Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / 28e6ff57d4d52c4b2934f223731a0c24dc14a624 / . / docs / ProgrammersManual.html
blob: 46b82ddc21fb6f23c335d58a578b248562ae68c7 [file] [log] [blame]
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Programmer's Manual</title>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]6 <link rel="stylesheet" href="llvm.css" type="text/css">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]7</head>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]8<body>
9
10<div class="doc_title">
11 LLVM Programmer's Manual
12</div>
13
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]14<ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]15 <li><a href="#introduction">Introduction</a></li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]16 <li><a href="#general">General Information</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]17 <ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]18 <li><a href="#stl">The C++ Standard Template Library</a></li>
19<!--
20 <li>The <tt>-time-passes</tt> option</li>
21 <li>How to use the LLVM Makefile system</li>
22 <li>How to write a regression test</li>
Chris Lattner61db4652004-12-08 19:05:44 +0000[diff] [blame]23
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]24-->
Chris Lattner84b7f8d2003-08-01 22:20:59 +0000[diff] [blame]25 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]26 </li>
27 <li><a href="#apis">Important and useful LLVM APIs</a>
28 <ul>
29 <li><a href="#isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt>
30and <tt>dyn_cast&lt;&gt;</tt> templates</a> </li>
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]31 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]32option</a>
33 <ul>
34 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
35and the <tt>-debug-only</tt> option</a> </li>
36 </ul>
37 </li>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]38 <li><a href="#Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]39option</a></li>
40<!--
41 <li>The <tt>InstVisitor</tt> template
42 <li>The general graph API
43-->
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]44 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]45 </ul>
46 </li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]47 <li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
48 <ul>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]49 <li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
50 <ul>
51 <li><a href="#dss_fixedarrays">Fixed Size Arrays</a></li>
52 <li><a href="#dss_heaparrays">Heap Allocated Arrays</a></li>
53 <li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
54 <li><a href="#dss_vector">&lt;vector&gt;</a></li>
55 <li><a href="#dss_deque">&lt;deque&gt;</a></li>
56 <li><a href="#dss_list">&lt;list&gt;</a></li>
57 <li><a href="#dss_ilist">llvm/ADT/ilist</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]58 <li><a href="#dss_other">Other Sequential Container Options</a></li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]59 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]60 <li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
61 <ul>
62 <li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
63 <li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
64 <li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]65 <li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]66 <li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
67 <li><a href="#dss_set">&lt;set&gt;</a></li>
68 <li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]69 <li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
70 <li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]71 </ul></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]72 <li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
73 <ul>
74 <li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]75 <li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]76 <li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
77 <li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
78 <li><a href="#dss_map">&lt;map&gt;</a></li>
79 <li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
80 </ul></li>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]81 <li><a href="#ds_bit">BitVector-like containers</a>
82 <ul>
83 <li><a href="#dss_bitvector">A dense bitvector</a></li>
84 <li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
85 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]86 </ul>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]87 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]88 <li><a href="#common">Helpful Hints for Common Operations</a>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]89 <ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]90 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
91 <ul>
92 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
93in a <tt>Function</tt></a> </li>
94 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
95in a <tt>BasicBlock</tt></a> </li>
96 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
97in a <tt>Function</tt></a> </li>
98 <li><a href="#iterate_convert">Turning an iterator into a
99class pointer</a> </li>
100 <li><a href="#iterate_complex">Finding call sites: a more
101complex example</a> </li>
102 <li><a href="#calls_and_invokes">Treating calls and invokes
103the same way</a> </li>
104 <li><a href="#iterate_chains">Iterating over def-use &amp;
105use-def chains</a> </li>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]106 <li><a href="#iterate_preds">Iterating over predecessors &amp;
107successors of blocks</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]108 </ul>
109 </li>
110 <li><a href="#simplechanges">Making simple changes</a>
111 <ul>
112 <li><a href="#schanges_creating">Creating and inserting new
113 <tt>Instruction</tt>s</a> </li>
114 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
115 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
116with another <tt>Value</tt></a> </li>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]117 <li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]118 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]119 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]120<!--
121 <li>Working with the Control Flow Graph
122 <ul>
123 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
124 <li>
125 <li>
126 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]127-->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]128 </ul>
129 </li>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]130
131 <li><a href="#advanced">Advanced Topics</a>
132 <ul>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]133 <li><a href="#TypeResolve">LLVM Type Resolution</a>
134 <ul>
135 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
136 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
137 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
138 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
139 </ul></li>
140
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]141 <li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> and <tt>TypeSymbolTable</tt> classes</a></li>
142 <li><a href="#UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a></li>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]143 </ul></li>
144
Joel Stanley9b96c442002-09-06 21:55:13 +0000[diff] [blame]145 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]146 <ul>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]147 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]148 <li><a href="#Module">The <tt>Module</tt> class</a></li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]149 <li><a href="#Value">The <tt>Value</tt> class</a>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]150 <ul>
151 <li><a href="#User">The <tt>User</tt> class</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]152 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]153 <li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
154 <li><a href="#Constant">The <tt>Constant</tt> class</a>
155 <ul>
156 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]157 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]158 <li><a href="#Function">The <tt>Function</tt> class</a></li>
159 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
160 </ul>
161 </li>
162 </ul>
163 </li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]164 </ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]165 </li>
166 <li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
167 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
168 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]169 </li>
170 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]171 </li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]172</ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]173
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]174<div class="doc_author">
175 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
Chris Lattner94c43592004-05-26 16:52:55 +0000[diff] [blame]176 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]177 <a href="mailto:ggreif@gmail.com">Gabor Greif</a>,
178 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a> and
Chris Lattner94c43592004-05-26 16:52:55 +0000[diff] [blame]179 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]180</div>
181
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]182<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]183<div class="doc_section">
184 <a name="introduction">Introduction </a>
185</div>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]186<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]187
188<div class="doc_text">
189
190<p>This document is meant to highlight some of the important classes and
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]191interfaces available in the LLVM source-base. This manual is not
192intended to explain what LLVM is, how it works, and what LLVM code looks
193like. It assumes that you know the basics of LLVM and are interested
194in writing transformations or otherwise analyzing or manipulating the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]195code.</p>
196
197<p>This document should get you oriented so that you can find your
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]198way in the continuously growing source code that makes up the LLVM
199infrastructure. Note that this manual is not intended to serve as a
200replacement for reading the source code, so if you think there should be
201a method in one of these classes to do something, but it's not listed,
202check the source. Links to the <a href="/doxygen/">doxygen</a> sources
203are provided to make this as easy as possible.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]204
205<p>The first section of this document describes general information that is
206useful to know when working in the LLVM infrastructure, and the second describes
207the Core LLVM classes. In the future this manual will be extended with
208information describing how to use extension libraries, such as dominator
209information, CFG traversal routines, and useful utilities like the <tt><a
210href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
211
212</div>
213
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]214<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]215<div class="doc_section">
216 <a name="general">General Information</a>
217</div>
218<!-- *********************************************************************** -->
219
220<div class="doc_text">
221
222<p>This section contains general information that is useful if you are working
223in the LLVM source-base, but that isn't specific to any particular API.</p>
224
225</div>
226
227<!-- ======================================================================= -->
228<div class="doc_subsection">
229 <a name="stl">The C++ Standard Template Library</a>
230</div>
231
232<div class="doc_text">
233
234<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]235perhaps much more than you are used to, or have seen before. Because of
236this, you might want to do a little background reading in the
237techniques used and capabilities of the library. There are many good
238pages that discuss the STL, and several books on the subject that you
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]239can get, so it will not be discussed in this document.</p>
240
241<p>Here are some useful links:</p>
242
243<ol>
244
245<li><a href="http://www.dinkumware.com/refxcpp.html">Dinkumware C++ Library
246reference</a> - an excellent reference for the STL and other parts of the
247standard C++ library.</li>
248
249<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]250O'Reilly book in the making. It has a decent
251Standard Library
252Reference 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]253published.</li>
254
255<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
256Questions</a></li>
257
258<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
259Contains a useful <a
260href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
261STL</a>.</li>
262
263<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
264Page</a></li>
265
Tanya Lattner79445ba2004-12-08 18:34:56 +0000[diff] [blame]266<li><a href="http://64.78.49.204/">
Reid Spencer096603a2004-05-26 08:41:35 +0000[diff] [blame]267Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
268the book).</a></li>
269
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]270</ol>
271
272<p>You are also encouraged to take a look at the <a
273href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
274to write maintainable code more than where to put your curly braces.</p>
275
276</div>
277
278<!-- ======================================================================= -->
279<div class="doc_subsection">
280 <a name="stl">Other useful references</a>
281</div>
282
283<div class="doc_text">
284
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]285<ol>
286<li><a href="http://www.psc.edu/%7Esemke/cvs_branches.html">CVS
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]287Branch and Tag Primer</a></li>
Misha Brukmana0f71e42004-06-18 18:39:00 +0000[diff] [blame]288<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
289static and shared libraries across platforms</a></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]290</ol>
291
292</div>
293
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]294<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]295<div class="doc_section">
296 <a name="apis">Important and useful LLVM APIs</a>
297</div>
298<!-- *********************************************************************** -->
299
300<div class="doc_text">
301
302<p>Here we highlight some LLVM APIs that are generally useful and good to
303know about when writing transformations.</p>
304
305</div>
306
307<!-- ======================================================================= -->
308<div class="doc_subsection">
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]309 <a name="isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt> and
310 <tt>dyn_cast&lt;&gt;</tt> templates</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]311</div>
312
313<div class="doc_text">
314
315<p>The LLVM source-base makes extensive use of a custom form of RTTI.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]316These templates have many similarities to the C++ <tt>dynamic_cast&lt;&gt;</tt>
317operator, but they don't have some drawbacks (primarily stemming from
318the fact that <tt>dynamic_cast&lt;&gt;</tt> only works on classes that
319have a v-table). Because they are used so often, you must know what they
320do and how they work. All of these templates are defined in the <a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]321 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]322file (note that you very rarely have to include this file directly).</p>
323
324<dl>
325 <dt><tt>isa&lt;&gt;</tt>: </dt>
326
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]327 <dd><p>The <tt>isa&lt;&gt;</tt> operator works exactly like the Java
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]328 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
329 a reference or pointer points to an instance of the specified class. This can
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]330 be very useful for constraint checking of various sorts (example below).</p>
331 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]332
333 <dt><tt>cast&lt;&gt;</tt>: </dt>
334
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]335 <dd><p>The <tt>cast&lt;&gt;</tt> operator is a "checked cast" operation. It
Chris Lattner28e6ff52008-06-20 05:03:17 +0000[diff] [blame^]336 converts a pointer or reference from a base class to a derived class, causing
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]337 an assertion failure if it is not really an instance of the right type. This
338 should be used in cases where you have some information that makes you believe
339 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]340 and <tt>cast&lt;&gt;</tt> template is:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]341
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]342<div class="doc_code">
343<pre>
344static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
345 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))
346 return true;
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]347
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]348 // <i>Otherwise, it must be an instruction...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]349 return !L-&gt;contains(cast&lt;<a href="#Instruction">Instruction</a>&gt;(V)-&gt;getParent());
350}
351</pre>
352</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]353
354 <p>Note that you should <b>not</b> use an <tt>isa&lt;&gt;</tt> test followed
355 by a <tt>cast&lt;&gt;</tt>, for that use the <tt>dyn_cast&lt;&gt;</tt>
356 operator.</p>
357
358 </dd>
359
360 <dt><tt>dyn_cast&lt;&gt;</tt>:</dt>
361
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]362 <dd><p>The <tt>dyn_cast&lt;&gt;</tt> operator is a "checking cast" operation.
