Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / b011c66da4412327619352f2b4a625b574b592b1 / . / docs / ProgrammersManual.html
blob: 571d4ba27aff5b73c5fd0f52e8d4371d42a16df4 [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>
65 <li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
66 <li><a href="#dss_set">&lt;set&gt;</a></li>
67 <li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]68 <li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
69 <li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]70 </ul></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]71 <li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
72 <ul>
73 <li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]74 <li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]75 <li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
76 <li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
77 <li><a href="#dss_map">&lt;map&gt;</a></li>
78 <li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
79 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]80 </ul>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]81 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]82 <li><a href="#common">Helpful Hints for Common Operations</a>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]83 <ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]84 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
85 <ul>
86 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
87in a <tt>Function</tt></a> </li>
88 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
89in a <tt>BasicBlock</tt></a> </li>
90 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
91in a <tt>Function</tt></a> </li>
92 <li><a href="#iterate_convert">Turning an iterator into a
93class pointer</a> </li>
94 <li><a href="#iterate_complex">Finding call sites: a more
95complex example</a> </li>
96 <li><a href="#calls_and_invokes">Treating calls and invokes
97the same way</a> </li>
98 <li><a href="#iterate_chains">Iterating over def-use &amp;
99use-def chains</a> </li>
100 </ul>
101 </li>
102 <li><a href="#simplechanges">Making simple changes</a>
103 <ul>
104 <li><a href="#schanges_creating">Creating and inserting new
105 <tt>Instruction</tt>s</a> </li>
106 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
107 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
108with another <tt>Value</tt></a> </li>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame^]109 <li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]110 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]111 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]112<!--
113 <li>Working with the Control Flow Graph
114 <ul>
115 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
116 <li>
117 <li>
118 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]119-->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]120 </ul>
121 </li>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]122
123 <li><a href="#advanced">Advanced Topics</a>
124 <ul>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]125 <li><a href="#TypeResolve">LLVM Type Resolution</a>
126 <ul>
127 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
128 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
129 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
130 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
131 </ul></li>
132
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]133 <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]134 </ul></li>
135
Joel Stanley9b96c442002-09-06 21:55:13 +0000[diff] [blame]136 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]137 <ul>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]138 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]139 <li><a href="#Module">The <tt>Module</tt> class</a></li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]140 <li><a href="#Value">The <tt>Value</tt> class</a>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]141 <ul>
142 <li><a href="#User">The <tt>User</tt> class</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]143 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]144 <li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
145 <li><a href="#Constant">The <tt>Constant</tt> class</a>
146 <ul>
147 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]148 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]149 <li><a href="#Function">The <tt>Function</tt> class</a></li>
150 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
151 </ul>
152 </li>
153 </ul>
154 </li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]155 </ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]156 </li>
157 <li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
158 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
159 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]160 </li>
161 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]162 </li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]163</ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]164
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]165<div class="doc_author">
166 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
Chris Lattner94c43592004-05-26 16:52:55 +0000[diff] [blame]167 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
168 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>, and
169 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]170</div>
171
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]172<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]173<div class="doc_section">
174 <a name="introduction">Introduction </a>
175</div>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]176<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]177
178<div class="doc_text">
179
180<p>This document is meant to highlight some of the important classes and
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]181interfaces available in the LLVM source-base. This manual is not
182intended to explain what LLVM is, how it works, and what LLVM code looks
183like. It assumes that you know the basics of LLVM and are interested
184in writing transformations or otherwise analyzing or manipulating the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]185code.</p>
186
187<p>This document should get you oriented so that you can find your
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]188way in the continuously growing source code that makes up the LLVM
189infrastructure. Note that this manual is not intended to serve as a
190replacement for reading the source code, so if you think there should be
191a method in one of these classes to do something, but it's not listed,
192check the source. Links to the <a href="/doxygen/">doxygen</a> sources
193are provided to make this as easy as possible.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]194
195<p>The first section of this document describes general information that is
196useful to know when working in the LLVM infrastructure, and the second describes
197the Core LLVM classes. In the future this manual will be extended with
198information describing how to use extension libraries, such as dominator
199information, CFG traversal routines, and useful utilities like the <tt><a
200href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
201
202</div>
203
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]204<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]205<div class="doc_section">
206 <a name="general">General Information</a>
207</div>
208<!-- *********************************************************************** -->
209
210<div class="doc_text">
211
212<p>This section contains general information that is useful if you are working
213in the LLVM source-base, but that isn't specific to any particular API.</p>
214
215</div>
216
217<!-- ======================================================================= -->
218<div class="doc_subsection">
219 <a name="stl">The C++ Standard Template Library</a>
220</div>
221
222<div class="doc_text">
223
224<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]225perhaps much more than you are used to, or have seen before. Because of
226this, you might want to do a little background reading in the
227techniques used and capabilities of the library. There are many good
228pages that discuss the STL, and several books on the subject that you
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]229can get, so it will not be discussed in this document.</p>
230
231<p>Here are some useful links:</p>
232
233<ol>
234
235<li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
236reference</a> - an excellent reference for the STL and other parts of the
237standard C++ library.</li>
238
239<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]240O'Reilly book in the making. It has a decent
241Standard Library
242Reference 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]243published.</li>
244
245<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
246Questions</a></li>
247
248<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
249Contains a useful <a
250href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
251STL</a>.</li>
252
253<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
254Page</a></li>
255
Tanya Lattner79445ba2004-12-08 18:34:56 +0000[diff] [blame]256<li><a href="http://64.78.49.204/">
Reid Spencer096603a2004-05-26 08:41:35 +0000[diff] [blame]257Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
258the book).</a></li>
259
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]260</ol>
261
262<p>You are also encouraged to take a look at the <a
263href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
264to write maintainable code more than where to put your curly braces.</p>
265
266</div>
267
268<!-- ======================================================================= -->
269<div class="doc_subsection">
270 <a name="stl">Other useful references</a>
271</div>
272
273<div class="doc_text">
274
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]275<ol>
276<li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]277Branch and Tag Primer</a></li>
Misha Brukmana0f71e42004-06-18 18:39:00 +0000[diff] [blame]278<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
279static and shared libraries across platforms</a></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]280</ol>
281
282</div>
283
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]284<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]285<div class="doc_section">
286 <a name="apis">Important and useful LLVM APIs</a>
287</div>
288<!-- *********************************************************************** -->
289
290<div class="doc_text">
291
292<p>Here we highlight some LLVM APIs that are generally useful and good to
293know about when writing transformations.</p>
294
295</div>
296
297<!-- ======================================================================= -->
298<div class="doc_subsection">
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]299 <a name="isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt> and
300 <tt>dyn_cast&lt;&gt;</tt> templates</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]301</div>
302
303<div class="doc_text">
304
305<p>The LLVM source-base makes extensive use of a custom form of RTTI.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]306These templates have many similarities to the C++ <tt>dynamic_cast&lt;&gt;</tt>
307operator, but they don't have some drawbacks (primarily stemming from
308the fact that <tt>dynamic_cast&lt;&gt;</tt> only works on classes that
309have a v-table). Because they are used so often, you must know what they
310do and how they work. All of these templates are defined in the <a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]311 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]312file (note that you very rarely have to include this file directly).</p>
313
314<dl>
315 <dt><tt>isa&lt;&gt;</tt>: </dt>
316
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]317 <dd><p>The <tt>isa&lt;&gt;</tt> operator works exactly like the Java
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]318 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
319 a reference or pointer points to an instance of the specified class. This can
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]320 be very useful for constraint checking of various sorts (example below).</p>
321 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]322
323 <dt><tt>cast&lt;&gt;</tt>: </dt>
324
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]325 <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]326 converts a pointer or reference from a base class to a derived cast, causing
327 an assertion failure if it is not really an instance of the right type. This
328 should be used in cases where you have some information that makes you believe
329 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]330 and <tt>cast&lt;&gt;</tt> template is:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]331
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]332<div class="doc_code">
333<pre>
334static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
335 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))
336 return true;
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]337
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]338 // <i>Otherwise, it must be an instruction...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]339 return !L-&gt;contains(cast&lt;<a href="#Instruction">Instruction</a>&gt;(V)-&gt;getParent());
340}
341</pre>
342</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]343
344 <p>Note that you should <b>not</b> use an <tt>isa&lt;&gt;</tt> test followed
345 by a <tt>cast&lt;&gt;</tt>, for that use the <tt>dyn_cast&lt;&gt;</tt>
346 operator.</p>
347
348 </dd>
349
350 <dt><tt>dyn_cast&lt;&gt;</tt>:</dt>
351
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]352 <dd><p>The <tt>dyn_cast&lt;&gt;</tt> operator is a "checking cast" operation.
353 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]354 pointer to it (this operator does not work with references). If the operand is
355 not of the correct type, a null pointer is returned. Thus, this works very
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]356 much like the <tt>dynamic_cast&lt;&gt;</tt> operator in C++, and should be
357 used in the same circumstances. Typically, the <tt>dyn_cast&lt;&gt;</tt>
358 operator is used in an <tt>if</tt> statement or some other flow control
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]359 statement like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]360
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]361<div class="doc_code">
362<pre>
363if (<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]364 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]365}
366</pre>
367</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]368
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]369 <p>This form of the <tt>if</tt> statement effectively combines together a call
370 to <tt>isa&lt;&gt;</tt> and a call to <tt>cast&lt;&gt;</tt> into one
371 statement, which is very convenient.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]372
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]373 <p>Note that the <tt>dyn_cast&lt;&gt;</tt> operator, like C++'s
374 <tt>dynamic_cast&lt;&gt;</tt> or Java's <tt>instanceof</tt> operator, can be
375 abused. In particular, you should not use big chained <tt>if/then/else</tt>
376 blocks to check for lots of different variants of classes. If you find
377 yourself wanting to do this, it is much cleaner and more efficient to use the
378 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]379
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]380 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]381
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]382 <dt><tt>cast_or_null&lt;&gt;</tt>: </dt>
383
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]384 <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]385 <tt>cast&lt;&gt;</tt> operator, except that it allows for a null pointer as an
386 argument (which it then propagates). This can sometimes be useful, allowing
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]387 you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]388
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]389 <dt><tt>dyn_cast_or_null&lt;&gt;</tt>: </dt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]390
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]391 <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]392 <tt>dyn_cast&lt;&gt;</tt> operator, except that it allows for a null pointer
393 as an argument (which it then propagates). This can sometimes be useful,
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]394 allowing you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]395
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]396</dl>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]397
398<p>These five templates can be used with any classes, whether they have a
399v-table or not. To add support for these templates, you simply need to add
400<tt>classof</tt> static methods to the class you are interested casting
401to. Describing this is currently outside the scope of this document, but there
402are lots of examples in the LLVM source base.</p>
403
404</div>
405
406<!-- ======================================================================= -->
407<div class="doc_subsection">
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]408 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]409</div>
410
411<div class="doc_text">
412
413<p>Often when working on your pass you will put a bunch of debugging printouts
414and other code into your pass. After you get it working, you want to remove
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]415it, 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]416across).</p>
417
418<p> Naturally, because of this, you don't want to delete the debug printouts,
419but you don't want them to always be noisy. A standard compromise is to comment
420them out, allowing you to enable them if you need them in the future.</p>
421
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]422<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]423file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
424this problem. Basically, you can put arbitrary code into the argument of the
425<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
426tool) is run with the '<tt>-debug</tt>' command line argument:</p>
427
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]428<div class="doc_code">
429<pre>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]430DOUT &lt;&lt; "I am here!\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]431</pre>
432</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]433
434<p>Then you can run your pass like this:</p>
435
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]436<div class="doc_code">
437<pre>
438$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]439<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]440$ opt &lt; a.bc &gt; /dev/null -mypass -debug
441I am here!
