a>
2894</div>
2895
2896<div class="doc_text">
2897
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2898<ul>
2899 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]2900 *Ty, LinkageTypes Linkage, const std::string &amp;N = "", Module* Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2901
2902 <p>Constructor used when you need to create new <tt>Function</tt>s to add
2903 the the program. The constructor must specify the type of the function to
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]2904 create and what type of linkage the function should have. The <a
2905 href="#FunctionType"><tt>FunctionType</tt></a> argument
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2906 specifies the formal arguments and return value for the function. The same
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2907 <a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2908 create multiple functions. The <tt>Parent</tt> argument specifies the Module
2909 in which the function is defined. If this argument is provided, the function
2910 will automatically be inserted into that module's list of
2911 functions.</p></li>
2912
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2913 <li><tt>bool isExternal()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2914
2915 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
2916 function is "external", it does not have a body, and thus must be resolved
2917 by linking with a function defined in a different translation unit.</p></li>
2918
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2919 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2920 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2921
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]2922 <tt>begin()</tt>, <tt>end()</tt>
2923 <tt>size()</tt>, <tt>empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2924
2925 <p>These are forwarding methods that make it easy to access the contents of
2926 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
2927 list.</p></li>
2928
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2929 <li><tt>Function::BasicBlockListType &amp;getBasicBlockList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2930
2931 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
2932 is necessary to use when you need to update the list or perform a complex
2933 action that doesn't have a forwarding method.</p></li>
2934
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]2935 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2936iterator<br>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]2937 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2938
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]2939 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]2940 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2941
2942 <p>These are forwarding methods that make it easy to access the contents of
2943 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
2944 list.</p></li>
2945
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2946 <li><tt>Function::ArgumentListType &amp;getArgumentList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2947
2948 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
2949 necessary to use when you need to update the list or perform a complex
2950 action that doesn't have a forwarding method.</p></li>
2951
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2952 <li><tt><a href="#BasicBlock">BasicBlock</a> &amp;getEntryBlock()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2953
2954 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
2955 function. Because the entry block for the function is always the first
2956 block, this returns the first block of the <tt>Function</tt>.</p></li>
2957
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2958 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
2959 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2960
2961 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
2962 <tt>Function</tt> and returns the return type of the function, or the <a
2963 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
2964 function.</p></li>
2965
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2966 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2967
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2968 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2969 for this <tt>Function</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2970</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2971
2972</div>
2973
2974<!-- ======================================================================= -->
2975<div class="doc_subsection">
2976 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
2977</div>
2978
2979<div class="doc_text">
2980
2981<p><tt>#include "<a
2982href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
2983<br>
Tanya Lattnera3da7772004-06-22 08:02:25 +0000[diff] [blame]2984doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2985 Class</a><br>
2986Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
2987<a href="#Constant"><tt>Constant</tt></a>,
2988<a href="#User"><tt>User</tt></a>,
2989<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2990
2991<p>Global variables are represented with the (suprise suprise)
2992<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
2993subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
2994always referenced by their address (global values must live in memory, so their
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2995"name" refers to their constant address). See
2996<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
2997variables may have an initial value (which must be a
2998<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
2999they may be marked as "constant" themselves (indicating that their contents
3000never change at runtime).</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3001</div>
3002
3003<!-- _______________________________________________________________________ -->
3004<div class="doc_subsubsection">
3005 <a name="m_GlobalVariable">Important Public Members of the
3006 <tt>GlobalVariable</tt> class</a>
3007</div>
3008
3009<div class="doc_text">
3010
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3011<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3012 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
3013 isConstant, LinkageTypes&amp; Linkage, <a href="#Constant">Constant</a>
3014 *Initializer = 0, const std::string &amp;Name = "", Module* Parent = 0)</tt>
3015
3016 <p>Create a new global variable of the specified type. If
3017 <tt>isConstant</tt> is true then the global variable will be marked as
3018 unchanging for the program. The Linkage parameter specifies the type of
3019 linkage (internal, external, weak, linkonce, appending) for the variable. If
3020 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage,&nbsp; then
3021 the resultant global variable will have internal linkage. AppendingLinkage
3022 concatenates together all instances (in different translation units) of the
3023 variable into a single variable but is only applicable to arrays. &nbsp;See
3024 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
3025 further details on linkage types. Optionally an initializer, a name, and the
3026 module to put the variable into may be specified for the global variable as
3027 well.</p></li>
3028