363 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]364 pointer to it (this operator does not work with references). If the operand is
365 not of the correct type, a null pointer is returned. Thus, this works very
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]366 much like the <tt>dynamic_cast&lt;&gt;</tt> operator in C++, and should be
367 used in the same circumstances. Typically, the <tt>dyn_cast&lt;&gt;</tt>
368 operator is used in an <tt>if</tt> statement or some other flow control
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]369 statement like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]370
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]371<div class="doc_code">
372<pre>
373if (<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]374 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]375}
376</pre>
377</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]378
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]379 <p>This form of the <tt>if</tt> statement effectively combines together a call
380 to <tt>isa&lt;&gt;</tt> and a call to <tt>cast&lt;&gt;</tt> into one
381 statement, which is very convenient.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]382
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]383 <p>Note that the <tt>dyn_cast&lt;&gt;</tt> operator, like C++'s
384 <tt>dynamic_cast&lt;&gt;</tt> or Java's <tt>instanceof</tt> operator, can be
385 abused. In particular, you should not use big chained <tt>if/then/else</tt>
386 blocks to check for lots of different variants of classes. If you find
387 yourself wanting to do this, it is much cleaner and more efficient to use the
388 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]389
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]390 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]391
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]392 <dt><tt>cast_or_null&lt;&gt;</tt>: </dt>
393
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]394 <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]395 <tt>cast&lt;&gt;</tt> operator, except that it allows for a null pointer as an
396 argument (which it then propagates). This can sometimes be useful, allowing
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]397 you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]398
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]399 <dt><tt>dyn_cast_or_null&lt;&gt;</tt>: </dt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]400
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]401 <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]402 <tt>dyn_cast&lt;&gt;</tt> operator, except that it allows for a null pointer
403 as an argument (which it then propagates). This can sometimes be useful,
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]404 allowing you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]405
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]406</dl>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]407
408<p>These five templates can be used with any classes, whether they have a
409v-table or not. To add support for these templates, you simply need to add
410<tt>classof</tt> static methods to the class you are interested casting
411to. Describing this is currently outside the scope of this document, but there
412are lots of examples in the LLVM source base.</p>
413
414</div>
415
416<!-- ======================================================================= -->
417<div class="doc_subsection">
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]418 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]419</div>
420
421<div class="doc_text">
422
423<p>Often when working on your pass you will put a bunch of debugging printouts
424and other code into your pass. After you get it working, you want to remove
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]425it, 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]426across).</p>
427
428<p> Naturally, because of this, you don't want to delete the debug printouts,
429but you don't want them to always be noisy. A standard compromise is to comment
430them out, allowing you to enable them if you need them in the future.</p>
431
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]432<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]433file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
434this problem. Basically, you can put arbitrary code into the argument of the
435<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
436tool) is run with the '<tt>-debug</tt>' command line argument:</p>
437
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]438<div class="doc_code">
439<pre>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]440DOUT &lt;&lt; "I am here!\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]441</pre>
442</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]443
444<p>Then you can run your pass like this:</p>
445
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]446<div class="doc_code">
447<pre>
448$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]449<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]450$ opt &lt; a.bc &gt; /dev/null -mypass -debug
451I am here!
452</pre>
453</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]454
455<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
456to not have to create "yet another" command line option for the debug output for
457your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
458so they do not cause a performance impact at all (for the same reason, they
459should also not contain side-effects!).</p>
460
461<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
462enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
463"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
464program hasn't been started yet, you can always just run it with
465<tt>-debug</tt>.</p>
466
467</div>
468
469<!-- _______________________________________________________________________ -->
470<div class="doc_subsubsection">
Chris Lattnerc9151082005-04-26 22:57:07 +0000[diff] [blame]471 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]472 the <tt>-debug-only</tt> option</a>
473</div>
474
475<div class="doc_text">
476
477<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
478just turns on <b>too much</b> information (such as when working on the code
479generator). If you want to enable debug information with more fine-grained
480control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
481option as follows:</p>
482
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]483<div class="doc_code">
484<pre>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]485DOUT &lt;&lt; "No debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]486#undef DEBUG_TYPE
487#define DEBUG_TYPE "foo"
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]488DOUT &lt;&lt; "'foo' debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]489#undef DEBUG_TYPE
490#define DEBUG_TYPE "bar"
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]491DOUT &lt;&lt; "'bar' debug type\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]492#undef DEBUG_TYPE
493#define DEBUG_TYPE ""
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]494DOUT &lt;&lt; "No debug type (2)\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]495</pre>
496</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]497
498<p>Then you can run your pass like this:</p>
499
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]500<div class="doc_code">
501<pre>
502$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]503<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]504$ opt &lt; a.bc &gt; /dev/null -mypass -debug
505No debug type
506'foo' debug type
507'bar' debug type
508No debug type (2)
509$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=foo
510'foo' debug type
511$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=bar
512'bar' debug type
513</pre>
514</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]515
516<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
517a 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]518you <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]519<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
520"bar", because there is no system in place to ensure that names do not
521conflict. If two different modules use the same string, they will all be turned
522on when the name is specified. This allows, for example, all debug information
523for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]524even if the source lives in multiple files.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]525
526</div>
527
528<!-- ======================================================================= -->
529<div class="doc_subsection">
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]530 <a name="Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]531 option</a>
532</div>
533
534<div class="doc_text">
535
536<p>The "<tt><a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]537href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]538provides a class named <tt>Statistic</tt> that is used as a unified way to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]539keep track of what the LLVM compiler is doing and how effective various
540optimizations are. It is useful to see what optimizations are contributing to
541making a particular program run faster.</p>
542
543<p>Often you may run your pass on some big program, and you're interested to see
544how many times it makes a certain transformation. Although you can do this with
545hand 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]546for big programs. Using the <tt>Statistic</tt> class makes it very easy to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]547keep track of this information, and the calculated information is presented in a
548uniform manner with the rest of the passes being executed.</p>
549
550<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
551it are as follows:</p>
552
553<ol>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]554 <li><p>Define your statistic like this:</p>
555
556<div class="doc_code">
557<pre>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]558#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
559STATISTIC(NumXForms, "The # of times I did stuff");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]560</pre>
561</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]562
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]563 <p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
564 specified by the first argument. The pass name is taken from the DEBUG_TYPE
565 macro, and the description is taken from the second argument. The variable
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]566 defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]567
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]568 <li><p>Whenever you make a transformation, bump the counter:</p>
569
570<div class="doc_code">
571<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]572++NumXForms; // <i>I did stuff!</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]573</pre>
574</div>
575
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]576 </li>
577 </ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]578
579 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
580 statistics gathered, use the '<tt>-stats</tt>' option:</p>
581
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]582<div class="doc_code">
583<pre>
584$ opt -stats -mypassname &lt; program.bc &gt; /dev/null
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]585<i>... statistics output ...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]586</pre>
587</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]588
Reid Spencer6b6c73e2007-02-09 16:00:28 +0000[diff] [blame]589 <p> When running <tt>opt</tt> on a C file from the SPEC benchmark
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]590suite, it gives a report that looks like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]591
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]592<div class="doc_code">
593<pre>
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]594 7646 bitcodewriter - Number of normal instructions
595 725 bitcodewriter - Number of oversized instructions
596 129996 bitcodewriter - Number of bitcode bytes written
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]597 2817 raise - Number of insts DCEd or constprop'd
598 3213 raise - Number of cast-of-self removed
599 5046 raise - Number of expression trees converted
600 75 raise - Number of other getelementptr's formed
601 138 raise - Number of load/store peepholes
602 42 deadtypeelim - Number of unused typenames removed from symtab
603 392 funcresolve - Number of varargs functions resolved
604 27 globaldce - Number of global variables removed
605 2 adce - Number of basic blocks removed
606 134 cee - Number of branches revectored
607 49 cee - Number of setcc instruction eliminated
608 532 gcse - Number of loads removed
609 2919 gcse - Number of instructions removed
610 86 indvars - Number of canonical indvars added
611 87 indvars - Number of aux indvars removed
612 25 instcombine - Number of dead inst eliminate
613 434 instcombine - Number of insts combined
614 248 licm - Number of load insts hoisted
615 1298 licm - Number of insts hoisted to a loop pre-header
616 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
617 75 mem2reg - Number of alloca's promoted
618 1444 cfgsimplify - Number of blocks simplified
619</pre>
620</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]621
622<p>Obviously, with so many optimizations, having a unified framework for this
623stuff is very nice. Making your pass fit well into the framework makes it more
624maintainable and useful.</p>
625
626</div>
627
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]628<!-- ======================================================================= -->
629<div class="doc_subsection">
630 <a name="ViewGraph">Viewing graphs while debugging code</a>
631</div>
632
633<div class="doc_text">
634
635<p>Several of the important data structures in LLVM are graphs: for example
636CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
637LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
638<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
639DAGs</a>. In many cases, while debugging various parts of the compiler, it is
640nice to instantly visualize these graphs.</p>
641
642<p>LLVM provides several callbacks that are available in a debug build to do
643exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
644the current LLVM tool will pop up a window containing the CFG for the function
645where each basic block is a node in the graph, and each node contains the
646instructions in the block. Similarly, there also exists
647<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
648<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
649and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]650you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]651up a window. Alternatively, you can sprinkle calls to these functions in your
652code in places you want to debug.</p>
653
654<p>Getting this to work requires a small amount of configuration. On Unix
655systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
656toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
657Mac OS/X, download and install the Mac OS/X <a
658href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]659<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]660it) to your path. Once in your system and path are set up, rerun the LLVM
661configure script and rebuild LLVM to enable this functionality.</p>
662
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]663<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
664<i>interesting</i> nodes in large complex graphs. From gdb, if you
665<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]666next <tt>call DAG.viewGraph()</tt> would highlight the node in the
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]667specified color (choices of colors can be found at <a
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]668href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]669complex node attributes can be provided with <tt>call
670DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
671found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
672Attributes</a>.) If you want to restart and clear all the current graph
673attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
674
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]675</div>
676
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]677<!-- *********************************************************************** -->
678<div class="doc_section">
679 <a name="datastructure">Picking the Right Data Structure for a Task</a>
680</div>
681<!-- *********************************************************************** -->
682
683<div class="doc_text">
684
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]685<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
686 and we commonly use STL data structures. This section describes the trade-offs
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]687 you should consider when you pick one.</p>
688
689<p>
690The first step is a choose your own adventure: do you want a sequential
691container, a set-like container, or a map-like container? The most important
692thing when choosing a container is the algorithmic properties of how you plan to
693access the container. Based on that, you should use:</p>
694
695<ul>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]696<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]697 of an value based on another value. Map-like containers also support
698 efficient queries for containment (whether a key is in the map). Map-like
699 containers generally do not support efficient reverse mapping (values to
700 keys). If you need that, use two maps. Some map-like containers also
701 support efficient iteration through the keys in sorted order. Map-like
702 containers are the most expensive sort, only use them if you need one of
703 these capabilities.</li>
704
705<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
706 stuff into a container that automatically eliminates duplicates. Some
707 set-like containers support efficient iteration through the elements in
708 sorted order. Set-like containers are more expensive than sequential
709 containers.