442</pre>
443</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]444
445<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
446to not have to create "yet another" command line option for the debug output for
447your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
448so they do not cause a performance impact at all (for the same reason, they
449should also not contain side-effects!).</p>
450
451<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
452enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
453"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
454program hasn't been started yet, you can always just run it with
455<tt>-debug</tt>.</p>
456
457</div>
458
459<!-- _______________________________________________________________________ -->
460<div class="doc_subsubsection">
Chris Lattnerc9151082005-04-26 22:57:07 +0000[diff] [blame]461 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]462 the <tt>-debug-only</tt> option</a>
463</div>
464
465<div class="doc_text">
466
467<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
468just turns on <b>too much</b> information (such as when working on the code
469generator). If you want to enable debug information with more fine-grained
470control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
471option as follows:</p>
472
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]473<div class="doc_code">
474<pre>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]475DOUT &lt;&lt; "No debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]476#undef DEBUG_TYPE
477#define DEBUG_TYPE "foo"
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]478DOUT &lt;&lt; "'foo' debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]479#undef DEBUG_TYPE
480#define DEBUG_TYPE "bar"
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]481DOUT &lt;&lt; "'bar' debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]482#undef DEBUG_TYPE
483#define DEBUG_TYPE ""
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]484DOUT &lt;&lt; "No debug type (2)\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]485</pre>
486</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]487
488<p>Then you can run your pass like this:</p>
489
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]490<div class="doc_code">
491<pre>
492$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]493<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]494$ opt &lt; a.bc &gt; /dev/null -mypass -debug
495No debug type
496'foo' debug type
497'bar' debug type
498No debug type (2)
499$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=foo
500'foo' debug type
501$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=bar
502'bar' debug type
503</pre>
504</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]505
506<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
507a 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]508you <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]509<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
510"bar", because there is no system in place to ensure that names do not
511conflict. If two different modules use the same string, they will all be turned
512on when the name is specified. This allows, for example, all debug information
513for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]514even if the source lives in multiple files.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]515
516</div>
517
518<!-- ======================================================================= -->
519<div class="doc_subsection">
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]520 <a name="Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]521 option</a>
522</div>
523
524<div class="doc_text">
525
526<p>The "<tt><a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]527href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]528provides a class named <tt>Statistic</tt> that is used as a unified way to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]529keep track of what the LLVM compiler is doing and how effective various
530optimizations are. It is useful to see what optimizations are contributing to
531making a particular program run faster.</p>
532
533<p>Often you may run your pass on some big program, and you're interested to see
534how many times it makes a certain transformation. Although you can do this with
535hand 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]536for big programs. Using the <tt>Statistic</tt> class makes it very easy to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]537keep track of this information, and the calculated information is presented in a
538uniform manner with the rest of the passes being executed.</p>
539
540<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
541it are as follows:</p>
542
543<ol>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]544 <li><p>Define your statistic like this:</p>
545
546<div class="doc_code">
547<pre>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]548#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
549STATISTIC(NumXForms, "The # of times I did stuff");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]550</pre>
551</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]552
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]553 <p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
554 specified by the first argument. The pass name is taken from the DEBUG_TYPE
555 macro, and the description is taken from the second argument. The variable
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]556 defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]557
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]558 <li><p>Whenever you make a transformation, bump the counter:</p>
559
560<div class="doc_code">
561<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]562++NumXForms; // <i>I did stuff!</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]563</pre>
564</div>
565
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]566 </li>
567 </ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]568
569 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
570 statistics gathered, use the '<tt>-stats</tt>' option:</p>
571
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]572<div class="doc_code">
573<pre>
574$ opt -stats -mypassname &lt; program.bc &gt; /dev/null
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]575<i>... statistics output ...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]576</pre>
577</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]578
Reid Spencer6b6c73e2007-02-09 16:00:28 +0000[diff] [blame]579 <p> When running <tt>opt</tt> on a C file from the SPEC benchmark
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]580suite, it gives a report that looks like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]581
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]582<div class="doc_code">
583<pre>
584 7646 bytecodewriter - Number of normal instructions
585 725 bytecodewriter - Number of oversized instructions
586 129996 bytecodewriter - Number of bytecode bytes written
587 2817 raise - Number of insts DCEd or constprop'd
588 3213 raise - Number of cast-of-self removed
589 5046 raise - Number of expression trees converted
590 75 raise - Number of other getelementptr's formed
591 138 raise - Number of load/store peepholes
592 42 deadtypeelim - Number of unused typenames removed from symtab
593 392 funcresolve - Number of varargs functions resolved
594 27 globaldce - Number of global variables removed
595 2 adce - Number of basic blocks removed
596 134 cee - Number of branches revectored
597 49 cee - Number of setcc instruction eliminated
598 532 gcse - Number of loads removed
599 2919 gcse - Number of instructions removed
600 86 indvars - Number of canonical indvars added
601 87 indvars - Number of aux indvars removed
602 25 instcombine - Number of dead inst eliminate
603 434 instcombine - Number of insts combined
604 248 licm - Number of load insts hoisted
605 1298 licm - Number of insts hoisted to a loop pre-header
606 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
607 75 mem2reg - Number of alloca's promoted
608 1444 cfgsimplify - Number of blocks simplified
609</pre>
610</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]611
612<p>Obviously, with so many optimizations, having a unified framework for this
613stuff is very nice. Making your pass fit well into the framework makes it more
614maintainable and useful.</p>
615
616</div>
617
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]618<!-- ======================================================================= -->
619<div class="doc_subsection">
620 <a name="ViewGraph">Viewing graphs while debugging code</a>
621</div>
622
623<div class="doc_text">
624
625<p>Several of the important data structures in LLVM are graphs: for example
626CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
627LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
628<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
629DAGs</a>. In many cases, while debugging various parts of the compiler, it is
630nice to instantly visualize these graphs.</p>
631
632<p>LLVM provides several callbacks that are available in a debug build to do
633exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
634the current LLVM tool will pop up a window containing the CFG for the function
635where each basic block is a node in the graph, and each node contains the
636instructions in the block. Similarly, there also exists
637<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
638<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
639and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]640you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]641up a window. Alternatively, you can sprinkle calls to these functions in your
642code in places you want to debug.</p>
643
644<p>Getting this to work requires a small amount of configuration. On Unix
645systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
646toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
647Mac OS/X, download and install the Mac OS/X <a
648href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]649<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]650it) to your path. Once in your system and path are set up, rerun the LLVM
651configure script and rebuild LLVM to enable this functionality.</p>
652
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]653<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
654<i>interesting</i> nodes in large complex graphs. From gdb, if you
655<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]656next <tt>call DAG.viewGraph()</tt> would highlight the node in the
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]657specified color (choices of colors can be found at <a
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]658href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]659complex node attributes can be provided with <tt>call
660DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
661found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
662Attributes</a>.) If you want to restart and clear all the current graph
663attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
664
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]665</div>
666
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]667<!-- *********************************************************************** -->
668<div class="doc_section">
669 <a name="datastructure">Picking the Right Data Structure for a Task</a>
670</div>
671<!-- *********************************************************************** -->
672
673<div class="doc_text">
674
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]675<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
676 and we commonly use STL data structures. This section describes the trade-offs
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]677 you should consider when you pick one.</p>
678
679<p>
680The first step is a choose your own adventure: do you want a sequential
681container, a set-like container, or a map-like container? The most important
682thing when choosing a container is the algorithmic properties of how you plan to
683access the container. Based on that, you should use:</p>
684
685<ul>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]686<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]687 of an value based on another value. Map-like containers also support
688 efficient queries for containment (whether a key is in the map). Map-like
689 containers generally do not support efficient reverse mapping (values to
690 keys). If you need that, use two maps. Some map-like containers also
691 support efficient iteration through the keys in sorted order. Map-like
692 containers are the most expensive sort, only use them if you need one of
693 these capabilities.</li>
694
695<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
696 stuff into a container that automatically eliminates duplicates. Some
697 set-like containers support efficient iteration through the elements in
698 sorted order. Set-like containers are more expensive than sequential
699 containers.
700</li>
701
702<li>a <a href="#ds_sequential">sequential</a> container provides
703 the most efficient way to add elements and keeps track of the order they are
704 added to the collection. They permit duplicates and support efficient
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]705 iteration, but do not support efficient look-up based on a key.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]706</li>
707
708</ul>
709
710<p>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]711Once the proper category of container is determined, you can fine tune the
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]712memory use, constant factors, and cache behaviors of access by intelligently
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]713picking a member of the category. Note that constant factors and cache behavior
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]714can be a big deal. If you have a vector that usually only contains a few
715elements (but could contain many), for example, it's much better to use
716<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
717. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
718cost of adding the elements to the container. </p>
719
720</div>
721
722<!-- ======================================================================= -->
723<div class="doc_subsection">
724 <a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
725</div>
726
727<div class="doc_text">
728There are a variety of sequential containers available for you, based on your
729needs. Pick the first in this section that will do what you want.
730</div>
731
732<!-- _______________________________________________________________________ -->
733<div class="doc_subsubsection">
734 <a name="dss_fixedarrays">Fixed Size Arrays</a>
735</div>
736
737<div class="doc_text">
738<p>Fixed size arrays are very simple and very fast. They are good if you know
739exactly how many elements you have, or you have a (low) upper bound on how many
740you have.</p>
741</div>
742
743<!-- _______________________________________________________________________ -->
744<div class="doc_subsubsection">
745 <a name="dss_heaparrays">Heap Allocated Arrays</a>
746</div>
747
748<div class="doc_text">
749<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
750the number of elements is variable, if you know how many elements you will need
751before the array is allocated, and if the array is usually large (if not,
752consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
753allocated array is the cost of the new/delete (aka malloc/free). Also note that
754if you are allocating an array of a type with a constructor, the constructor and
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]755destructors will be run for every element in the array (re-sizable vectors only
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]756construct those elements actually used).</p>
757</div>
758
759<!-- _______________________________________________________________________ -->
760<div class="doc_subsubsection">
761 <a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
762</div>
763
764<div class="doc_text">
765<p><tt>SmallVector&lt;Type, N&gt;</tt> is a simple class that looks and smells
766just like <tt>vector&lt;Type&gt;</tt>:
767it supports efficient iteration, lays out elements in memory order (so you can
768do pointer arithmetic between elements), supports efficient push_back/pop_back
769operations, supports efficient random access to its elements, etc.</p>
770
771<p>The advantage of SmallVector is that it allocates space for
772some number of elements (N) <b>in the object itself</b>. Because of this, if
773the SmallVector is dynamically smaller than N, no malloc is performed. This can
774be a big win in cases where the malloc/free call is far more expensive than the
775code that fiddles around with the elements.</p>
776
777<p>This is good for vectors that are "usually small" (e.g. the number of
778predecessors/successors of a block is usually less than 8). On the other hand,
779this makes the size of the SmallVector itself large, so you don't want to
780allocate lots of them (doing so will waste a lot of space). As such,
781SmallVectors are most useful when on the stack.</p>
782
783<p>SmallVector also provides a nice portable and efficient replacement for
784<tt>alloca</tt>.</p>
785
786</div>
787
788<!-- _______________________________________________________________________ -->
789<div class="doc_subsubsection">
790 <a name="dss_vector">&lt;vector&gt;</a>
791</div>
792
793<div class="doc_text">
794<p>
795std::vector is well loved and respected. It is useful when SmallVector isn't:
796when the size of the vector is often large (thus the small optimization will
797rarely be a benefit) or if you will be allocating many instances of the vector
798itself (which would waste space for elements that aren't in the container).
799vector is also useful when interfacing with code that expects vectors :).
800</p>
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]801
802<p>One worthwhile note about std::vector: avoid code like this:</p>
803
804<div class="doc_code">
805<pre>
806for ( ... ) {
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]807 std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]808 use V;
809}
810</pre>
811</div>
812
813<p>Instead, write this as:</p>
814
815<div class="doc_code">
816<pre>
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]817std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]818for ( ... ) {
819 use V;
820 V.clear();
821}
822</pre>
823</div>
824
825<p>Doing so will save (at least) one heap allocation and free per iteration of
826the loop.</p>
827
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]828</div>
829
830<!-- _______________________________________________________________________ -->
831<div class="doc_subsubsection">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]832 <a name="dss_deque">&lt;deque&gt;</a>
833</div>
834
835<div class="doc_text">
836<p>std::deque is, in some senses, a generalized version of std::vector. Like
837std::vector, it provides constant time random access and other similar
838properties, but it also provides efficient access to the front of the list. It
839does not guarantee continuity of elements within memory.</p>
840
841<p>In exchange for this extra flexibility, std::deque has significantly higher
842constant factor costs than std::vector. If possible, use std::vector or
843something cheaper.</p>
844</div>
845
846<!-- _______________________________________________________________________ -->
847<div class="doc_subsubsection">
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]848 <a name="dss_list">&lt;list&gt;</a>
849</div>
850
851<div class="doc_text">
852<p>std::list is an extremely inefficient class that is rarely useful.