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3029 <li><tt>bool isConstant() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3030
3031 <p>Returns true if this is a global variable that is known not to
3032 be modified at runtime.</p></li>
3033
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3034 <li><tt>bool hasInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3035
3036 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
3037
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3038 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3039
3040 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
3041 to call this method if there is no initializer.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3042</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3043
3044</div>
3045
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3046
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3047<!-- ======================================================================= -->
3048<div class="doc_subsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3049 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3050</div>
3051
3052<div class="doc_text">
3053
3054<p><tt>#include "<a
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3055href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
3056doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
3057Class</a><br>
3058Superclass: <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3059
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3060<p>This class represents a single entry multiple exit section of the code,
3061commonly known as a basic block by the compiler community. The
3062<tt>BasicBlock</tt> class maintains a list of <a
3063href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
3064Matching the language definition, the last element of this list of instructions
3065is always a terminator instruction (a subclass of the <a
3066href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
3067
3068<p>In addition to tracking the list of instructions that make up the block, the
3069<tt>BasicBlock</tt> class also keeps track of the <a
3070href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
3071
3072<p>Note that <tt>BasicBlock</tt>s themselves are <a
3073href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
3074like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
3075<tt>label</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3076
3077</div>
3078
3079<!-- _______________________________________________________________________ -->
3080<div class="doc_subsubsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3081 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
3082 class</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3083</div>
3084
3085<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3086<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3087
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3088<li><tt>BasicBlock(const std::string &amp;Name = "", </tt><tt><a
3089 href="#Function">Function</a> *Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3090
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3091<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
3092insertion into a function. The constructor optionally takes a name for the new
3093block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
3094the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
3095automatically inserted at the end of the specified <a
3096href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
3097manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3098
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3099<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
3100<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
3101<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
3102<tt>size()</tt>, <tt>empty()</tt>
3103STL-style functions for accessing the instruction list.
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3104
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3105<p>These methods and typedefs are forwarding functions that have the same
3106semantics as the standard library methods of the same names. These methods
3107expose the underlying instruction list of a basic block in a way that is easy to
3108manipulate. To get the full complement of container operations (including
3109operations to update the list), you must use the <tt>getInstList()</tt>
3110method.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3111
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3112<li><tt>BasicBlock::InstListType &amp;getInstList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3113
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3114<p>This method is used to get access to the underlying container that actually
3115holds the Instructions. This method must be used when there isn't a forwarding
3116function in the <tt>BasicBlock</tt> class for the operation that you would like
3117to perform. Because there are no forwarding functions for "updating"
3118operations, you need to use this if you want to update the contents of a
3119<tt>BasicBlock</tt>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3120
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3121<li><tt><a href="#Function">Function</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3122
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3123<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
3124embedded into, or a null pointer if it is homeless.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3125
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3126<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3127
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3128<p> Returns a pointer to the terminator instruction that appears at the end of
3129the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
3130instruction in the block is not a terminator, then a null pointer is
3131returned.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3132
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3133</ul>
3134
3135</div>
3136
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3137
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3138<!-- ======================================================================= -->
3139<div class="doc_subsection">
3140 <a name="Argument">The <tt>Argument</tt> class</a>
3141</div>
3142
3143<div class="doc_text">
3144
3145<p>This subclass of Value defines the interface for incoming formal
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]3146arguments to a function. A Function maintains a list of its formal
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3147arguments. An argument has a pointer to the parent Function.</p>
3148
3149</div>
3150
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3151<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3152<hr>
3153<address>
3154 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3155 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3156 <a href="http://validator.w3.org/check/referer"><img
3157 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3158
3159 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
3160 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]3161 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3162 Last modified: $Date$
3163</address>
3164
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3165</body>
3166</html>