710</li>
711
712<li>a <a href="#ds_sequential">sequential</a> container provides
713 the most efficient way to add elements and keeps track of the order they are
714 added to the collection. They permit duplicates and support efficient
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]715 iteration, but do not support efficient look-up based on a key.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]716</li>
717
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]718<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
719 perform set operations on sets of numeric id's, while automatically
720 eliminating duplicates. Bit containers require a maximum of 1 bit for each
721 identifier you want to store.
722</li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]723</ul>
724
725<p>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]726Once the proper category of container is determined, you can fine tune the
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]727memory use, constant factors, and cache behaviors of access by intelligently
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]728picking a member of the category. Note that constant factors and cache behavior
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]729can be a big deal. If you have a vector that usually only contains a few
730elements (but could contain many), for example, it's much better to use
731<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
732. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
733cost of adding the elements to the container. </p>
734
735</div>
736
737<!-- ======================================================================= -->
738<div class="doc_subsection">
739 <a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
740</div>
741
742<div class="doc_text">
743There are a variety of sequential containers available for you, based on your
744needs. Pick the first in this section that will do what you want.
745</div>
746
747<!-- _______________________________________________________________________ -->
748<div class="doc_subsubsection">
749 <a name="dss_fixedarrays">Fixed Size Arrays</a>
750</div>
751
752<div class="doc_text">
753<p>Fixed size arrays are very simple and very fast. They are good if you know
754exactly how many elements you have, or you have a (low) upper bound on how many
755you have.</p>
756</div>
757
758<!-- _______________________________________________________________________ -->
759<div class="doc_subsubsection">
760 <a name="dss_heaparrays">Heap Allocated Arrays</a>
761</div>
762
763<div class="doc_text">
764<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
765the number of elements is variable, if you know how many elements you will need
766before the array is allocated, and if the array is usually large (if not,
767consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
768allocated array is the cost of the new/delete (aka malloc/free). Also note that
769if you are allocating an array of a type with a constructor, the constructor and
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]770destructors will be run for every element in the array (re-sizable vectors only
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]771construct those elements actually used).</p>
772</div>
773
774<!-- _______________________________________________________________________ -->
775<div class="doc_subsubsection">
776 <a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
777</div>
778
779<div class="doc_text">
780<p><tt>SmallVector&lt;Type, N&gt;</tt> is a simple class that looks and smells
781just like <tt>vector&lt;Type&gt;</tt>:
782it supports efficient iteration, lays out elements in memory order (so you can
783do pointer arithmetic between elements), supports efficient push_back/pop_back
784operations, supports efficient random access to its elements, etc.</p>
785
786<p>The advantage of SmallVector is that it allocates space for
787some number of elements (N) <b>in the object itself</b>. Because of this, if
788the SmallVector is dynamically smaller than N, no malloc is performed. This can
789be a big win in cases where the malloc/free call is far more expensive than the
790code that fiddles around with the elements.</p>
791
792<p>This is good for vectors that are "usually small" (e.g. the number of
793predecessors/successors of a block is usually less than 8). On the other hand,
794this makes the size of the SmallVector itself large, so you don't want to
795allocate lots of them (doing so will waste a lot of space). As such,
796SmallVectors are most useful when on the stack.</p>
797
798<p>SmallVector also provides a nice portable and efficient replacement for
799<tt>alloca</tt>.</p>
800
801</div>
802
803<!-- _______________________________________________________________________ -->
804<div class="doc_subsubsection">
805 <a name="dss_vector">&lt;vector&gt;</a>
806</div>
807
808<div class="doc_text">
809<p>
810std::vector is well loved and respected. It is useful when SmallVector isn't:
811when the size of the vector is often large (thus the small optimization will
812rarely be a benefit) or if you will be allocating many instances of the vector
813itself (which would waste space for elements that aren't in the container).
814vector is also useful when interfacing with code that expects vectors :).
815</p>
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]816
817<p>One worthwhile note about std::vector: avoid code like this:</p>
818
819<div class="doc_code">
820<pre>
821for ( ... ) {
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]822 std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]823 use V;
824}
825</pre>
826</div>
827
828<p>Instead, write this as:</p>
829
830<div class="doc_code">
831<pre>
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]832std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]833for ( ... ) {
834 use V;
835 V.clear();
836}
837</pre>
838</div>
839
840<p>Doing so will save (at least) one heap allocation and free per iteration of
841the loop.</p>
842
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]843</div>
844
845<!-- _______________________________________________________________________ -->
846<div class="doc_subsubsection">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]847 <a name="dss_deque">&lt;deque&gt;</a>
848</div>
849
850<div class="doc_text">
851<p>std::deque is, in some senses, a generalized version of std::vector. Like
852std::vector, it provides constant time random access and other similar
853properties, but it also provides efficient access to the front of the list. It
854does not guarantee continuity of elements within memory.</p>
855
856<p>In exchange for this extra flexibility, std::deque has significantly higher
857constant factor costs than std::vector. If possible, use std::vector or
858something cheaper.</p>
859</div>
860
861<!-- _______________________________________________________________________ -->
862<div class="doc_subsubsection">
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]863 <a name="dss_list">&lt;list&gt;</a>
864</div>
865
866<div class="doc_text">
867<p>std::list is an extremely inefficient class that is rarely useful.
868It performs a heap allocation for every element inserted into it, thus having an
869extremely high constant factor, particularly for small data types. std::list
870also only supports bidirectional iteration, not random access iteration.</p>
871
872<p>In exchange for this high cost, std::list supports efficient access to both
873ends of the list (like std::deque, but unlike std::vector or SmallVector). In
874addition, the iterator invalidation characteristics of std::list are stronger
875than that of a vector class: inserting or removing an element into the list does
876not invalidate iterator or pointers to other elements in the list.</p>
877</div>
878
879<!-- _______________________________________________________________________ -->
880<div class="doc_subsubsection">
881 <a name="dss_ilist">llvm/ADT/ilist</a>
882</div>
883
884<div class="doc_text">
885<p><tt>ilist&lt;T&gt;</tt> implements an 'intrusive' doubly-linked list. It is
886intrusive, because it requires the element to store and provide access to the
887prev/next pointers for the list.</p>
888
889<p>ilist has the same drawbacks as std::list, and additionally requires an
890ilist_traits implementation for the element type, but it provides some novel
891characteristics. In particular, it can efficiently store polymorphic objects,
892the traits class is informed when an element is inserted or removed from the
893list, and ilists are guaranteed to support a constant-time splice operation.