853It performs a heap allocation for every element inserted into it, thus having an
854extremely high constant factor, particularly for small data types. std::list
855also only supports bidirectional iteration, not random access iteration.</p>
856
857<p>In exchange for this high cost, std::list supports efficient access to both
858ends of the list (like std::deque, but unlike std::vector or SmallVector). In
859addition, the iterator invalidation characteristics of std::list are stronger
860than that of a vector class: inserting or removing an element into the list does
861not invalidate iterator or pointers to other elements in the list.</p>
862</div>
863
864<!-- _______________________________________________________________________ -->
865<div class="doc_subsubsection">
866 <a name="dss_ilist">llvm/ADT/ilist</a>
867</div>
868
869<div class="doc_text">
870<p><tt>ilist&lt;T&gt;</tt> implements an 'intrusive' doubly-linked list. It is
871intrusive, because it requires the element to store and provide access to the
872prev/next pointers for the list.</p>
873
874<p>ilist has the same drawbacks as std::list, and additionally requires an
875ilist_traits implementation for the element type, but it provides some novel
876characteristics. In particular, it can efficiently store polymorphic objects,
877the traits class is informed when an element is inserted or removed from the
878list, and ilists are guaranteed to support a constant-time splice operation.
879</p>
880
881<p>These properties are exactly what we want for things like Instructions and
882basic blocks, which is why these are implemented with ilists.</p>
883</div>
884
885<!-- _______________________________________________________________________ -->
886<div class="doc_subsubsection">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]887 <a name="dss_other">Other Sequential Container options</a>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]888</div>
889
890<div class="doc_text">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]891<p>Other STL containers are available, such as std::string.</p>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]892
893<p>There are also various STL adapter classes such as std::queue,
894std::priority_queue, std::stack, etc. These provide simplified access to an
895underlying container but don't affect the cost of the container itself.</p>
896
897</div>
898
899
900<!-- ======================================================================= -->
901<div class="doc_subsection">
902 <a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
903</div>
904
905<div class="doc_text">
906
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]907<p>Set-like containers are useful when you need to canonicalize multiple values
908into a single representation. There are several different choices for how to do
909this, providing various trade-offs.</p>
910
911</div>
912
913
914<!-- _______________________________________________________________________ -->
915<div class="doc_subsubsection">
916 <a name="dss_sortedvectorset">A sorted 'vector'</a>
917</div>
918
919<div class="doc_text">
920
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]921<p>If you intend to insert a lot of elements, then do a lot of queries, a
922great approach is to use a vector (or other sequential container) with
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]923std::sort+std::unique to remove duplicates. This approach works really well if
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]924your usage pattern has these two distinct phases (insert then query), and can be
925coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
926</p>
927
928<p>
929This combination provides the several nice properties: the result data is
930contiguous in memory (good for cache locality), has few allocations, is easy to
931address (iterators in the final vector are just indices or pointers), and can be
932efficiently queried with a standard binary or radix search.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]933
934</div>
935
936<!-- _______________________________________________________________________ -->
937<div class="doc_subsubsection">
938 <a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
939</div>
940
941<div class="doc_text">
942
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]943<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]944are 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]945has space for N elements in place (thus, if the set is dynamically smaller than
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]946N, no malloc traffic is required) and accesses them with a simple linear search.
947When the set grows beyond 'N' elements, it allocates a more expensive representation that
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]948guarantees efficient access (for most types, it falls back to std::set, but for
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]949pointers it uses something far better, <a
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]950href="#dss_smallptrset">SmallPtrSet</a>).</p>
951
952<p>The magic of this class is that it handles small sets extremely efficiently,
953but gracefully handles extremely large sets without loss of efficiency. The
954drawback is that the interface is quite small: it supports insertion, queries
955and erasing, but does not support iteration.</p>
956
957</div>
958
959<!-- _______________________________________________________________________ -->
960<div class="doc_subsubsection">
961 <a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
962</div>
963
964<div class="doc_text">
965
966<p>SmallPtrSet has all the advantages of SmallSet (and a SmallSet of pointers is
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]967transparently implemented with a SmallPtrSet), but also supports iterators. If
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]968more than 'N' insertions are performed, a single quadratically
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]969probed hash table is allocated and grows as needed, providing extremely
970efficient access (constant time insertion/deleting/queries with low constant
971factors) and is very stingy with malloc traffic.</p>
972
973<p>Note that, unlike std::set, the iterators of SmallPtrSet are invalidated
974whenever an insertion occurs. Also, the values visited by the iterators are not
975visited in sorted order.</p>
976
977</div>
978
979<!-- _______________________________________________________________________ -->
980<div class="doc_subsubsection">
981 <a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
982</div>
983
984<div class="doc_text">
985
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]986<p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]987FoldingSet is an aggregate class that is really good at uniquing
988expensive-to-create or polymorphic objects. It is a combination of a chained
989hash table with intrusive links (uniqued objects are required to inherit from
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]990FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
991its ID process.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]992
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]993<p>Consider a case where you want to implement a "getOrCreateFoo" method for
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]994a complex object (for example, a node in the code generator). The client has a
995description of *what* it wants to generate (it knows the opcode and all the
996operands), 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]997only to find out it already exists, at which point we would have to delete it
998and return the node that already exists.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]999</p>
1000
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1001<p>To support this style of client, FoldingSet perform a query with a
1002FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1003element that we want to query for. The query either returns the element
1004matching the ID or it returns an opaque ID that indicates where insertion should
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1005take place. Construction of the ID usually does not require heap traffic.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1006
1007<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
1008in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1009Because the elements are individually allocated, pointers to the elements are
1010stable: inserting or removing elements does not invalidate any pointers to other
1011elements.
1012</p>
1013
1014</div>
1015
1016<!-- _______________________________________________________________________ -->
1017<div class="doc_subsubsection">
1018 <a name="dss_set">&lt;set&gt;</a>
1019</div>
1020
1021<div class="doc_text">
1022
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1023<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
1024many things but great at nothing. std::set allocates memory for each element
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1025inserted (thus it is very malloc intensive) and typically stores three pointers
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1026per element in the set (thus adding a large amount of per-element space
1027overhead). It offers guaranteed log(n) performance, which is not particularly
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1028fast from a complexity standpoint (particularly if the elements of the set are
1029expensive to compare, like strings), and has extremely high constant factors for
1030lookup, insertion and removal.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1031
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1032<p>The advantages of std::set are that its iterators are stable (deleting or
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1033inserting an element from the set does not affect iterators or pointers to other
1034elements) and that iteration over the set is guaranteed to be in sorted order.
1035If the elements in the set are large, then the relative overhead of the pointers
1036and malloc traffic is not a big deal, but if the elements of the set are small,
1037std::set is almost never a good choice.</p>
1038
1039</div>
1040
1041<!-- _______________________________________________________________________ -->
1042<div class="doc_subsubsection">
1043 <a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
1044</div>
1045
1046<div class="doc_text">
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1047<p>LLVM's SetVector&lt;Type&gt; is an adapter class that combines your choice of
1048a set-like container along with a <a href="#ds_sequential">Sequential
1049Container</a>. The important property
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1050that this provides is efficient insertion with uniquing (duplicate elements are
1051ignored) with iteration support. It implements this by inserting elements into
1052both a set-like container and the sequential container, using the set-like
1053container for uniquing and the sequential container for iteration.
1054</p>
1055
1056<p>The difference between SetVector and other sets is that the order of
1057iteration is guaranteed to match the order of insertion into the SetVector.
1058This property is really important for things like sets of pointers. Because
1059pointer values are non-deterministic (e.g. vary across runs of the program on
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1060different machines), iterating over the pointers in the set will
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1061not be in a well-defined order.</p>
1062
1063<p>
1064The drawback of SetVector is that it requires twice as much space as a normal
1065set and has the sum of constant factors from the set-like container and the
1066sequential container that it uses. Use it *only* if you need to iterate over
1067the elements in a deterministic order. SetVector is also expensive to delete
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1068elements out of (linear time), unless you use it's "pop_back" method, which is
1069faster.
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1070</p>
1071
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1072<p>SetVector is an adapter class that defaults to using std::vector and std::set
1073for the underlying containers, so it is quite expensive. However,
1074<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
1075defaults to using a SmallVector and SmallSet of a specified size. If you use
1076this, and if your sets are dynamically smaller than N, you will save a lot of
1077heap traffic.</p>
1078
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1079</div>
1080
1081<!-- _______________________________________________________________________ -->
1082<div class="doc_subsubsection">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1083 <a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
1084</div>
1085
1086<div class="doc_text">
1087
1088<p>
1089UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
1090retains a unique ID for each element inserted into the set. It internally
1091contains a map and a vector, and it assigns a unique ID for each value inserted
1092into the set.</p>
1093
1094<p>UniqueVector is very expensive: its cost is the sum of the cost of
1095maintaining both the map and vector, it has high complexity, high constant
1096factors, and produces a lot of malloc traffic. It should be avoided.</p>
1097
1098</div>
1099
1100
1101<!-- _______________________________________________________________________ -->
1102<div class="doc_subsubsection">
1103 <a name="dss_otherset">Other Set-Like Container Options</a>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1104</div>
1105
1106<div class="doc_text">
1107
1108<p>
1109The STL provides several other options, such as std::multiset and the various
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1110"hash_set" like containers (whether from C++ TR1 or from the SGI library).</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1111
1112<p>std::multiset is useful if you're not interested in elimination of
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1113duplicates, but has all the drawbacks of std::set. A sorted vector (where you
1114don't delete duplicate entries) or some other approach is almost always
1115better.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1116
1117<p>The various hash_set implementations (exposed portably by
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1118"llvm/ADT/hash_set") is a simple chained hashtable. This algorithm is as malloc
1119intensive as std::set (performing an allocation for each element inserted,
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1120thus having really high constant factors) but (usually) provides O(1)
1121insertion/deletion of elements. This can be useful if your elements are large
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1122(thus making the constant-factor cost relatively low) or if comparisons are
1123expensive. Element iteration does not visit elements in a useful order.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1124
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1125</div>
1126
1127<!-- ======================================================================= -->
1128<div class="doc_subsection">
1129 <a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
1130</div>
1131
1132<div class="doc_text">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1133Map-like containers are useful when you want to associate data to a key. As
1134usual, there are a lot of different ways to do this. :)
1135</div>
1136
1137<!-- _______________________________________________________________________ -->
1138<div class="doc_subsubsection">
1139 <a name="dss_sortedvectormap">A sorted 'vector'</a>
1140</div>
1141
1142<div class="doc_text">
1143
1144<p>
1145If your usage pattern follows a strict insert-then-query approach, you can
1146trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
1147for set-like containers</a>. The only difference is that your query function
1148(which uses std::lower_bound to get efficient log(n) lookup) should only compare
1149the key, not both the key and value. This yields the same advantages as sorted
1150vectors for sets.