894</p>
895
896<p>These properties are exactly what we want for things like Instructions and
897basic blocks, which is why these are implemented with ilists.</p>
898</div>
899
900<!-- _______________________________________________________________________ -->
901<div class="doc_subsubsection">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]902 <a name="dss_other">Other Sequential Container options</a>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]903</div>
904
905<div class="doc_text">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]906<p>Other STL containers are available, such as std::string.</p>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]907
908<p>There are also various STL adapter classes such as std::queue,
909std::priority_queue, std::stack, etc. These provide simplified access to an
910underlying container but don't affect the cost of the container itself.</p>
911
912</div>
913
914
915<!-- ======================================================================= -->
916<div class="doc_subsection">
917 <a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
918</div>
919
920<div class="doc_text">
921
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]922<p>Set-like containers are useful when you need to canonicalize multiple values
923into a single representation. There are several different choices for how to do
924this, providing various trade-offs.</p>
925
926</div>
927
928
929<!-- _______________________________________________________________________ -->
930<div class="doc_subsubsection">
931 <a name="dss_sortedvectorset">A sorted 'vector'</a>
932</div>
933
934<div class="doc_text">
935
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]936<p>If you intend to insert a lot of elements, then do a lot of queries, a
937great approach is to use a vector (or other sequential container) with
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]938std::sort+std::unique to remove duplicates. This approach works really well if
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]939your usage pattern has these two distinct phases (insert then query), and can be
940coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
941</p>
942
943<p>
944This combination provides the several nice properties: the result data is
945contiguous in memory (good for cache locality), has few allocations, is easy to
946address (iterators in the final vector are just indices or pointers), and can be
947efficiently queried with a standard binary or radix search.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]948
949</div>
950
951<!-- _______________________________________________________________________ -->
952<div class="doc_subsubsection">
953 <a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
954</div>
955
956<div class="doc_text">
957
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]958<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]959are 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]960has space for N elements in place (thus, if the set is dynamically smaller than
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]961N, no malloc traffic is required) and accesses them with a simple linear search.
962When the set grows beyond 'N' elements, it allocates a more expensive representation that
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]963guarantees efficient access (for most types, it falls back to std::set, but for
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]964pointers it uses something far better, <a
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]965href="#dss_smallptrset">SmallPtrSet</a>).</p>
966
967<p>The magic of this class is that it handles small sets extremely efficiently,
968but gracefully handles extremely large sets without loss of efficiency. The
969drawback is that the interface is quite small: it supports insertion, queries
970and erasing, but does not support iteration.</p>
971
972</div>
973
974<!-- _______________________________________________________________________ -->
975<div class="doc_subsubsection">
976 <a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
977</div>
978
979<div class="doc_text">
980
981<p>SmallPtrSet has all the advantages of SmallSet (and a SmallSet of pointers is
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]982transparently implemented with a SmallPtrSet), but also supports iterators. If
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]983more than 'N' insertions are performed, a single quadratically
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]984probed hash table is allocated and grows as needed, providing extremely
985efficient access (constant time insertion/deleting/queries with low constant
986factors) and is very stingy with malloc traffic.</p>
987
988<p>Note that, unlike std::set, the iterators of SmallPtrSet are invalidated
989whenever an insertion occurs. Also, the values visited by the iterators are not
990visited in sorted order.</p>
991
992</div>
993
994<!-- _______________________________________________________________________ -->
995<div class="doc_subsubsection">
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]996 <a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
997</div>
998
999<div class="doc_text">
1000
1001<p>
1002DenseSet is a simple quadratically probed hash table. It excels at supporting
1003small values: it uses a single allocation to hold all of the pairs that
1004are currently inserted in the set. DenseSet is a great way to unique small
1005values that are not simple pointers (use <a
1006href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
1007the same requirements for the value type that <a
1008href="#dss_densemap">DenseMap</a> has.
1009</p>
1010
1011</div>
1012
1013<!-- _______________________________________________________________________ -->
1014<div class="doc_subsubsection">
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1015 <a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
1016</div>
1017
1018<div class="doc_text">
1019
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1020<p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1021FoldingSet is an aggregate class that is really good at uniquing
1022expensive-to-create or polymorphic objects. It is a combination of a chained
1023hash table with intrusive links (uniqued objects are required to inherit from
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1024FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
1025its ID process.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1026
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1027<p>Consider a case where you want to implement a "getOrCreateFoo" method for
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1028a complex object (for example, a node in the code generator). The client has a
1029description of *what* it wants to generate (it knows the opcode and all the
1030operands), 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]1031only to find out it already exists, at which point we would have to delete it
1032and return the node that already exists.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1033</p>
1034
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1035<p>To support this style of client, FoldingSet perform a query with a
1036FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1037element that we want to query for. The query either returns the element
1038matching the ID or it returns an opaque ID that indicates where insertion should
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1039take place. Construction of the ID usually does not require heap traffic.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1040
1041<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
1042in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1043Because the elements are individually allocated, pointers to the elements are
1044stable: inserting or removing elements does not invalidate any pointers to other
1045elements.
1046</p>
1047
1048</div>
1049
1050<!-- _______________________________________________________________________ -->
1051<div class="doc_subsubsection">
1052 <a name="dss_set">&lt;set&gt;</a>
1053</div>
1054
1055<div class="doc_text">
1056
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1057<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
1058many things but great at nothing. std::set allocates memory for each element
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1059inserted (thus it is very malloc intensive) and typically stores three pointers
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1060per element in the set (thus adding a large amount of per-element space
1061overhead). It offers guaranteed log(n) performance, which is not particularly
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1062fast from a complexity standpoint (particularly if the elements of the set are
1063expensive to compare, like strings), and has extremely high constant factors for
1064lookup, insertion and removal.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1065
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1066<p>The advantages of std::set are that its iterators are stable (deleting or
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1067inserting an element from the set does not affect iterators or pointers to other
1068elements) and that iteration over the set is guaranteed to be in sorted order.
1069If the elements in the set are large, then the relative overhead of the pointers
1070and malloc traffic is not a big deal, but if the elements of the set are small,
1071std::set is almost never a good choice.</p>
1072
1073</div>
1074
1075<!-- _______________________________________________________________________ -->
1076<div class="doc_subsubsection">
1077 <a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
1078</div>
1079
1080<div class="doc_text">
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1081<p>LLVM's SetVector&lt;Type&gt; is an adapter class that combines your choice of
1082a set-like container along with a <a href="#ds_sequential">Sequential
1083Container</a>. The important property
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1084that this provides is efficient insertion with uniquing (duplicate elements are
1085ignored) with iteration support. It implements this by inserting elements into
1086both a set-like container and the sequential container, using the set-like
1087container for uniquing and the sequential container for iteration.
1088</p>
1089
1090<p>The difference between SetVector and other sets is that the order of
1091iteration is guaranteed to match the order of insertion into the SetVector.
1092This property is really important for things like sets of pointers. Because
1093pointer values are non-deterministic (e.g. vary across runs of the program on
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1094different machines), iterating over the pointers in the set will
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1095not be in a well-defined order.</p>
1096
1097<p>
1098The drawback of SetVector is that it requires twice as much space as a normal
1099set and has the sum of constant factors from the set-like container and the
1100sequential container that it uses. Use it *only* if you need to iterate over
1101the elements in a deterministic order. SetVector is also expensive to delete
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1102elements out of (linear time), unless you use it's "pop_back" method, which is
1103faster.
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1104</p>
1105
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1106<p>SetVector is an adapter class that defaults to using std::vector and std::set
1107for the underlying containers, so it is quite expensive. However,
1108<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
1109defaults to using a SmallVector and SmallSet of a specified size. If you use
1110this, and if your sets are dynamically smaller than N, you will save a lot of
1111heap traffic.</p>
1112
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1113</div>
1114
1115<!-- _______________________________________________________________________ -->
1116<div class="doc_subsubsection">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1117 <a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
1118</div>
1119
1120<div class="doc_text">
1121
1122<p>
1123UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
1124retains a unique ID for each element inserted into the set. It internally
1125contains a map and a vector, and it assigns a unique ID for each value inserted
1126into the set.</p>
1127
1128<p>UniqueVector is very expensive: its cost is the sum of the cost of
1129maintaining both the map and vector, it has high complexity, high constant
1130factors, and produces a lot of malloc traffic. It should be avoided.</p>
1131
1132</div>
1133
1134
1135<!-- _______________________________________________________________________ -->
1136<div class="doc_subsubsection">
1137 <a name="dss_otherset">Other Set-Like Container Options</a>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1138</div>
1139
1140<div class="doc_text">
1141
1142<p>
1143The STL provides several other options, such as std::multiset and the various
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1144"hash_set" like containers (whether from C++ TR1 or from the SGI library).</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1145
1146<p>std::multiset is useful if you're not interested in elimination of
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1147duplicates, but has all the drawbacks of std::set. A sorted vector (where you
1148don't delete duplicate entries) or some other approach is almost always
1149better.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1150
1151<p>The various hash_set implementations (exposed portably by
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1152"llvm/ADT/hash_set") is a simple chained hashtable. This algorithm is as malloc
1153intensive as std::set (performing an allocation for each element inserted,
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1154thus having really high constant factors) but (usually) provides O(1)
1155insertion/deletion of elements. This can be useful if your elements are large
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1156(thus making the constant-factor cost relatively low) or if comparisons are
1157expensive. Element iteration does not visit elements in a useful order.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1158
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1159</div>
1160
1161<!-- ======================================================================= -->
1162<div class="doc_subsection">
1163 <a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
1164</div>
1165
1166<div class="doc_text">
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1167Map-like containers are useful when you want to associate data to a key. As
1168usual, there are a lot of different ways to do this. :)
1169</div>
1170
1171<!-- _______________________________________________________________________ -->
1172<div class="doc_subsubsection">
1173 <a name="dss_sortedvectormap">A sorted 'vector'</a>
1174</div>
1175
1176<div class="doc_text">
1177
1178<p>
1179If your usage pattern follows a strict insert-then-query approach, you can
1180trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
1181for set-like containers</a>. The only difference is that your query function
1182(which uses std::lower_bound to get efficient log(n) lookup) should only compare
1183the key, not both the key and value. This yields the same advantages as sorted
1184vectors for sets.
1185</p>
1186</div>
1187
1188<!-- _______________________________________________________________________ -->
1189<div class="doc_subsubsection">
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1190 <a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1191</div>
1192
1193<div class="doc_text">
1194
1195<p>
1196Strings are commonly used as keys in maps, and they are difficult to support
1197efficiently: they are variable length, inefficient to hash and compare when
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1198long, expensive to copy, etc. StringMap is a specialized container designed to
1199cope with these issues. It supports mapping an arbitrary range of bytes to an
1200arbitrary other object.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1201
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1202<p>The StringMap implementation uses a quadratically-probed hash table, where
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1203the buckets store a pointer to the heap allocated entries (and some other
1204stuff). The entries in the map must be heap allocated because the strings are
1205variable length. The string data (key) and the element object (value) are
1206stored in the same allocation with the string data immediately after the element
1207object. This container guarantees the "<tt>(char*)(&amp;Value+1)</tt>" points
1208to the key string for a value.</p>
1209
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1210<p>The StringMap is very fast for several reasons: quadratic probing is very
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1211cache efficient for lookups, the hash value of strings in buckets is not
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1212recomputed when lookup up an element, StringMap rarely has to touch the
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1213memory for unrelated objects when looking up a value (even when hash collisions
1214happen), hash table growth does not recompute the hash values for strings
1215already in the table, and each pair in the map is store in a single allocation
1216(the string data is stored in the same allocation as the Value of a pair).</p>
1217
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1218<p>StringMap also provides query methods that take byte ranges, so it only ever
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1219copies a string if a value is inserted into the table.</p>
1220</div>
1221
1222<!-- _______________________________________________________________________ -->
1223<div class="doc_subsubsection">
1224 <a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
1225</div>
1226
1227<div class="doc_text">
1228<p>
1229IndexedMap is a specialized container for mapping small dense integers (or
1230values that can be mapped to small dense integers) to some other type. It is
1231internally implemented as a vector with a mapping function that maps the keys to
1232the dense integer range.
1233</p>
1234
1235<p>
1236This is useful for cases like virtual registers in the LLVM code generator: they
1237have a dense mapping that is offset by a compile-time constant (the first
1238virtual register ID).</p>
1239
1240</div>
1241
1242<!-- _______________________________________________________________________ -->
1243<div class="doc_subsubsection">
1244 <a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
1245</div>
1246
1247<div class="doc_text">
1248
1249<p>
1250DenseMap is a simple quadratically probed hash table. It excels at supporting
1251small keys and values: it uses a single allocation to hold all of the pairs that
1252are currently inserted in the map. DenseMap is a great way to map pointers to
1253pointers, or map other small types to each other.
1254</p>
1255
1256<p>
1257There are several aspects of DenseMap that you should be aware of, however. The
1258iterators in a densemap are invalidated whenever an insertion occurs, unlike
1259map. Also, because DenseMap allocates space for a large number of key/value
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1260pairs (it starts with 64 by default), it will waste a lot of space if your keys
1261or values are large. Finally, you must implement a partial specialization of
Chris Lattner76c1b972007-09-17 18:34:04 +0000[diff] [blame]1262DenseMapInfo for the key that you want, if it isn't already supported. This
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1263is required to tell DenseMap about two special marker values (which can never be
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1264inserted into the map) that it needs internally.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1265
1266</div>
1267
1268<!-- _______________________________________________________________________ -->
1269<div class="doc_subsubsection">
1270 <a name="dss_map">&lt;map&gt;</a>
1271</div>
1272
1273<div class="doc_text">
1274
1275<p>
1276std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
1277a single allocation per pair inserted into the map, it offers log(n) lookup with
1278an extremely large constant factor, imposes a space penalty of 3 pointers per
1279pair in the map, etc.</p>
1280
1281<p>std::map is most useful when your keys or values are very large, if you need
1282to iterate over the collection in sorted order, or if you need stable iterators
1283into the map (i.e. they don't get invalidated if an insertion or deletion of
1284another element takes place).</p>
1285
1286</div>
1287
1288<!-- _______________________________________________________________________ -->
1289<div class="doc_subsubsection">
1290 <a name="dss_othermap">Other Map-Like Container Options</a>
1291</div>
1292
1293<div class="doc_text">
1294
1295<p>
1296The STL provides several other options, such as std::multimap and the various
1297"hash_map" like containers (whether from C++ TR1 or from the SGI library).</p>
1298
1299<p>std::multimap is useful if you want to map a key to multiple values, but has
1300all the drawbacks of std::map. A sorted vector or some other approach is almost
1301always better.</p>
1302
1303<p>The various hash_map implementations (exposed portably by
1304"llvm/ADT/hash_map") are simple chained hash tables. This algorithm is as
1305malloc intensive as std::map (performing an allocation for each element
1306inserted, thus having really high constant factors) but (usually) provides O(1)
1307insertion/deletion of elements. This can be useful if your elements are large
1308(thus making the constant-factor cost relatively low) or if comparisons are
1309expensive. Element iteration does not visit elements in a useful order.</p>
1310
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1311</div>
1312
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1313<!-- ======================================================================= -->
1314<div class="doc_subsection">
1315 <a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
1316</div>
1317
1318<div class="doc_text">
Chris Lattner7086ce72007-09-25 22:37:50 +0000[diff] [blame]1319<p>Unlike the other containers, there are only two bit storage containers, and
1320choosing when to use each is relatively straightforward.</p>
1321
1322<p>One additional option is
1323<tt>std::vector&lt;bool&gt;</tt>: we discourage its use for two reasons 1) the
1324implementation in many common compilers (e.g. commonly available versions of
1325GCC) is extremely inefficient and 2) the C++ standards committee is likely to
1326deprecate this container and/or change it significantly somehow. In any case,
1327please don't use it.</p>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1328</div>
1329
1330<!-- _______________________________________________________________________ -->
1331<div class="doc_subsubsection">
1332 <a name="dss_bitvector">BitVector</a>
1333</div>
1334
1335<div class="doc_text">
1336<p> The BitVector container provides a fixed size set of bits for manipulation.
1337It supports individual bit setting/testing, as well as set operations. The set
1338operations take time O(size of bitvector), but operations are performed one word
1339at a time, instead of one bit at a time. This makes the BitVector very fast for
1340set operations compared to other containers. Use the BitVector when you expect
1341the number of set bits to be high (IE a dense set).
1342</p>
1343</div>
1344
1345<!-- _______________________________________________________________________ -->
1346<div class="doc_subsubsection">
1347 <a name="dss_sparsebitvector">SparseBitVector</a>
1348</div>
1349
1350<div class="doc_text">
1351<p> The SparseBitVector container is much like BitVector, with one major
1352difference: Only the bits that are set, are stored. This makes the
1353SparseBitVector much more space efficient than BitVector when the set is sparse,
1354as well as making set operations O(number of set bits) instead of O(size of
1355universe). The downside to the SparseBitVector is that setting and testing of random bits is O(N), and on large SparseBitVectors, this can be slower than BitVector. In our implementation, setting or testing bits in sorted order
1356(either forwards or reverse) is O(1) worst case. Testing and setting bits within 128 bits (depends on size) of the current bit is also O(1). As a general statement, testing/setting bits in a SparseBitVector is O(distance away from last set bit).
1357</p>
1358</div>
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]1359
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1360<!-- *********************************************************************** -->
1361<div class="doc_section">
1362 <a name="common">Helpful Hints for Common Operations</a>
1363</div>
1364<!-- *********************************************************************** -->
1365
1366<div class="doc_text">
1367
1368<p>This section describes how to perform some very simple transformations of
1369LLVM code. This is meant to give examples of common idioms used, showing the
1370practical side of LLVM transformations. <p> Because this is a "how-to" section,
1371you should also read about the main classes that you will be working with. The
1372<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
1373and descriptions of the main classes that you should know about.</p>
1374
1375</div>
1376
1377<!-- NOTE: this section should be heavy on example code -->
1378<!-- ======================================================================= -->
1379<div class="doc_subsection">
1380 <a name="inspection">Basic Inspection and Traversal Routines</a>
1381</div>
1382
1383<div class="doc_text">
1384
1385<p>The LLVM compiler infrastructure have many different data structures that may
1386be traversed. Following the example of the C++ standard template library, the
1387techniques used to traverse these various data structures are all basically the
1388same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
1389method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
1390function returns an iterator pointing to one past the last valid element of the
1391sequence, and there is some <tt>XXXiterator</tt> data type that is common
1392between the two operations.</p>
1393
1394<p>Because the pattern for iteration is common across many different aspects of
1395the program representation, the standard template library algorithms may be used
1396on them, and it is easier to remember how to iterate. First we show a few common
1397examples of the data structures that need to be traversed. Other data
1398structures are traversed in very similar ways.</p>
1399
1400</div>
1401
1402<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1403<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1404 <a name="iterate_function">Iterating over the </a><a
1405 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
1406 href="#Function"><tt>Function</tt></a>
1407</div>
1408
1409<div class="doc_text">
1410
1411<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
1412transform in some way; in particular, you'd like to manipulate its
1413<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
1414the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
1415an example that prints the name of a <tt>BasicBlock</tt> and the number of
1416<tt>Instruction</tt>s it contains:</p>
1417
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1418<div class="doc_code">
1419<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1420// <i>func is a pointer to a Function instance</i>
1421for (Function::iterator i = func-&gt;begin(), e = func-&gt;end(); i != e; ++i)
1422 // <i>Print out the name of the basic block if it has one, and then the</i>
1423 // <i>number of instructions that it contains</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1424 llvm::cerr &lt;&lt; "Basic block (name=" &lt;&lt; i-&gt;getName() &lt;&lt; ") has "
1425 &lt;&lt; i-&gt;size() &lt;&lt; " instructions.\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1426</pre>
1427</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1428
1429<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]1430invoking member functions of the <tt>Instruction</tt> class. This is
1431because the indirection operator is overloaded for the iterator
Chris Lattner7496ec52003-08-05 22:54:23 +0000[diff] [blame]1432classes. In the above code, the expression <tt>i-&gt;size()</tt> is
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1433exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
1434
1435</div>
1436
1437<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1438<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1439 <a name="iterate_basicblock">Iterating over the </a><a
1440 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1441 href="#BasicBlock"><tt>BasicBlock</tt></a>
1442</div>
1443
1444<div class="doc_text">
1445
1446<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
1447easy to iterate over the individual instructions that make up
1448<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
1449a <tt>BasicBlock</tt>:</p>
1450
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1451<div class="doc_code">
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]1452<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1453// <i>blk is a pointer to a BasicBlock instance</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1454for (BasicBlock::iterator i = blk-&gt;begin(), e = blk-&gt;end(); i != e; ++i)
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1455 // <i>The next statement works since operator&lt;&lt;(ostream&amp;,...)</i>
1456 // <i>is overloaded for Instruction&amp;</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1457 llvm::cerr &lt;&lt; *i &lt;&lt; "\n";
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]1458</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1459</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1460
1461<p>However, this isn't really the best way to print out the contents of a
1462<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
1463anything you'll care about, you could have just invoked the print routine on the
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1464basic block itself: <tt>llvm::cerr &lt;&lt; *blk &lt;&lt; "\n";</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1465
1466</div>
1467
1468<!-- _______________________________________________________________________ -->
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1469<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1470 <a name="iterate_institer">Iterating over the </a><a
1471 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1472 href="#Function"><tt>Function</tt></a>