1151</p>
1152</div>
1153
1154<!-- _______________________________________________________________________ -->
1155<div class="doc_subsubsection">
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1156 <a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1157</div>
1158
1159<div class="doc_text">
1160
1161<p>
1162Strings are commonly used as keys in maps, and they are difficult to support
1163efficiently: they are variable length, inefficient to hash and compare when
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1164long, expensive to copy, etc. StringMap is a specialized container designed to
1165cope with these issues. It supports mapping an arbitrary range of bytes to an
1166arbitrary other object.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1167
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1168<p>The StringMap implementation uses a quadratically-probed hash table, where
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1169the buckets store a pointer to the heap allocated entries (and some other
1170stuff). The entries in the map must be heap allocated because the strings are
1171variable length. The string data (key) and the element object (value) are
1172stored in the same allocation with the string data immediately after the element
1173object. This container guarantees the "<tt>(char*)(&amp;Value+1)</tt>" points
1174to the key string for a value.</p>
1175
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1176<p>The StringMap is very fast for several reasons: quadratic probing is very
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1177cache efficient for lookups, the hash value of strings in buckets is not
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1178recomputed when lookup up an element, StringMap rarely has to touch the
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1179memory for unrelated objects when looking up a value (even when hash collisions
1180happen), hash table growth does not recompute the hash values for strings
1181already in the table, and each pair in the map is store in a single allocation
1182(the string data is stored in the same allocation as the Value of a pair).</p>
1183
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1184<p>StringMap also provides query methods that take byte ranges, so it only ever
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1185copies a string if a value is inserted into the table.</p>
1186</div>
1187
1188<!-- _______________________________________________________________________ -->
1189<div class="doc_subsubsection">
1190 <a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
1191</div>
1192
1193<div class="doc_text">
1194<p>
1195IndexedMap is a specialized container for mapping small dense integers (or
1196values that can be mapped to small dense integers) to some other type. It is
1197internally implemented as a vector with a mapping function that maps the keys to
1198the dense integer range.
1199</p>
1200
1201<p>
1202This is useful for cases like virtual registers in the LLVM code generator: they
1203have a dense mapping that is offset by a compile-time constant (the first
1204virtual register ID).</p>
1205
1206</div>
1207
1208<!-- _______________________________________________________________________ -->
1209<div class="doc_subsubsection">
1210 <a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
1211</div>
1212
1213<div class="doc_text">
1214
1215<p>
1216DenseMap is a simple quadratically probed hash table. It excels at supporting
1217small keys and values: it uses a single allocation to hold all of the pairs that
1218are currently inserted in the map. DenseMap is a great way to map pointers to
1219pointers, or map other small types to each other.
1220</p>
1221
1222<p>
1223There are several aspects of DenseMap that you should be aware of, however. The
1224iterators in a densemap are invalidated whenever an insertion occurs, unlike
1225map. Also, because DenseMap allocates space for a large number of key/value
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1226pairs (it starts with 64 by default), it will waste a lot of space if your keys
1227or values are large. Finally, you must implement a partial specialization of
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1228DenseMapKeyInfo for the key that you want, if it isn't already supported. This
1229is required to tell DenseMap about two special marker values (which can never be
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1230inserted into the map) that it needs internally.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1231
1232</div>
1233
1234<!-- _______________________________________________________________________ -->
1235<div class="doc_subsubsection">
1236 <a name="dss_map">&lt;map&gt;</a>
1237</div>
1238
1239<div class="doc_text">
1240
1241<p>
1242std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
1243a single allocation per pair inserted into the map, it offers log(n) lookup with
1244an extremely large constant factor, imposes a space penalty of 3 pointers per
1245pair in the map, etc.</p>
1246
1247<p>std::map is most useful when your keys or values are very large, if you need
1248to iterate over the collection in sorted order, or if you need stable iterators
1249into the map (i.e. they don't get invalidated if an insertion or deletion of
1250another element takes place).</p>
1251
1252</div>
1253
1254<!-- _______________________________________________________________________ -->
1255<div class="doc_subsubsection">
1256 <a name="dss_othermap">Other Map-Like Container Options</a>
1257</div>
1258
1259<div class="doc_text">
1260
1261<p>
1262The STL provides several other options, such as std::multimap and the various
1263"hash_map" like containers (whether from C++ TR1 or from the SGI library).</p>
1264
1265<p>std::multimap is useful if you want to map a key to multiple values, but has
1266all the drawbacks of std::map. A sorted vector or some other approach is almost
1267always better.</p>
1268
1269<p>The various hash_map implementations (exposed portably by
1270"llvm/ADT/hash_map") are simple chained hash tables. This algorithm is as
1271malloc intensive as std::map (performing an allocation for each element
1272inserted, thus having really high constant factors) but (usually) provides O(1)
1273insertion/deletion of elements. This can be useful if your elements are large
1274(thus making the constant-factor cost relatively low) or if comparisons are
1275expensive. Element iteration does not visit elements in a useful order.</p>
1276
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1277</div>
1278
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]1279
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1280<!-- *********************************************************************** -->
1281<div class="doc_section">
1282 <a name="common">Helpful Hints for Common Operations</a>
1283</div>
1284<!-- *********************************************************************** -->
1285
1286<div class="doc_text">
1287
1288<p>This section describes how to perform some very simple transformations of
1289LLVM code. This is meant to give examples of common idioms used, showing the
1290practical side of LLVM transformations. <p> Because this is a "how-to" section,
1291you should also read about the main classes that you will be working with. The
1292<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
1293and descriptions of the main classes that you should know about.</p>
1294
1295</div>
1296
1297<!-- NOTE: this section should be heavy on example code -->
1298<!-- ======================================================================= -->
1299<div class="doc_subsection">
1300 <a name="inspection">Basic Inspection and Traversal Routines</a>
1301</div>
1302
1303<div class="doc_text">
1304
1305<p>The LLVM compiler infrastructure have many different data structures that may
1306be traversed. Following the example of the C++ standard template library, the
1307techniques used to traverse these various data structures are all basically the
1308same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
1309method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
1310function returns an iterator pointing to one past the last valid element of the
1311sequence, and there is some <tt>XXXiterator</tt> data type that is common
1312between the two operations.</p>
1313
1314<p>Because the pattern for iteration is common across many different aspects of
1315the program representation, the standard template library algorithms may be used
1316on them, and it is easier to remember how to iterate. First we show a few common
1317examples of the data structures that need to be traversed. Other data
1318structures are traversed in very similar ways.</p>
1319
1320</div>
1321
1322<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1323<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1324 <a name="iterate_function">Iterating over the </a><a
1325 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
1326 href="#Function"><tt>Function</tt></a>
1327</div>
1328
1329<div class="doc_text">
1330
1331<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
1332transform in some way; in particular, you'd like to manipulate its
1333<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
1334the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
1335an example that prints the name of a <tt>BasicBlock</tt> and the number of
1336<tt>Instruction</tt>s it contains:</p>
1337
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1338<div class="doc_code">
1339<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1340// <i>func is a pointer to a Function instance</i>
1341for (Function::iterator i = func-&gt;begin(), e = func-&gt;end(); i != e; ++i)
1342 // <i>Print out the name of the basic block if it has one, and then the</i>
1343 // <i>number of instructions that it contains</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1344 llvm::cerr &lt;&lt; "Basic block (name=" &lt;&lt; i-&gt;getName() &lt;&lt; ") has "
1345 &lt;&lt; i-&gt;size() &lt;&lt; " instructions.\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1346</pre>
1347</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1348
1349<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]1350invoking member functions of the <tt>Instruction</tt> class. This is
1351because the indirection operator is overloaded for the iterator
Chris Lattner7496ec52003-08-05 22:54:23 +0000[diff] [blame]1352classes. In the above code, the expression <tt>i-&gt;size()</tt> is
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1353exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
1354
1355</div>
1356
1357<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1358<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1359 <a name="iterate_basicblock">Iterating over the </a><a
1360 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1361 href="#BasicBlock"><tt>BasicBlock</tt></a>
1362</div>
1363
1364<div class="doc_text">
1365
1366<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
1367easy to iterate over the individual instructions that make up
1368<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
1369a <tt>BasicBlock</tt>:</p>
1370
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1371<div class="doc_code">
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]1372<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1373// <i>blk is a pointer to a BasicBlock instance</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1374for (BasicBlock::iterator i = blk-&gt;begin(), e = blk-&gt;end(); i != e; ++i)
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1375 // <i>The next statement works since operator&lt;&lt;(ostream&amp;,...)</i>
1376 // <i>is overloaded for Instruction&amp;</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1377 llvm::cerr &lt;&lt; *i &lt;&lt; "\n";
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]1378</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1379</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1380
1381<p>However, this isn't really the best way to print out the contents of a
1382<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
1383anything you'll care about, you could have just invoked the print routine on the
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1384basic block itself: <tt>llvm::cerr &lt;&lt; *blk &lt;&lt; "\n";</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1385
1386</div>
1387
1388<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1389<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1390 <a name="iterate_institer">Iterating over the </a><a
1391 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1392 href="#Function"><tt>Function</tt></a>
1393</div>
1394
1395<div class="doc_text">
1396
1397<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
1398<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
1399<tt>InstIterator</tt> should be used instead. You'll need to include <a
1400href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
1401and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1402small 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]1403
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1404<div class="doc_code">
1405<pre>
1406#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
1407
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1408// <i>F is a pointer to a Function instance</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1409for (inst_iterator i = inst_begin(F), e = inst_end(F); i != e; ++i)
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1410 llvm::cerr &lt;&lt; *i &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1411</pre>
1412</div>
1413
1414<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]1415work list with its initial contents. For example, if you wanted to
1416initialize a work list to contain all instructions in a <tt>Function</tt>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1417F, all you would need to do is something like:</p>
1418
1419<div class="doc_code">
1420<pre>
1421std::set&lt;Instruction*&gt; worklist;
1422worklist.insert(inst_begin(F), inst_end(F));
1423</pre>
1424</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1425
1426<p>The STL set <tt>worklist</tt> would now contain all instructions in the
1427<tt>Function</tt> pointed to by F.</p>
1428
1429</div>
1430
1431<!-- _______________________________________________________________________ -->
1432<div class="doc_subsubsection">
1433 <a name="iterate_convert">Turning an iterator into a class pointer (and
1434 vice-versa)</a>
1435</div>
1436
1437<div class="doc_text">
1438
1439<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]1440instance when all you've got at hand is an iterator. Well, extracting
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1441a reference or a pointer from an iterator is very straight-forward.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1442Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1443is a <tt>BasicBlock::const_iterator</tt>:</p>
1444
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1445<div class="doc_code">
1446<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1447Instruction&amp; inst = *i; // <i>Grab reference to instruction reference</i>
1448Instruction* pinst = &amp;*i; // <i>Grab pointer to instruction reference</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1449const Instruction&amp; inst = *j;
1450</pre>
1451</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1452
1453<p>However, the iterators you'll be working with in the LLVM framework are
1454special: they will automatically convert to a ptr-to-instance type whenever they
1455need to. Instead of dereferencing the iterator and then taking the address of
1456the result, you can simply assign the iterator to the proper pointer type and
1457you get the dereference and address-of operation as a result of the assignment
1458(behind the scenes, this is a result of overloading casting mechanisms). Thus
1459the last line of the last example,</p>
1460
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1461<div class="doc_code">
1462<pre>
1463Instruction* pinst = &amp;*i;
1464</pre>
1465</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1466
1467<p>is semantically equivalent to</p>
1468
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1469<div class="doc_code">
1470<pre>
1471Instruction* pinst = i;
1472</pre>
1473</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1474
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1475<p>It's also possible to turn a class pointer into the corresponding iterator,
1476and this is a constant time operation (very efficient). The following code
1477snippet illustrates use of the conversion constructors provided by LLVM
1478iterators. By using these, you can explicitly grab the iterator of something
1479without actually obtaining it via iteration over some structure:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1480
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1481<div class="doc_code">
1482<pre>
1483void printNextInstruction(Instruction* inst) {
1484 BasicBlock::iterator it(inst);
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1485 ++it; // <i>After this line, it refers to the instruction after *inst</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1486 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]1487}
1488</pre>
1489</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1490
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1491</div>
1492
1493<!--_______________________________________________________________________-->
1494<div class="doc_subsubsection">
1495 <a name="iterate_complex">Finding call sites: a slightly more complex
1496 example</a>
1497</div>
1498
1499<div class="doc_text">
1500
1501<p>Say that you're writing a FunctionPass and would like to count all the
1502locations in the entire module (that is, across every <tt>Function</tt>) where a
1503certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
1504learn 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]1505much more straight-forward manner, but this example will allow us to explore how
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1506you'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]1507is what we want to do:</p>
1508
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1509<div class="doc_code">
1510<pre>
1511initialize callCounter to zero
1512for each Function f in the Module
1513 for each BasicBlock b in f
1514 for each Instruction i in b
1515 if (i is a CallInst and calls the given function)
1516 increment callCounter
1517</pre>
1518</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1519
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1520<p>And the actual code is (remember, because we're writing a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1521<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1522override the <tt>runOnFunction</tt> method):</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1523
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1524<div class="doc_code">
1525<pre>
1526Function* targetFunc = ...;
1527
1528class OurFunctionPass : public FunctionPass {
1529 public:
1530 OurFunctionPass(): callCounter(0) { }
1531
1532 virtual runOnFunction(Function&amp; F) {
1533 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
1534 for (BasicBlock::iterator i = b-&gt;begin(); ie = b-&gt;end(); i != ie; ++i) {
1535 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a>&lt;<a
1536 href="#CallInst">CallInst</a>&gt;(&amp;*i)) {
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1537 // <i>We know we've encountered a call instruction, so we</i>
1538 // <i>need to determine if it's a call to the</i>
1539 // <i>function pointed to by m_func or not</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1540
1541 if (callInst-&gt;getCalledFunction() == targetFunc)
1542 ++callCounter;
1543 }
1544 }
1545 }
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1546 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1547
1548 private:
1549 unsigned callCounter;
1550};
1551</pre>
1552</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1553
1554</div>
1555
Brian Gaekef1972c62003-11-07 19:25:45 +0000[diff] [blame]1556<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1557<div class="doc_subsubsection">
1558 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
1559</div>
1560
1561<div class="doc_text">
1562
1563<p>You may have noticed that the previous example was a bit oversimplified in
1564that it did not deal with call sites generated by 'invoke' instructions. In
1565this, and in other situations, you may find that you want to treat
1566<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
1567most-specific common base class is <tt>Instruction</tt>, which includes lots of
1568less closely-related things. For these cases, LLVM provides a handy wrapper
1569class called <a
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]1570href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1571It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
1572methods that provide functionality common to <tt>CallInst</tt>s and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1573<tt>InvokeInst</tt>s.</p>
1574
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1575<p>This class has "value semantics": it should be passed by value, not by
1576reference and it should not be dynamically allocated or deallocated using
1577<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
1578assignable and constructable, with costs equivalents to that of a bare pointer.