1473</div>
1474
1475<div class="doc_text">
1476
1477<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
1478<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
1479<tt>InstIterator</tt> should be used instead. You'll need to include <a
1480href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
1481and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1482small 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]1483
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1484<div class="doc_code">
1485<pre>
1486#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
1487
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1488// <i>F is a pointer to a Function instance</i>
Chris Lattnerda021aa2008-06-04 18:20:42 +0000[diff] [blame]1489for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1490 llvm::cerr &lt;&lt; *I &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1491</pre>
1492</div>
1493
1494<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]1495work list with its initial contents. For example, if you wanted to
1496initialize a work list to contain all instructions in a <tt>Function</tt>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1497F, all you would need to do is something like:</p>
1498
1499<div class="doc_code">
1500<pre>
1501std::set&lt;Instruction*&gt; worklist;
Chris Lattnerda021aa2008-06-04 18:20:42 +0000[diff] [blame]1502// or better yet, SmallPtrSet&lt;Instruction*, 64&gt; worklist;
1503
1504for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1505 worklist.insert(&amp;*I);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1506</pre>
1507</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1508
1509<p>The STL set <tt>worklist</tt> would now contain all instructions in the
1510<tt>Function</tt> pointed to by F.</p>
1511
1512</div>
1513
1514<!-- _______________________________________________________________________ -->
1515<div class="doc_subsubsection">
1516 <a name="iterate_convert">Turning an iterator into a class pointer (and
1517 vice-versa)</a>
1518</div>
1519
1520<div class="doc_text">
1521
1522<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]1523instance when all you've got at hand is an iterator. Well, extracting
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1524a reference or a pointer from an iterator is very straight-forward.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1525Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1526is a <tt>BasicBlock::const_iterator</tt>:</p>
1527
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1528<div class="doc_code">
1529<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1530Instruction&amp; inst = *i; // <i>Grab reference to instruction reference</i>
1531Instruction* pinst = &amp;*i; // <i>Grab pointer to instruction reference</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1532const Instruction&amp; inst = *j;
1533</pre>
1534</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1535
1536<p>However, the iterators you'll be working with in the LLVM framework are
1537special: they will automatically convert to a ptr-to-instance type whenever they
1538need to. Instead of dereferencing the iterator and then taking the address of
1539the result, you can simply assign the iterator to the proper pointer type and
1540you get the dereference and address-of operation as a result of the assignment
1541(behind the scenes, this is a result of overloading casting mechanisms). Thus
1542the last line of the last example,</p>
1543
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1544<div class="doc_code">
1545<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1546Instruction *pinst = &amp;*i;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1547</pre>
1548</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1549
1550<p>is semantically equivalent to</p>
1551
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1552<div class="doc_code">
1553<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1554Instruction *pinst = i;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1555</pre>
1556</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1557
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1558<p>It's also possible to turn a class pointer into the corresponding iterator,
1559and this is a constant time operation (very efficient). The following code
1560snippet illustrates use of the conversion constructors provided by LLVM
1561iterators. By using these, you can explicitly grab the iterator of something
1562without actually obtaining it via iteration over some structure:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1563
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1564<div class="doc_code">
1565<pre>
1566void printNextInstruction(Instruction* inst) {
1567 BasicBlock::iterator it(inst);
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1568 ++it; // <i>After this line, it refers to the instruction after *inst</i>
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1569 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]1570}
1571</pre>
1572</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1573
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1574</div>
1575
1576<!--_______________________________________________________________________-->
1577<div class="doc_subsubsection">
1578 <a name="iterate_complex">Finding call sites: a slightly more complex
1579 example</a>
1580</div>
1581
1582<div class="doc_text">
1583
1584<p>Say that you're writing a FunctionPass and would like to count all the
1585locations in the entire module (that is, across every <tt>Function</tt>) where a
1586certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
1587learn 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]1588much more straight-forward manner, but this example will allow us to explore how
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1589you'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]1590is what we want to do:</p>
1591
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1592<div class="doc_code">
1593<pre>
1594initialize callCounter to zero
1595for each Function f in the Module
1596 for each BasicBlock b in f
1597 for each Instruction i in b
1598 if (i is a CallInst and calls the given function)
1599 increment callCounter
1600</pre>
1601</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1602
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1603<p>And the actual code is (remember, because we're writing a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1604<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1605override the <tt>runOnFunction</tt> method):</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1606
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1607<div class="doc_code">
1608<pre>
1609Function* targetFunc = ...;
1610
1611class OurFunctionPass : public FunctionPass {
1612 public:
1613 OurFunctionPass(): callCounter(0) { }
1614
1615 virtual runOnFunction(Function&amp; F) {
1616 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
1617 for (BasicBlock::iterator i = b-&gt;begin(); ie = b-&gt;end(); i != ie; ++i) {
1618 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a>&lt;<a
1619 href="#CallInst">CallInst</a>&gt;(&amp;*i)) {
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1620 // <i>We know we've encountered a call instruction, so we</i>
1621 // <i>need to determine if it's a call to the</i>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1622 // <i>function pointed to by m_func or not.</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1623 if (callInst-&gt;getCalledFunction() == targetFunc)
1624 ++callCounter;
1625 }
1626 }
1627 }
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1628 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1629
1630 private:
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1631 unsigned callCounter;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1632};
1633</pre>
1634</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1635
1636</div>
1637
Brian Gaekef1972c62003-11-07 19:25:45 +0000[diff] [blame]1638<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1639<div class="doc_subsubsection">
1640 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
1641</div>
1642
1643<div class="doc_text">
1644
1645<p>You may have noticed that the previous example was a bit oversimplified in
1646that it did not deal with call sites generated by 'invoke' instructions. In
1647this, and in other situations, you may find that you want to treat
1648<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
1649most-specific common base class is <tt>Instruction</tt>, which includes lots of
1650less closely-related things. For these cases, LLVM provides a handy wrapper
1651class called <a
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]1652href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1653It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
1654methods that provide functionality common to <tt>CallInst</tt>s and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1655<tt>InvokeInst</tt>s.</p>
1656
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1657<p>This class has "value semantics": it should be passed by value, not by
1658reference and it should not be dynamically allocated or deallocated using
1659<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
1660assignable and constructable, with costs equivalents to that of a bare pointer.
1661If you look at its definition, it has only a single pointer member.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1662
1663</div>
1664
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]1665<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1666<div class="doc_subsubsection">
1667 <a name="iterate_chains">Iterating over def-use &amp; use-def chains</a>
1668</div>
1669
1670<div class="doc_text">
1671
1672<p>Frequently, we might have an instance of the <a
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]1673href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1674determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
1675<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
1676For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
1677particular function <tt>foo</tt>. Finding all of the instructions that
1678<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
1679of <tt>F</tt>:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1680
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1681<div class="doc_code">
1682<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1683Function *F = ...;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1684
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1685for (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]1686 if (Instruction *Inst = dyn_cast&lt;Instruction&gt;(*i)) {
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]1687 llvm::cerr &lt;&lt; "F is used in instruction:\n";
1688 llvm::cerr &lt;&lt; *Inst &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1689 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1690</pre>
1691</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1692
1693<p>Alternately, it's common to have an instance of the <a
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1694href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1695<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
1696<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
1697<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
1698all of the values that a particular instruction uses (that is, the operands of
1699the particular <tt>Instruction</tt>):</p>
1700
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1701<div class="doc_code">
1702<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1703Instruction *pi = ...;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1704
1705for (User::op_iterator i = pi-&gt;op_begin(), e = pi-&gt;op_end(); i != e; ++i) {
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1706 Value *v = *i;
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1707 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1708}
1709</pre>
1710</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1711
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]1712<!--
1713 def-use chains ("finding all users of"): Value::use_begin/use_end
1714 use-def chains ("finding all values used"): User::op_begin/op_end [op=operand]
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1715-->
1716
1717</div>
1718
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]1719<!--_______________________________________________________________________-->
1720<div class="doc_subsubsection">
1721 <a name="iterate_preds">Iterating over predecessors &amp;
1722successors of blocks</a>
1723</div>
1724
1725<div class="doc_text">
1726
1727<p>Iterating over the predecessors and successors of a block is quite easy
1728with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
1729this to iterate over all predecessors of BB:</p>
1730
1731<div class="doc_code">
1732<pre>
1733#include "llvm/Support/CFG.h"
1734BasicBlock *BB = ...;
1735
1736for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
1737 BasicBlock *Pred = *PI;
1738 // <i>...</i>
1739}
1740</pre>
1741</div>
1742
1743<p>Similarly, to iterate over successors use
1744succ_iterator/succ_begin/succ_end.</p>
1745
1746</div>
1747
1748
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1749<!-- ======================================================================= -->
1750<div class="doc_subsection">
1751 <a name="simplechanges">Making simple changes</a>
1752</div>
1753
1754<div class="doc_text">
1755
1756<p>There are some primitive transformation operations present in the LLVM
Joel Stanley753eb712002-09-11 22:32:24 +0000[diff] [blame]1757infrastructure that are worth knowing about. When performing
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1758transformations, it's fairly common to manipulate the contents of basic
1759blocks. This section describes some of the common methods for doing so
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1760and gives example code.</p>
1761
1762</div>
1763
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1764<!--_______________________________________________________________________-->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1765<div class="doc_subsubsection">
1766 <a name="schanges_creating">Creating and inserting new
1767 <tt>Instruction</tt>s</a>
1768</div>
1769
1770<div class="doc_text">
1771
1772<p><i>Instantiating Instructions</i></p>
1773
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1774<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1775constructor for the kind of instruction to instantiate and provide the necessary
1776parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
1777(const-ptr-to) <tt>Type</tt>. Thus:</p>
1778
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1779<div class="doc_code">
1780<pre>
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1781AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1782</pre>
1783</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1784
1785<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1786one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1787subclass is likely to have varying default parameters which change the semantics
1788of the instruction, so refer to the <a
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]1789href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1790Instruction</a> that you're interested in instantiating.</p>
1791
1792<p><i>Naming values</i></p>
1793
1794<p>It is very useful to name the values of instructions when you're able to, as
1795this facilitates the debugging of your transformations. If you end up looking
1796at generated LLVM machine code, you definitely want to have logical names
1797associated with the results of instructions! By supplying a value for the
1798<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
1799associate a logical name with the result of the instruction's execution at
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1800run time. For example, say that I'm writing a transformation that dynamically
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1801allocates space for an integer on the stack, and that integer is going to be
1802used as some kind of index by some other code. To accomplish this, I place an
1803<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
1804<tt>Function</tt>, and I'm intending to use it within the same
1805<tt>Function</tt>. I might do:</p>
1806
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1807<div class="doc_code">
1808<pre>
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1809AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1810</pre>
1811</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1812
1813<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1814execution value, which is a pointer to an integer on the run time stack.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1815
1816<p><i>Inserting instructions</i></p>
1817
1818<p>There are essentially two ways to insert an <tt>Instruction</tt>
1819into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
1820
Joel Stanley9dd1ad62002-09-18 03:17:23 +0000[diff] [blame]1821<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1822 <li>Insertion into an explicit instruction list
1823
1824 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
1825 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
1826 before <tt>*pi</tt>, we do the following: </p>
1827
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1828<div class="doc_code">
1829<pre>
1830BasicBlock *pb = ...;
1831Instruction *pi = ...;
1832Instruction *newInst = new Instruction(...);
1833
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1834pb-&gt;getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1835</pre>
1836</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1837
1838 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
1839 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
1840 classes provide constructors which take a pointer to a
1841 <tt>BasicBlock</tt> to be appended to. For example code that
1842 looked like: </p>
1843
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1844<div class="doc_code">
1845<pre>
1846BasicBlock *pb = ...;
1847Instruction *newInst = new Instruction(...);
1848
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]1849pb-&gt;getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1850</pre>
1851</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1852
1853 <p>becomes: </p>
1854
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1855<div class="doc_code">
1856<pre>
1857BasicBlock *pb = ...;
1858Instruction *newInst = new Instruction(..., pb);
1859</pre>
1860</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]1861
1862 <p>which is much cleaner, especially if you are creating
1863 long instruction streams.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1864
1865 <li>Insertion into an implicit instruction list
1866
1867 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
1868 are implicitly associated with an existing instruction list: the instruction
1869 list of the enclosing basic block. Thus, we could have accomplished the same
1870 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
1871 </p>
1872
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1873<div class="doc_code">
1874<pre>
1875Instruction *pi = ...;
1876Instruction *newInst = new Instruction(...);
1877
1878pi-&gt;getParent()-&gt;getInstList().insert(pi, newInst);
1879</pre>
1880</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1881
1882 <p>In fact, this sequence of steps occurs so frequently that the
1883 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
1884 constructors which take (as a default parameter) a pointer to an
1885 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
1886 precede. That is, <tt>Instruction</tt> constructors are capable of
1887 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
1888 provided instruction, immediately before that instruction. Using an
1889 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
1890 parameter, the above code becomes:</p>
1891
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1892<div class="doc_code">
1893<pre>
1894Instruction* pi = ...;
1895Instruction* newInst = new Instruction(..., pi);
1896</pre>
1897</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1898
1899 <p>which is much cleaner, especially if you're creating a lot of
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1900 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1901</ul>
1902
1903</div>
1904
1905<!--_______________________________________________________________________-->
1906<div class="doc_subsubsection">
1907 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
1908</div>
1909
1910<div class="doc_text">
1911
1912<p>Deleting an instruction from an existing sequence of instructions that form a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]1913<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward. First,
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1914you must have a pointer to the instruction that you wish to delete. Second, you
1915need to obtain the pointer to that instruction's basic block. You use the
1916pointer to the basic block to get its list of instructions and then use the
1917erase function to remove your instruction. For example:</p>
1918
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1919<div class="doc_code">
1920<pre>
1921<a href="#Instruction">Instruction</a> *I = .. ;
Chris Lattner9f8ec252008-02-15 22:57:17 +0000[diff] [blame]1922I-&gt;eraseFromParent();
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1923</pre>
1924</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1925
1926</div>
1927
1928<!--_______________________________________________________________________-->
1929<div class="doc_subsubsection">
1930 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
1931 <tt>Value</tt></a>
1932</div>
1933
1934<div class="doc_text">
1935
1936<p><i>Replacing individual instructions</i></p>
1937
1938<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]1939permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1940and <tt>ReplaceInstWithInst</tt>.</p>
1941
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1942<h4><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1943
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1944<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1945 <li><tt>ReplaceInstWithValue</tt>
1946
1947 <p>This function replaces all uses (within a basic block) of a given
1948 instruction with a value, and then removes the original instruction. The
1949 following example illustrates the replacement of the result of a particular
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]1950 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1951 pointer to an integer.</p>
1952
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1953<div class="doc_code">
1954<pre>
1955AllocaInst* instToReplace = ...;
1956BasicBlock::iterator ii(instToReplace);
1957
1958ReplaceInstWithValue(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1959 Constant::getNullValue(PointerType::get(Type::Int32Ty)));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1960</pre></div></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1961
1962 <li><tt>ReplaceInstWithInst</tt>
1963
1964 <p>This function replaces a particular instruction with another
1965 instruction. The following example illustrates the replacement of one
1966 <tt>AllocaInst</tt> with another.</p>
1967
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1968<div class="doc_code">
1969<pre>
1970AllocaInst* instToReplace = ...;
1971BasicBlock::iterator ii(instToReplace);
1972
1973ReplaceInstWithInst(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]1974 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]1975</pre></div></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1976</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1977
1978<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
1979
1980<p>You can use <tt>Value::replaceAllUsesWith</tt> and
1981<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]1982doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]1983and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]1984information.</p>
1985
1986<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
1987include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
1988ReplaceInstWithValue, ReplaceInstWithInst -->
1989
1990</div>
1991
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]1992<!--_______________________________________________________________________-->
1993<div class="doc_subsubsection">
1994 <a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
1995</div>
1996
1997<div class="doc_text">
1998
Tanya Lattnerc5dfcdb2007-06-20 20:46:37 +0000[diff] [blame]1999<p>Deleting a global variable from a module is just as easy as deleting an
2000Instruction. First, you must have a pointer to the global variable that you wish
2001 to delete. You use this pointer to erase it from its parent, the module.
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2002 For example:</p>
2003
2004<div class="doc_code">
2005<pre>
2006<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2007
Tanya Lattnerc5dfcdb2007-06-20 20:46:37 +0000[diff] [blame]2008GV-&gt;eraseFromParent();
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2009</pre>
2010</div>
2011
2012</div>
2013
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2014<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2015<div class="doc_section">
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2016 <a name="advanced">Advanced Topics</a>
2017</div>
2018<!-- *********************************************************************** -->
2019
2020<div class="doc_text">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2021<p>
2022This section describes some of the advanced or obscure API's that most clients
2023do not need to be aware of. These API's tend manage the inner workings of the
2024LLVM system, and only need to be accessed in unusual circumstances.
2025</p>
2026</div>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2027
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2028<!-- ======================================================================= -->
2029<div class="doc_subsection">
2030 <a name="TypeResolve">LLVM Type Resolution</a>
2031</div>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2032
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2033<div class="doc_text">
2034
2035<p>
2036The LLVM type system has a very simple goal: allow clients to compare types for
2037structural equality with a simple pointer comparison (aka a shallow compare).
2038This goal makes clients much simpler and faster, and is used throughout the LLVM
2039system.
2040</p>
2041
2042<p>
2043Unfortunately achieving this goal is not a simple matter. In particular,
2044recursive types and late resolution of opaque types makes the situation very
2045difficult to handle. Fortunately, for the most part, our implementation makes
2046most clients able to be completely unaware of the nasty internal details. The
2047primary case where clients are exposed to the inner workings of it are when
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]2048building a recursive type. In addition to this case, the LLVM bitcode reader,
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2049assembly parser, and linker also have to be aware of the inner workings of this
2050system.
2051</p>
2052
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]2053<p>
2054For our purposes below, we need three concepts. First, an "Opaque Type" is
2055exactly as defined in the <a href="LangRef.html#t_opaque">language
2056reference</a>. Second an "Abstract Type" is any type which includes an
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2057opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
2058Third, 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]2059float }</tt>").
2060</p>
2061
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2062</div>
2063
2064<!-- ______________________________________________________________________ -->
2065<div class="doc_subsubsection">
2066 <a name="BuildRecType">Basic Recursive Type Construction</a>
2067</div>
2068
2069<div class="doc_text">
2070
2071<p>
2072Because the most common question is "how do I build a recursive type with LLVM",
2073we answer it now and explain it as we go. Here we include enough to cause this
2074to be emitted to an output .ll file:
2075</p>
2076
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2077<div class="doc_code">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2078<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2079%mylist = type { %mylist*, i32 }
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2080</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2081</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2082
2083<p>
2084To build this, use the following LLVM APIs:
2085</p>
2086
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2087<div class="doc_code">
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2088<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2089// <i>Create the initial outer struct</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2090<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
2091std::vector&lt;const Type*&gt; Elts;
2092Elts.push_back(PointerType::get(StructTy));
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]2093Elts.push_back(Type::Int32Ty);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2094StructType *NewSTy = StructType::get(Elts);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2095
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2096// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2097// <i>the struct and the opaque type are actually the same.</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2098cast&lt;OpaqueType&gt;(StructTy.get())-&gt;<a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2099
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2100// <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]2101// <i>kept up-to-date</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2102NewSTy = cast&lt;StructType&gt;(StructTy.get());
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2103
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2104// <i>Add a name for the type to the module symbol table (optional)</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2105MyModule-&gt;addTypeName("mylist", NewSTy);
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2106</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2107</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2108
2109<p>
2110This code shows the basic approach used to build recursive types: build a
2111non-recursive type using 'opaque', then use type unification to close the cycle.