1579If you look at its definition, it has only a single pointer member.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1580
1581</div>
1582
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]1583<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1584<div class="doc_subsubsection">
1585 <a name="iterate_chains">Iterating over def-use &amp; use-def chains</a>
1586</div>
1587
1588<div class="doc_text">
1589
1590<p>Frequently, we might have an instance of the <a
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]1591href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1592determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
1593<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
1594For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
1595particular function <tt>foo</tt>. Finding all of the instructions that
1596<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
1597of <tt>F</tt>:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1598
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1599<div class="doc_code">
1600<pre>
1601Function* F = ...;
1602
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1603for (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]1604 if (Instruction *Inst = dyn_cast&lt;Instruction&gt;(*i)) {
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1605 llvm::cerr &lt;&lt; "F is used in instruction:\n";
1606 llvm::cerr &lt;&lt; *Inst &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1607 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1608</pre>
1609</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1610
1611<p>Alternately, it's common to have an instance of the <a
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1612href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1613<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
1614<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
1615<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
1616all of the values that a particular instruction uses (that is, the operands of
1617the particular <tt>Instruction</tt>):</p>
1618
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1619<div class="doc_code">
1620<pre>
1621Instruction* pi = ...;
1622
1623for (User::op_iterator i = pi-&gt;op_begin(), e = pi-&gt;op_end(); i != e; ++i) {
1624 Value* v = *i;
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1625 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1626}
1627</pre>
1628</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1629
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]1630<!--
1631 def-use chains ("finding all users of"): Value::use_begin/use_end
1632 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1633-->
1634
1635</div>
1636
1637<!-- ======================================================================= -->
1638<div class="doc_subsection">
1639 <a name="simplechanges">Making simple changes</a>
1640</div>
1641
1642<div class="doc_text">
1643
1644<p>There are some primitive transformation operations present in the LLVM
Joel Stanley753eb712002-09-11 22:32:24 +0000[diff] [blame]1645infrastructure that are worth knowing about. When performing
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1646transformations, it's fairly common to manipulate the contents of basic
1647blocks. This section describes some of the common methods for doing so
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1648and gives example code.</p>
1649
1650</div>
1651
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1652<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1653<div class="doc_subsubsection">
1654 <a name="schanges_creating">Creating and inserting new
1655 <tt>Instruction</tt>s</a>
1656</div>
1657
1658<div class="doc_text">
1659
1660<p><i>Instantiating Instructions</i></p>
1661
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1662<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1663constructor for the kind of instruction to instantiate and provide the necessary
1664parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
1665(const-ptr-to) <tt>Type</tt>. Thus:</p>
1666
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1667<div class="doc_code">
1668<pre>
1669AllocaInst* ai = new AllocaInst(Type::IntTy);
1670</pre>
1671</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1672
1673<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1674one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1675subclass is likely to have varying default parameters which change the semantics
1676of the instruction, so refer to the <a
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]1677href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1678Instruction</a> that you're interested in instantiating.</p>
1679
1680<p><i>Naming values</i></p>
1681
1682<p>It is very useful to name the values of instructions when you're able to, as
1683this facilitates the debugging of your transformations. If you end up looking
1684at generated LLVM machine code, you definitely want to have logical names
1685associated with the results of instructions! By supplying a value for the
1686<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
1687associate a logical name with the result of the instruction's execution at
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1688run time. For example, say that I'm writing a transformation that dynamically
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1689allocates space for an integer on the stack, and that integer is going to be
1690used as some kind of index by some other code. To accomplish this, I place an
1691<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
1692<tt>Function</tt>, and I'm intending to use it within the same
1693<tt>Function</tt>. I might do:</p>
1694
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1695<div class="doc_code">
1696<pre>
1697AllocaInst* pa = new AllocaInst(Type::IntTy, 0, "indexLoc");
1698</pre>
1699</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1700
1701<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1702execution value, which is a pointer to an integer on the run time stack.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1703
1704<p><i>Inserting instructions</i></p>
1705
1706<p>There are essentially two ways to insert an <tt>Instruction</tt>
1707into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
1708
Joel Stanley9dd1ad62002-09-18 03:17:23 +0000[diff] [blame]1709<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1710 <li>Insertion into an explicit instruction list
1711
1712 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
1713 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
1714 before <tt>*pi</tt>, we do the following: </p>
1715
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1716<div class="doc_code">
1717<pre>
1718BasicBlock *pb = ...;
1719Instruction *pi = ...;
1720Instruction *newInst = new Instruction(...);
1721
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1722pb-&gt;getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1723</pre>
1724</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1725
1726 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
1727 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
1728 classes provide constructors which take a pointer to a
1729 <tt>BasicBlock</tt> to be appended to. For example code that
1730 looked like: </p>
1731
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1732<div class="doc_code">
1733<pre>
1734BasicBlock *pb = ...;
1735Instruction *newInst = new Instruction(...);
1736
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1737pb-&gt;getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1738</pre>
1739</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1740
1741 <p>becomes: </p>
1742
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1743<div class="doc_code">
1744<pre>
1745BasicBlock *pb = ...;
1746Instruction *newInst = new Instruction(..., pb);
1747</pre>
1748</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1749
1750 <p>which is much cleaner, especially if you are creating
1751 long instruction streams.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1752
1753 <li>Insertion into an implicit instruction list
1754
1755 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
1756 are implicitly associated with an existing instruction list: the instruction
1757 list of the enclosing basic block. Thus, we could have accomplished the same
1758 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
1759 </p>
1760
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1761<div class="doc_code">
1762<pre>
1763Instruction *pi = ...;
1764Instruction *newInst = new Instruction(...);
1765
1766pi-&gt;getParent()-&gt;getInstList().insert(pi, newInst);
1767</pre>
1768</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1769
1770 <p>In fact, this sequence of steps occurs so frequently that the
1771 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
1772 constructors which take (as a default parameter) a pointer to an
1773 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
1774 precede. That is, <tt>Instruction</tt> constructors are capable of
1775 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
1776 provided instruction, immediately before that instruction. Using an
1777 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
1778 parameter, the above code becomes:</p>
1779
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1780<div class="doc_code">
1781<pre>
1782Instruction* pi = ...;
1783Instruction* newInst = new Instruction(..., pi);
1784</pre>
1785</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1786
1787 <p>which is much cleaner, especially if you're creating a lot of
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1788 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1789</ul>
1790
1791</div>
1792
1793<!--_______________________________________________________________________-->
1794<div class="doc_subsubsection">
1795 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
1796</div>
1797
1798<div class="doc_text">
1799
1800<p>Deleting an instruction from an existing sequence of instructions that form a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1801<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1802you must have a pointer to the instruction that you wish to delete. Second, you
1803need to obtain the pointer to that instruction's basic block. You use the
1804pointer to the basic block to get its list of instructions and then use the
1805erase function to remove your instruction. For example:</p>
1806
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1807<div class="doc_code">
1808<pre>
1809<a href="#Instruction">Instruction</a> *I = .. ;
1810<a href="#BasicBlock">BasicBlock</a> *BB = I-&gt;getParent();
1811
1812BB-&gt;getInstList().erase(I);
1813</pre>
1814</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1815
1816</div>
1817
1818<!--_______________________________________________________________________-->
1819<div class="doc_subsubsection">
1820 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
1821 <tt>Value</tt></a>
1822</div>
1823
1824<div class="doc_text">
1825
1826<p><i>Replacing individual instructions</i></p>
1827
1828<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]1829permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1830and <tt>ReplaceInstWithInst</tt>.</p>
1831
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1832<h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1833
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1834<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1835 <li><tt>ReplaceInstWithValue</tt>
1836
1837 <p>This function replaces all uses (within a basic block) of a given
1838 instruction with a value, and then removes the original instruction. The
1839 following example illustrates the replacement of the result of a particular
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]1840 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1841 pointer to an integer.</p>
1842
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1843<div class="doc_code">
1844<pre>
1845AllocaInst* instToReplace = ...;
1846BasicBlock::iterator ii(instToReplace);
1847
1848ReplaceInstWithValue(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
1849 Constant::getNullValue(PointerType::get(Type::IntTy)));
1850</pre></div></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1851
1852 <li><tt>ReplaceInstWithInst</tt>
1853
1854 <p>This function replaces a particular instruction with another
1855 instruction. The following example illustrates the replacement of one
1856 <tt>AllocaInst</tt> with another.</p>
1857
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1858<div class="doc_code">
1859<pre>
1860AllocaInst* instToReplace = ...;
1861BasicBlock::iterator ii(instToReplace);
1862
1863ReplaceInstWithInst(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
1864 new AllocaInst(Type::IntTy, 0, "ptrToReplacedInt"));
1865</pre></div></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1866</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1867
1868<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
1869
1870<p>You can use <tt>Value::replaceAllUsesWith</tt> and
1871<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]1872doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1873and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1874information.</p>
1875
1876<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
1877include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
1878ReplaceInstWithValue, ReplaceInstWithInst -->
1879
1880</div>
1881
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame^]1882<!--_______________________________________________________________________-->
1883<div class="doc_subsubsection">
1884 <a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
1885</div>
1886
1887<div class="doc_text">
1888
1889<p>Deleting a global variable from a module is similar to deleting an
1890instruction. First, you must have a pointer to the global variable that you wish
1891 to delete. Second, you must have a pointer to the module the global variable
1892 belongs to. You use the pointer to the module to get its list of global
1893 variables and then use the erase function to remove your global variable.
1894 For example:</p>
1895
1896<div class="doc_code">
1897<pre>
1898<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
1899<a href="#Module">Module</a> *M = GV-&gt;getParent();
1900
1901M-&gt;getGlobaleList().erase(GV);
1902</pre>
1903</div>
1904
1905</div>
1906
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]1907<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1908<div class="doc_section">
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]1909 <a name="advanced">Advanced Topics</a>
1910</div>
1911<!-- *********************************************************************** -->
1912
1913<div class="doc_text">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1914<p>
1915This section describes some of the advanced or obscure API's that most clients
1916do not need to be aware of. These API's tend manage the inner workings of the
1917LLVM system, and only need to be accessed in unusual circumstances.
1918</p>
1919</div>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]1920
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1921<!-- ======================================================================= -->
1922<div class="doc_subsection">
1923 <a name="TypeResolve">LLVM Type Resolution</a>
1924</div>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]1925
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1926<div class="doc_text">
1927
1928<p>
1929The LLVM type system has a very simple goal: allow clients to compare types for
1930structural equality with a simple pointer comparison (aka a shallow compare).
1931This goal makes clients much simpler and faster, and is used throughout the LLVM
1932system.
1933</p>
1934
1935<p>
1936Unfortunately achieving this goal is not a simple matter. In particular,
1937recursive types and late resolution of opaque types makes the situation very
1938difficult to handle. Fortunately, for the most part, our implementation makes
1939most clients able to be completely unaware of the nasty internal details. The
1940primary case where clients are exposed to the inner workings of it are when
1941building a recursive type. In addition to this case, the LLVM bytecode reader,
1942assembly parser, and linker also have to be aware of the inner workings of this
1943system.
1944</p>
1945
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]1946<p>
1947For our purposes below, we need three concepts. First, an "Opaque Type" is
1948exactly as defined in the <a href="LangRef.html#t_opaque">language
1949reference</a>. Second an "Abstract Type" is any type which includes an
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]1950opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
1951Third, 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]1952float }</tt>").
1953</p>
1954
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1955</div>
1956
1957<!-- ______________________________________________________________________ -->
1958<div class="doc_subsubsection">
1959 <a name="BuildRecType">Basic Recursive Type Construction</a>
1960</div>
1961
1962<div class="doc_text">
1963
1964<p>
1965Because the most common question is "how do I build a recursive type with LLVM",
1966we answer it now and explain it as we go. Here we include enough to cause this
1967to be emitted to an output .ll file:
1968</p>
1969
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1970<div class="doc_code">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1971<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]1972%mylist = type { %mylist*, i32 }
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1973</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1974</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1975
1976<p>
1977To build this, use the following LLVM APIs:
1978</p>
1979
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1980<div class="doc_code">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1981<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1982// <i>Create the initial outer struct</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1983<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
1984std::vector&lt;const Type*&gt; Elts;
1985Elts.push_back(PointerType::get(StructTy));
1986Elts.push_back(Type::IntTy);
1987StructType *NewSTy = StructType::get(Elts);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1988
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]1989// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1990// <i>the struct and the opaque type are actually the same.</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1991cast&lt;OpaqueType&gt;(StructTy.get())-&gt;<a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1992
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1993// <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]1994// <i>kept up-to-date</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1995NewSTy = cast&lt;StructType&gt;(StructTy.get());
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1996
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1997// <i>Add a name for the type to the module symbol table (optional)</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1998MyModule-&gt;addTypeName("mylist", NewSTy);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]1999</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2000</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2001
2002<p>
2003This code shows the basic approach used to build recursive types: build a
2004non-recursive type using 'opaque', then use type unification to close the cycle.
2005The type unification step is performed by the <tt><a
Chris Lattneraff26d12007-02-03 03:06:52 +0000[diff] [blame]2006href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2007described next. After that, we describe the <a
2008href="#PATypeHolder">PATypeHolder class</a>.
2009</p>
2010
2011</div>
2012
2013<!-- ______________________________________________________________________ -->
2014<div class="doc_subsubsection">
2015 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
2016</div>
2017
2018<div class="doc_text">
2019<p>
2020The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
2021While this method is actually a member of the DerivedType class, it is most
2022often used on OpaqueType instances. Type unification is actually a recursive
2023process. After unification, types can become structurally isomorphic to
2024existing types, and all duplicates are deleted (to preserve pointer equality).
2025</p>
2026
2027<p>
2028In the example above, the OpaqueType object is definitely deleted.
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2029Additionally, if there is an "{ \2*, i32}" type already created in the system,
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2030the pointer and struct type created are <b>also</b> deleted. Obviously whenever
2031a type is deleted, any "Type*" pointers in the program are invalidated. As
2032such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
2033live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
2034types can never move or be deleted). To deal with this, the <a
2035href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
2036reference to a possibly refined type, and the <a
2037href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
2038complex datastructures.
2039</p>
2040
2041</div>
2042
2043<!-- ______________________________________________________________________ -->
2044<div class="doc_subsubsection">
2045 <a name="PATypeHolder">The PATypeHolder Class</a>
2046</div>
2047
2048<div class="doc_text">
2049<p>
2050PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
2051happily goes about nuking types that become isomorphic to existing types, it
2052automatically updates all PATypeHolder objects to point to the new type. In the
2053example above, this allows the code to maintain a pointer to the resultant
2054resolved recursive type, even though the Type*'s are potentially invalidated.
2055</p>
2056
2057<p>
2058PATypeHolder is an extremely light-weight object that uses a lazy union-find
2059implementation to update pointers. For example the pointer from a Value to its
2060Type is maintained by PATypeHolder objects.
2061</p>
2062
2063</div>
2064
2065<!-- ______________________________________________________________________ -->
2066<div class="doc_subsubsection">
2067 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
2068</div>
2069
2070<div class="doc_text">
2071
2072<p>
2073Some data structures need more to perform more complex updates when types get
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2074resolved. To support this, a class can derive from the AbstractTypeUser class.
2075This class
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2076allows it to get callbacks when certain types are resolved. To register to get
2077callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]2078methods can be called on a type. Note that these methods only work for <i>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2079 abstract</i> types. Concrete types (those that do not include any opaque
2080objects) can never be refined.
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2081</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2082</div>
2083
2084
2085<!-- ======================================================================= -->
2086<div class="doc_subsection">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2087 <a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
2088 <tt>TypeSymbolTable</tt> classes</a>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2089</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2090
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2091<div class="doc_text">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2092<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
2093ValueSymbolTable</a></tt> class provides a symbol table that the <a
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2094href="#Function"><tt>Function</tt></a> and <a href="#Module">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2095<tt>Module</tt></a> classes use for naming value definitions. The symbol table
2096can provide a name for any <a href="#Value"><tt>Value</tt></a>.
2097The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
2098TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
2099names for types.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2100
Reid Spencera6362242007-01-07 00:41:39 +0000[diff] [blame]2101<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
2102by most clients. It should only be used when iteration over the symbol table
2103names themselves are required, which is very special purpose. Note that not
2104all LLVM
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2105<a href="#Value">Value</a>s have names, and those without names (i.e. they have
2106an empty name) do not exist in the symbol table.
2107</p>
2108
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2109<p>These symbol tables support iteration over the values/types in the symbol
2110table with <tt>begin/end/iterator</tt> and supports querying to see if a
2111specific name is in the symbol table (with <tt>lookup</tt>). The
2112<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
2113simply call <tt>setName</tt> on a value, which will autoinsert it into the
2114appropriate symbol table. For types, use the Module::addTypeName method to
2115insert entries into the symbol table.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2116
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2117</div>
2118
2119
2120
2121<!-- *********************************************************************** -->
2122<div class="doc_section">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2123 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
2124</div>
2125<!-- *********************************************************************** -->
2126
2127<div class="doc_text">
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2128<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
2129<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2130
2131<p>The Core LLVM classes are the primary means of representing the program
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2132being inspected or transformed. The core LLVM classes are defined in
2133header files in the <tt>include/llvm/</tt> directory, and implemented in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2134the <tt>lib/VMCore</tt> directory.</p>
2135
2136</div>
2137
2138<!-- ======================================================================= -->
2139<div class="doc_subsection">
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2140 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
2141</div>
2142
2143<div class="doc_text">
2144
2145 <p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
2146 a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
2147 through its subclasses. Certain primitive types (<tt>VoidType</tt>,
2148 <tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
2149 subclasses. They are hidden because they offer no useful functionality beyond
2150 what the <tt>Type</tt> class offers except to distinguish themselves from
2151 other subclasses of <tt>Type</tt>.</p>
2152 <p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
2153 named, but this is not a requirement. There exists exactly
2154 one instance of a given shape at any one time. This allows type equality to
2155 be performed with address equality of the Type Instance. That is, given two
2156 <tt>Type*</tt> values, the types are identical if the pointers are identical.
2157 </p>
2158</div>
2159
2160<!-- _______________________________________________________________________ -->
2161<div class="doc_subsubsection">
2162 <a name="m_Value">Important Public Methods</a>
2163</div>
2164
2165<div class="doc_text">
2166
2167<ul>
Chris Lattner8f79df32007-01-15 01:55:32 +0000[diff] [blame]2168 <li><tt>bool isInteger() const</tt>: Returns true for any integer type.</li>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2169
2170 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
2171 floating point types.</li>
2172
2173 <li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
2174 an OpaqueType anywhere in its definition).</li>
2175
2176 <li><tt>bool isSized()</tt>: Return true if the type has known size. Things
2177 that don't have a size are abstract types, labels and void.</li>
2178
2179</ul>
2180</div>
2181
2182<!-- _______________________________________________________________________ -->
2183<div class="doc_subsubsection">
2184 <a name="m_Value">Important Derived Types</a>
2185</div>
2186<div class="doc_text">
2187<dl>
2188 <dt><tt>IntegerType</tt></dt>
2189 <dd>Subclass of DerivedType that represents integer types of any bit width.
2190 Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
2191 <tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
2192 <ul>
2193 <li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
2194 type of a specific bit width.</li>
2195 <li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
2196 type.</li>
2197 </ul>
2198 </dd>
2199 <dt><tt>SequentialType</tt></dt>
2200 <dd>This is subclassed by ArrayType and PointerType
2201 <ul>
2202 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
2203 of the elements in the sequential type. </li>
2204 </ul>
2205 </dd>
2206 <dt><tt>ArrayType</tt></dt>
2207 <dd>This is a subclass of SequentialType and defines the interface for array
2208 types.
2209 <ul>
2210 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
2211 elements in the array. </li>
2212 </ul>
2213 </dd>
2214 <dt><tt>PointerType</tt></dt>
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]2215 <dd>Subclass of SequentialType for pointer types.</dd>
Reid Spencer9d6565a2007-02-15 02:26:10 +0000[diff] [blame]2216 <dt><tt>VectorType</tt></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2217 <dd>Subclass of SequentialType for vector types. A
2218 vector type is similar to an ArrayType but is distinguished because it is
2219 a first class type wherease ArrayType is not. Vector types are used for
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2220 vector operations and are usually small vectors of of an integer or floating
2221 point type.</dd>
2222 <dt><tt>StructType</tt></dt>
2223 <dd>Subclass of DerivedTypes for struct types.</dd>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2224 <dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2225 <dd>Subclass of DerivedTypes for function types.
2226 <ul>
2227 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
2228 function</li>
2229 <li><tt> const Type * getReturnType() const</tt>: Returns the
2230 return type of the function.</li>
2231 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
2232 the type of the ith parameter.</li>
2233 <li><tt> const unsigned getNumParams() const</tt>: Returns the
2234 number of formal parameters.</li>
2235 </ul>
2236 </dd>
2237 <dt><tt>OpaqueType</tt></dt>
2238 <dd>Sublcass of DerivedType for abstract types. This class
2239 defines no content and is used as a placeholder for some other type. Note
2240 that OpaqueType is used (temporarily) during type resolution for forward
2241 references of types. Once the referenced type is resolved, the OpaqueType
2242 is replaced with the actual type. OpaqueType can also be used for data
2243 abstraction. At link time opaque types can be resolved to actual types
2244 of the same name.</dd>
2245</dl>
2246</div>
2247
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2248
2249
2250<!-- ======================================================================= -->
2251<div class="doc_subsection">
2252 <a name="Module">The <tt>Module</tt> class</a>
2253</div>
2254
2255<div class="doc_text">
2256
2257<p><tt>#include "<a
2258href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
2259<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
2260
2261<p>The <tt>Module</tt> class represents the top level structure present in LLVM
2262programs. An LLVM module is effectively either a translation unit of the
2263original program or a combination of several translation units merged by the
2264linker. The <tt>Module</tt> class keeps track of a list of <a
2265href="#Function"><tt>Function</tt></a>s, a list of <a
2266href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
2267href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
2268helpful member functions that try to make common operations easy.</p>
2269
2270</div>
2271
2272<!-- _______________________________________________________________________ -->
2273<div class="doc_subsubsection">
2274 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
2275</div>
2276
2277<div class="doc_text">
2278
2279<ul>
2280 <li><tt>Module::Module(std::string name = "")</tt></li>
2281</ul>
2282
2283<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
2284provide a name for it (probably based on the name of the translation unit).</p>
2285
2286<ul>
2287 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
2288 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
2289
2290 <tt>begin()</tt>, <tt>end()</tt>
2291 <tt>size()</tt>, <tt>empty()</tt>
2292
2293 <p>These are forwarding methods that make it easy to access the contents of
2294 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
2295 list.</p></li>
2296
2297 <li><tt>Module::FunctionListType &amp;getFunctionList()</tt>
2298
2299 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
2300 necessary to use when you need to update the list or perform a complex
2301 action that doesn't have a forwarding method.</p>
2302
2303 <p><!-- Global Variable --></p></li>
2304</ul>
2305
2306<hr>
2307
2308<ul>
2309 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
2310
2311 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
2312
2313 <tt>global_begin()</tt>, <tt>global_end()</tt>
2314 <tt>global_size()</tt>, <tt>global_empty()</tt>
2315
2316 <p> These are forwarding methods that make it easy to access the contents of
2317 a <tt>Module</tt> object's <a
2318 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
2319
2320 <li><tt>Module::GlobalListType &amp;getGlobalList()</tt>
2321
2322 <p>Returns the list of <a
2323 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
2324 use when you need to update the list or perform a complex action that
2325 doesn't have a forwarding method.</p>
2326
2327 <p><!-- Symbol table stuff --> </p></li>
2328</ul>
2329
2330<hr>
2331
2332<ul>
2333 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
2334
2335 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2336 for this <tt>Module</tt>.</p>
2337
2338 <p><!-- Convenience methods --></p></li>
2339</ul>
2340
2341<hr>
2342
2343<ul>
2344 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
2345 &amp;Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
2346
2347 <p>Look up the specified function in the <tt>Module</tt> <a
2348 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
2349 <tt>null</tt>.</p></li>
2350
2351 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
2352 std::string &amp;Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
2353
2354 <p>Look up the specified function in the <tt>Module</tt> <a
2355 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
2356 external declaration for the function and return it.</p></li>
2357
2358 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
2359
2360 <p>If there is at least one entry in the <a
2361 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
2362 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
2363 string.</p></li>
2364
2365 <li><tt>bool addTypeName(const std::string &amp;Name, const <a
2366 href="#Type">Type</a> *Ty)</tt>
2367
2368 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2369 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
2370 name, true is returned and the <a
2371 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
2372</ul>
2373
2374</div>
2375
2376
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2377<!-- ======================================================================= -->
2378<div class="doc_subsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2379 <a name="Value">The <tt>Value</tt> class</a>
2380</div>
2381
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2382<div class="doc_text">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2383
2384<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
2385<br>
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]2386doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2387
2388<p>The <tt>Value</tt> class is the most important class in the LLVM Source
2389base. It represents a typed value that may be used (among other things) as an
2390operand to an instruction. There are many different types of <tt>Value</tt>s,
2391such as <a href="#Constant"><tt>Constant</tt></a>s,<a
2392href="#Argument"><tt>Argument</tt></a>s. Even <a
2393href="#Instruction"><tt>Instruction</tt></a>s and <a
2394href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
2395
2396<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
2397for a program. For example, an incoming argument to a function (represented
2398with an instance of the <a href="#Argument">Argument</a> class) is "used" by
2399every instruction in the function that references the argument. To keep track
2400of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
2401href="#User"><tt>User</tt></a>s that is using it (the <a
2402href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
2403graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
2404def-use information in the program, and is accessible through the <tt>use_</tt>*
2405methods, shown below.</p>
2406
2407<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
2408and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
2409method. In addition, all LLVM values can be named. The "name" of the
2410<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
2411
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2412<div class="doc_code">
2413<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2414%<b>foo</b> = add i32 1, 2
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2415</pre>
2416</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2417
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2418<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]2419that the name of any value may be missing (an empty string), so names should
2420<b>ONLY</b> be used for debugging (making the source code easier to read,
2421debugging printouts), they should not be used to keep track of values or map
2422between them. For this purpose, use a <tt>std::map</tt> of pointers to the
2423<tt>Value</tt> itself instead.</p>
2424
2425<p>One important aspect of LLVM is that there is no distinction between an SSA
2426variable and the operation that produces it. Because of this, any reference to
2427the value produced by an instruction (or the value available as an incoming
Chris Lattnerd5fc4fc2004-03-18 14:58:55 +0000[diff] [blame]2428argument, for example) is represented as a direct pointer to the instance of
2429the class that
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2430represents this value. Although this may take some getting used to, it
2431simplifies the representation and makes it easier to manipulate.</p>
2432
2433</div>
2434
2435<!-- _______________________________________________________________________ -->
2436<div class="doc_subsubsection">
2437 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
2438</div>
2439
2440<div class="doc_text">
2441
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2442<ul>
2443 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
2444use-list<br>
2445 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
2446the use-list<br>
2447 <tt>unsigned use_size()</tt> - Returns the number of users of the
2448value.<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2449 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2450 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
2451the use-list.<br>
2452 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
2453use-list.<br>
2454 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
2455element in the list.
2456 <p> These methods are the interface to access the def-use
2457information in LLVM. As with all other iterators in LLVM, the naming
2458conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2459 </li>
2460 <li><tt><a href="#Type">Type</a> *getType() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2461 <p>This method returns the Type of the Value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2462 </li>
2463 <li><tt>bool hasName() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2464 <tt>std::string getName() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2465 <tt>void setName(const std::string &amp;Name)</tt>
2466 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
2467be aware of the <a href="#nameWarning">precaution above</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2468 </li>
2469 <li><tt>void replaceAllUsesWith(Value *V)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2470
2471 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
2472 href="#User"><tt>User</tt>s</a> of the current value to refer to
2473 "<tt>V</tt>" instead. For example, if you detect that an instruction always
2474 produces a constant value (for example through constant folding), you can
2475 replace all uses of the instruction with the constant like this:</p>
2476
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2477<div class="doc_code">
2478<pre>
2479Inst-&gt;replaceAllUsesWith(ConstVal);
2480</pre>
2481</div>
2482
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2483</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2484
2485</div>
2486
2487<!-- ======================================================================= -->
2488<div class="doc_subsection">
2489 <a name="User">The <tt>User</tt> class</a>
2490</div>
2491
2492<div class="doc_text">
2493
2494<p>
2495<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]2496doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2497Superclass: <a href="#Value"><tt>Value</tt></a></p>
2498
2499<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
2500refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
2501that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
2502referring to. The <tt>User</tt> class itself is a subclass of
2503<tt>Value</tt>.</p>
2504
2505<p>The operands of a <tt>User</tt> point directly to the LLVM <a
2506href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
2507Single Assignment (SSA) form, there can only be one definition referred to,
2508allowing this direct connection. This connection provides the use-def
2509information in LLVM.</p>
2510
2511</div>
2512
2513<!-- _______________________________________________________________________ -->
2514<div class="doc_subsubsection">
2515 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
2516</div>
2517
2518<div class="doc_text">
2519
2520<p>The <tt>User</tt> class exposes the operand list in two ways: through
2521an index access interface and through an iterator based interface.</p>
2522
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2523<ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2524 <li><tt>Value *getOperand(unsigned i)</tt><br>
2525 <tt>unsigned getNumOperands()</tt>
2526 <p> These two methods expose the operands of the <tt>User</tt> in a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2527convenient form for direct access.</p></li>
2528
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2529 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
2530list<br>
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]2531 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
2532the operand list.<br>
2533 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2534operand list.
2535 <p> Together, these methods make up the iterator based interface to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2536the operands of a <tt>User</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2537</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2538
2539</div>
2540
2541<!-- ======================================================================= -->
2542<div class="doc_subsection">
2543 <a name="Instruction">The <tt>Instruction</tt> class</a>
2544</div>
2545
2546<div class="doc_text">
2547
2548<p><tt>#include "</tt><tt><a
2549href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2550doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2551Superclasses: <a href="#User"><tt>User</tt></a>, <a
2552href="#Value"><tt>Value</tt></a></p>
2553
2554<p>The <tt>Instruction</tt> class is the common base class for all LLVM
2555instructions. It provides only a few methods, but is a very commonly used
2556class. The primary data tracked by the <tt>Instruction</tt> class itself is the
2557opcode (instruction type) and the parent <a
2558href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
2559into. To represent a specific type of instruction, one of many subclasses of
2560<tt>Instruction</tt> are used.</p>
2561
2562<p> Because the <tt>Instruction</tt> class subclasses the <a
2563href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
2564way as for other <a href="#User"><tt>User</tt></a>s (with the
2565<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
2566<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
2567the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
2568file contains some meta-data about the various different types of instructions
2569in LLVM. It describes the enum values that are used as opcodes (for example
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2570<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2571concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
2572example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2573href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2574this file confuses doxygen, so these enum values don't show up correctly in the
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2575<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2576
2577</div>
2578
2579<!-- _______________________________________________________________________ -->
2580<div class="doc_subsubsection">
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2581 <a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
2582 class</a>
2583</div>
2584<div class="doc_text">
2585 <ul>
2586 <li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
2587 <p>This subclasses represents all two operand instructions whose operands
2588 must be the same type, except for the comparison instructions.</p></li>
2589 <li><tt><a name="CastInst">CastInst</a></tt>
2590 <p>This subclass is the parent of the 12 casting instructions. It provides
2591 common operations on cast instructions.</p>
2592 <li><tt><a name="CmpInst">CmpInst</a></tt>
2593 <p>This subclass respresents the two comparison instructions,
2594 <a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
2595 <a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
2596 <li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
2597 <p>This subclass is the parent of all terminator instructions (those which
2598 can terminate a block).</p>
2599 </ul>
2600 </div>
2601
2602<!-- _______________________________________________________________________ -->
2603<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2604 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
2605 class</a>
2606</div>
2607
2608<div class="doc_text">
2609
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2610<ul>
2611 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2612 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
2613this <tt>Instruction</tt> is embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2614 <li><tt>bool mayWriteToMemory()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2615 <p>Returns true if the instruction writes to memory, i.e. it is a
2616 <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]2617 <li><tt>unsigned getOpcode()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2618 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2619 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2620 <p>Returns another instance of the specified instruction, identical
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2621in all ways to the original except that the instruction has no parent
2622(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2623and it has no name</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2624</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2625
2626</div>
2627
2628<!-- ======================================================================= -->
2629<div class="doc_subsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2630 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2631</div>
2632
2633<div class="doc_text">
2634
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2635<p>Constant represents a base class for different types of constants. It
2636is subclassed by ConstantInt, ConstantArray, etc. for representing
2637the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
2638a subclass, which represents the address of a global variable or function.
2639</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2640
2641</div>
2642
2643<!-- _______________________________________________________________________ -->
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2644<div class="doc_subsubsection">Important Subclasses of Constant </div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2645<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2646<ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2647 <li>ConstantInt : This subclass of Constant represents an integer constant of
2648 any width.
2649 <ul>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]2650 <li><tt>const APInt&amp; getValue() const</tt>: Returns the underlying
2651 value of this constant, an APInt value.</li>
2652 <li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
2653 value to an int64_t via sign extension. If the value (not the bit width)
2654 of the APInt is too large to fit in an int64_t, an assertion will result.
2655 For this reason, use of this method is discouraged.</li>
2656 <li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
2657 value to a uint64_t via zero extension. IF the value (not the bit width)
2658 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]2659 For this reason, use of this method is discouraged.</li>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]2660 <li><tt>static ConstantInt* get(const APInt&amp; Val)</tt>: Returns the
2661 ConstantInt object that represents the value provided by <tt>Val</tt>.
2662 The type is implied as the IntegerType that corresponds to the bit width
2663 of <tt>Val</tt>.</li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2664 <li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
2665 Returns the ConstantInt object that represents the value provided by
2666 <tt>Val</tt> for integer type <tt>Ty</tt>.</li>
2667 </ul>
2668 </li>
2669 <li>ConstantFP : This class represents a floating point constant.
2670 <ul>
2671 <li><tt>double getValue() const</tt>: Returns the underlying value of
2672 this constant. </li>
2673 </ul>
2674 </li>
2675 <li>ConstantArray : This represents a constant array.
2676 <ul>
2677 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
2678 a vector of component constants that makeup this array. </li>
2679 </ul>
2680 </li>
2681 <li>ConstantStruct : This represents a constant struct.
2682 <ul>
2683 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
2684 a vector of component constants that makeup this array. </li>
2685 </ul>
2686 </li>
2687 <li>GlobalValue : This represents either a global variable or a function. In
2688 either case, the value is a constant fixed address (after linking).
2689 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2690</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2691</div>
2692
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2693
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2694<!-- ======================================================================= -->
2695<div class="doc_subsection">
2696 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
2697</div>
2698
2699<div class="doc_text">
2700
2701<p><tt>#include "<a
2702href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]2703doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
2704Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2705Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
2706<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]2707
2708<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
2709href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
2710visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
2711Because they are visible at global scope, they are also subject to linking with
2712other globals defined in different translation units. To control the linking
2713process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
2714<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]2715defined by the <tt>LinkageTypes</tt> enumeration.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2716
2717<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
2718<tt>static</tt> in C), it is not visible to code outside the current translation
2719unit, and does not participate in linking. If it has external linkage, it is
2720visible to external code, and does participate in linking. In addition to
2721linkage information, <tt>GlobalValue</tt>s keep track of which <a
2722href="#Module"><tt>Module</tt></a> they are currently part of.</p>
2723
2724<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
2725by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
2726global is always a pointer to its contents. It is important to remember this
2727when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
2728be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
2729subclass 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]2730i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2731the address of the first element of this array and the value of the
2732<tt>GlobalVariable</tt> are the same, they have different types. The
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2733<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
2734is <tt>i32.</tt> Because of this, accessing a global value requires you to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2735dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
2736can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
2737Language Reference Manual</a>.</p>
2738
2739</div>
2740
2741<!-- _______________________________________________________________________ -->
2742<div class="doc_subsubsection">
2743 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
2744 class</a>
2745</div>
2746
2747<div class="doc_text">
2748
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2749<ul>
2750 <li><tt>bool hasInternalLinkage() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2751 <tt>bool hasExternalLinkage() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2752 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
2753 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
2754 <p> </p>
2755 </li>
2756 <li><tt><a href="#Module">Module</a> *getParent()</tt>
2757 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2758GlobalValue is currently embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2759</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2760
2761</div>
2762
2763<!-- ======================================================================= -->
2764<div class="doc_subsection">
2765 <a name="Function">The <tt>Function</tt> class</a>
2766</div>
2767
2768<div class="doc_text">
2769
2770<p><tt>#include "<a
2771href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2772info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2773Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
2774<a href="#Constant"><tt>Constant</tt></a>,
2775<a href="#User"><tt>User</tt></a>,
2776<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2777
2778<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
2779actually one of the more complex classes in the LLVM heirarchy because it must
2780keep track of a large amount of data. The <tt>Function</tt> class keeps track
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2781of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
2782<a href="#Argument"><tt>Argument</tt></a>s, and a
2783<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2784
2785<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
2786commonly used part of <tt>Function</tt> objects. The list imposes an implicit
2787ordering of the blocks in the function, which indicate how the code will be
2788layed out by the backend. Additionally, the first <a
2789href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
2790<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
2791block. There are no implicit exit nodes, and in fact there may be multiple exit
2792nodes from a single <tt>Function</tt>. If the <a
2793href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
2794the <tt>Function</tt> is actually a function declaration: the actual body of the
2795function hasn't been linked in yet.</p>
2796
2797<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
2798<tt>Function</tt> class also keeps track of the list of formal <a
2799href="#Argument"><tt>Argument</tt></a>s that the function receives. This
2800container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
2801nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
2802the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
2803
2804<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
2805LLVM feature that is only used when you have to look up a value by name. Aside
2806from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
2807internally to make sure that there are not conflicts between the names of <a
2808href="#Instruction"><tt>Instruction</tt></a>s, <a
2809href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
2810href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
2811
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]2812<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
2813and therefore also a <a href="#Constant">Constant</a>. The value of the function
2814is its address (after linking) which is guaranteed to be constant.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2815</div>
2816
2817<!-- _______________________________________________________________________ -->
2818<div class="doc_subsubsection">
2819 <a name="m_Function">Important Public Members of the <tt>Function</tt>
2820 class</a>
2821</div>
2822
2823<div class="doc_text">
2824
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2825<ul>
2826 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]2827 *Ty, LinkageTypes Linkage, const std::string &amp;N = "", Module* Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2828
2829 <p>Constructor used when you need to create new <tt>Function</tt>s to add
2830 the the program. The constructor must specify the type of the function to
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]2831 create and what type of linkage the function should have. The <a
2832 href="#FunctionType"><tt>FunctionType</tt></a> argument
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2833 specifies the formal arguments and return value for the function. The same
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2834 <a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2835 create multiple functions. The <tt>Parent</tt> argument specifies the Module
2836 in which the function is defined. If this argument is provided, the function
2837 will automatically be inserted into that module's list of
2838 functions.</p></li>
2839
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2840 <li><tt>bool isExternal()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2841
2842 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
2843 function is "external", it does not have a body, and thus must be resolved
2844 by linking with a function defined in a different translation unit.</p></li>
2845
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2846 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2847 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2848
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]2849 <tt>begin()</tt>, <tt>end()</tt>
2850 <tt>size()</tt>, <tt>empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2851
2852 <p>These are forwarding methods that make it easy to access the contents of
2853 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
2854 list.</p></li>
2855
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2856 <li><tt>Function::BasicBlockListType &amp;getBasicBlockList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2857
2858 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
2859 is necessary to use when you need to update the list or perform a complex
2860 action that doesn't have a forwarding method.</p></li>
2861
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]2862 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2863iterator<br>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]2864 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2865
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]2866 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]2867 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2868
2869 <p>These are forwarding methods that make it easy to access the contents of
2870 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
2871 list.</p></li>
2872
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2873 <li><tt>Function::ArgumentListType &amp;getArgumentList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2874
2875 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
2876 necessary to use when you need to update the list or perform a complex
2877 action that doesn't have a forwarding method.</p></li>
2878
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2879 <li><tt><a href="#BasicBlock">BasicBlock</a> &amp;getEntryBlock()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2880
2881 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
2882 function. Because the entry block for the function is always the first
2883 block, this returns the first block of the <tt>Function</tt>.</p></li>
2884
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2885 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
2886 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2887
2888 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
2889 <tt>Function</tt> and returns the return type of the function, or the <a
2890 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
2891 function.</p></li>
2892
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2893 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2894
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2895 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2896 for this <tt>Function</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2897</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2898
2899</div>
2900
2901<!-- ======================================================================= -->
2902<div class="doc_subsection">
2903 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
2904</div>
2905
2906<div class="doc_text">
2907
2908<p><tt>#include "<a
2909href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
2910<br>
Tanya Lattnera3da7772004-06-22 08:02:25 +0000[diff] [blame]2911doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2912 Class</a><br>
2913Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
2914<a href="#Constant"><tt>Constant</tt></a>,
2915<a href="#User"><tt>User</tt></a>,
2916<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2917
2918<p>Global variables are represented with the (suprise suprise)
2919<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
2920subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
2921always referenced by their address (global values must live in memory, so their
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]2922"name" refers to their constant address). See
2923<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
2924variables may have an initial value (which must be a
2925<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
2926they may be marked as "constant" themselves (indicating that their contents
2927never change at runtime).</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2928</div>
2929
2930<!-- _______________________________________________________________________ -->
2931<div class="doc_subsubsection">
2932 <a name="m_GlobalVariable">Important Public Members of the
2933 <tt>GlobalVariable</tt> class</a>
2934</div>
2935
2936<div class="doc_text">
2937
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2938<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2939 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
2940 isConstant, LinkageTypes&amp; Linkage, <a href="#Constant">Constant</a>
2941 *Initializer = 0, const std::string &amp;Name = "", Module* Parent = 0)</tt>
2942
2943 <p>Create a new global variable of the specified type. If
2944 <tt>isConstant</tt> is true then the global variable will be marked as
2945 unchanging for the program. The Linkage parameter specifies the type of
2946 linkage (internal, external, weak, linkonce, appending) for the variable. If
2947 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage,&nbsp; then
2948 the resultant global variable will have internal linkage. AppendingLinkage
2949 concatenates together all instances (in different translation units) of the
2950 variable into a single variable but is only applicable to arrays. &nbsp;See
2951 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
2952 further details on linkage types. Optionally an initializer, a name, and the
2953 module to put the variable into may be specified for the global variable as
2954 well.</p></li>
2955
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2956 <li><tt>bool isConstant() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2957
2958 <p>Returns true if this is a global variable that is known not to
2959 be modified at runtime.</p></li>
2960
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2961 <li><tt>bool hasInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2962
2963 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
2964
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2965 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2966
2967 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
2968 to call this method if there is no initializer.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2969</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2970
2971</div>
2972
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2973
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2974<!-- ======================================================================= -->
2975<div class="doc_subsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2976 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2977</div>
2978
2979<div class="doc_text">
2980
2981<p><tt>#include "<a
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2982href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
2983doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
2984Class</a><br>
2985Superclass: <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2986
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2987<p>This class represents a single entry multiple exit section of the code,
2988commonly known as a basic block by the compiler community. The
2989<tt>BasicBlock</tt> class maintains a list of <a
2990href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
2991Matching the language definition, the last element of this list of instructions
2992is always a terminator instruction (a subclass of the <a
2993href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
2994
2995<p>In addition to tracking the list of instructions that make up the block, the
2996<tt>BasicBlock</tt> class also keeps track of the <a
2997href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
2998
2999<p>Note that <tt>BasicBlock</tt>s themselves are <a
3000href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
3001like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
3002<tt>label</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3003
3004</div>
3005
3006<!-- _______________________________________________________________________ -->
3007<div class="doc_subsubsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3008 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
3009 class</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3010</div>
3011
3012<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3013<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3014
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3015<li><tt>BasicBlock(const std::string &amp;Name = "", </tt><tt><a
3016 href="#Function">Function</a> *Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3017
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3018<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
3019insertion into a function. The constructor optionally takes a name for the new
3020block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
3021the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
3022automatically inserted at the end of the specified <a
3023href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
3024manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3025
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3026<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
3027<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
3028<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
3029<tt>size()</tt>, <tt>empty()</tt>
3030STL-style functions for accessing the instruction list.
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3031
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3032<p>These methods and typedefs are forwarding functions that have the same
3033semantics as the standard library methods of the same names. These methods
3034expose the underlying instruction list of a basic block in a way that is easy to
3035manipulate. To get the full complement of container operations (including
3036operations to update the list), you must use the <tt>getInstList()</tt>
3037method.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3038
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3039<li><tt>BasicBlock::InstListType &amp;getInstList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3040
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3041<p>This method is used to get access to the underlying container that actually
3042holds the Instructions. This method must be used when there isn't a forwarding
3043function in the <tt>BasicBlock</tt> class for the operation that you would like
3044to perform. Because there are no forwarding functions for "updating"
3045operations, you need to use this if you want to update the contents of a
3046<tt>BasicBlock</tt>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3047
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3048<li><tt><a href="#Function">Function</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3049
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3050<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
3051embedded into, or a null pointer if it is homeless.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3052
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3053<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3054
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3055<p> Returns a pointer to the terminator instruction that appears at the end of
3056the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
3057instruction in the block is not a terminator, then a null pointer is
3058returned.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3059
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3060</ul>
3061
3062</div>
3063
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3064
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3065<!-- ======================================================================= -->
3066<div class="doc_subsection">
3067 <a name="Argument">The <tt>Argument</tt> class</a>
3068</div>
3069
3070<div class="doc_text">
3071
3072<p>This subclass of Value defines the interface for incoming formal
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]3073arguments to a function. A Function maintains a list of its formal
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3074arguments. An argument has a pointer to the parent Function.</p>
3075
3076</div>
3077
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3078<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3079<hr>
3080<address>
3081 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3082 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3083 <a href="http://validator.w3.org/check/referer"><img
3084 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3085
3086 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
3087 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]3088 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3089 Last modified: $Date$
3090</address>
3091
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3092</body>
3093</html>