2112The type unification step is performed by the <tt><a
Chris Lattneraff26d12007-02-03 03:06:52 +0000[diff] [blame]2113href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2114described next. After that, we describe the <a
2115href="#PATypeHolder">PATypeHolder class</a>.
2116</p>
2117
2118</div>
2119
2120<!-- ______________________________________________________________________ -->
2121<div class="doc_subsubsection">
2122 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
2123</div>
2124
2125<div class="doc_text">
2126<p>
2127The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
2128While this method is actually a member of the DerivedType class, it is most
2129often used on OpaqueType instances. Type unification is actually a recursive
2130process. After unification, types can become structurally isomorphic to
2131existing types, and all duplicates are deleted (to preserve pointer equality).
2132</p>
2133
2134<p>
2135In the example above, the OpaqueType object is definitely deleted.
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2136Additionally, if there is an "{ \2*, i32}" type already created in the system,
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2137the pointer and struct type created are <b>also</b> deleted. Obviously whenever
2138a type is deleted, any "Type*" pointers in the program are invalidated. As
2139such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
2140live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
2141types can never move or be deleted). To deal with this, the <a
2142href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
2143reference to a possibly refined type, and the <a
2144href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
2145complex datastructures.
2146</p>
2147
2148</div>
2149
2150<!-- ______________________________________________________________________ -->
2151<div class="doc_subsubsection">
2152 <a name="PATypeHolder">The PATypeHolder Class</a>
2153</div>
2154
2155<div class="doc_text">
2156<p>
2157PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
2158happily goes about nuking types that become isomorphic to existing types, it
2159automatically updates all PATypeHolder objects to point to the new type. In the
2160example above, this allows the code to maintain a pointer to the resultant
2161resolved recursive type, even though the Type*'s are potentially invalidated.
2162</p>
2163
2164<p>
2165PATypeHolder is an extremely light-weight object that uses a lazy union-find
2166implementation to update pointers. For example the pointer from a Value to its
2167Type is maintained by PATypeHolder objects.
2168</p>
2169
2170</div>
2171
2172<!-- ______________________________________________________________________ -->
2173<div class="doc_subsubsection">
2174 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
2175</div>
2176
2177<div class="doc_text">
2178
2179<p>
2180Some data structures need more to perform more complex updates when types get
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2181resolved. To support this, a class can derive from the AbstractTypeUser class.
2182This class
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2183allows it to get callbacks when certain types are resolved. To register to get
2184callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
Chris Lattner0f876db2005-04-25 15:47:57 +0000[diff] [blame]2185methods can be called on a type. Note that these methods only work for <i>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2186 abstract</i> types. Concrete types (those that do not include any opaque
2187objects) can never be refined.
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2188</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2189</div>
2190
2191
2192<!-- ======================================================================= -->
2193<div class="doc_subsection">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2194 <a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
2195 <tt>TypeSymbolTable</tt> classes</a>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2196</div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2197
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2198<div class="doc_text">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2199<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
2200ValueSymbolTable</a></tt> class provides a symbol table that the <a
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2201href="#Function"><tt>Function</tt></a> and <a href="#Module">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2202<tt>Module</tt></a> classes use for naming value definitions. The symbol table
2203can provide a name for any <a href="#Value"><tt>Value</tt></a>.
2204The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
2205TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
2206names for types.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2207
Reid Spencera6362242007-01-07 00:41:39 +0000[diff] [blame]2208<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
2209by most clients. It should only be used when iteration over the symbol table
2210names themselves are required, which is very special purpose. Note that not
2211all LLVM
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2212<tt><a href="#Value">Value</a></tt>s have names, and those without names (i.e. they have
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2213an empty name) do not exist in the symbol table.
2214</p>
2215
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2216<p>These symbol tables support iteration over the values/types in the symbol
2217table with <tt>begin/end/iterator</tt> and supports querying to see if a
2218specific name is in the symbol table (with <tt>lookup</tt>). The
2219<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
2220simply call <tt>setName</tt> on a value, which will autoinsert it into the
2221appropriate symbol table. For types, use the Module::addTypeName method to
2222insert entries into the symbol table.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2223
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2224</div>
2225
2226
2227
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2228<!-- ======================================================================= -->
2229<div class="doc_subsection">
2230 <a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
2231</div>
2232
2233<div class="doc_text">
2234<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1User.html">
2235User</a></tt> class provides a base for expressing the ownership of <tt>User</tt>
2236towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
2237Value</a></tt>s. The <tt><a href="http://llvm.org/doxygen/classllvm_1_1Use.html">
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2238Use</a></tt> helper class is employed to do the bookkeeping and to facilitate <i>O(1)</i>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2239addition and removal.</p>
2240
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2241<!-- ______________________________________________________________________ -->
2242<div class="doc_subsubsection">
2243 <a name="PATypeHolder">Interaction and relationship between <tt>User</tt> and <tt>Use</tt> objects</a>
2244</div>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2245
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2246<div class="doc_text">
2247<p>
2248A subclass of <tt>User</tt> can choose between incorporating its <tt>Use</tt> objects
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2249or refer to them out-of-line by means of a pointer. A mixed variant
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2250(some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
2251that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
2252</p>
2253</div>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2254
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2255<p>
2256We have 2 different layouts in the <tt>User</tt> (sub)classes:
2257<ul>
2258<li><p>Layout a)
2259The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
2260object and there are a fixed number of them.</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2261
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2262<li><p>Layout b)
2263The <tt>Use</tt> object(s) are referenced by a pointer to an
2264array from the <tt>User</tt> object and there may be a variable
2265number of them.</p>
2266</ul>
2267<p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2268Initially each layout will possess a direct pointer to the
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2269start of the array of <tt>Use</tt>s. Though not mandatory for layout a),
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2270we stick to this redundancy for the sake of simplicity.
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2271The <tt>User</tt> object will also store the number of <tt>Use</tt> objects it
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2272has. (Theoretically this information can also be calculated
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2273given the scheme presented below.)</p>
2274<p>
2275Special forms of allocation operators (<tt>operator new</tt>)
2276will enforce the following memory layouts:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2277
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2278<ul>
2279<li><p>Layout a) will be modelled by prepending the <tt>User</tt> object by the <tt>Use[]</tt> array.</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2280
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2281<pre>
2282...---.---.---.---.-------...
2283 | P | P | P | P | User
2284'''---'---'---'---'-------'''
2285</pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2286
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2287<li><p>Layout b) will be modelled by pointing at the Use[] array.</p>
2288<pre>
2289.-------...
2290| User
2291'-------'''
2292 |
2293 v
2294 .---.---.---.---...
2295 | P | P | P | P |
2296 '---'---'---'---'''
2297</pre>
2298</ul>
2299<i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
2300 is stored in each <tt>Use</tt> object in the member <tt>Use::Prev</tt>)</i>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2301
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2302<!-- ______________________________________________________________________ -->
2303<div class="doc_subsubsection">
2304 <a name="PATypeHolder">The waymarking algorithm</a>
2305</div>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2306
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2307<div class="doc_text">
2308<p>
2309Since the <tt>Use</tt> objects will be deprived of the direct pointer to
2310their <tt>User</tt> objects, there must be a fast and exact method to
2311recover it. This is accomplished by the following scheme:</p>
2312</div>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2313
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2314A bit-encoding in the 2 LSBits (least significant bits) of the <tt>Use::Prev</tt> will allow to find the
2315start of the <tt>User</tt> object:
2316<ul>
2317<li><tt>00</tt> &mdash;&gt; binary digit 0</li>
2318<li><tt>01</tt> &mdash;&gt; binary digit 1</li>
2319<li><tt>10</tt> &mdash;&gt; stop and calculate (<tt>s</tt>)</li>
2320<li><tt>11</tt> &mdash;&gt; full stop (<tt>S</tt>)</li>
2321</ul>
2322<p>
2323Given a <tt>Use*</tt>, all we have to do is to walk till we get
2324a stop and we either have a <tt>User</tt> immediately behind or
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2325we have to walk to the next stop picking up digits
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2326and calculating the offset:</p>
2327<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2328.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
2329| 1 | s | 1 | 0 | 1 | 0 | s | 1 | 1 | 0 | s | 1 | 1 | s | 1 | S | User (or User*)
2330'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
2331 |+15 |+10 |+6 |+3 |+1
2332 | | | | |__>
2333 | | | |__________>
2334 | | |______________________>
2335 | |______________________________________>
2336 |__________________________________________________________>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2337</pre>
2338<p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2339Only the significant number of bits need to be stored between the
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2340stops, so that the <i>worst case is 20 memory accesses</i> when there are
23411000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2342
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2343<!-- ______________________________________________________________________ -->
2344<div class="doc_subsubsection">
2345 <a name="PATypeHolder">Reference implementation</a>
2346</div>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2347
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2348<div class="doc_text">
2349<p>
2350The following literate Haskell fragment demonstrates the concept:</p>
2351</div>
2352
2353<div class="doc_code">
2354<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2355> import Test.QuickCheck
2356>
2357> digits :: Int -> [Char] -> [Char]
2358> digits 0 acc = '0' : acc
2359> digits 1 acc = '1' : acc
2360> digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
2361>
2362> dist :: Int -> [Char] -> [Char]
2363> dist 0 [] = ['S']
2364> dist 0 acc = acc
2365> dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
2366> dist n acc = dist (n - 1) $ dist 1 acc
2367>
2368> takeLast n ss = reverse $ take n $ reverse ss
2369>
2370> test = takeLast 40 $ dist 20 []
2371>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2372</pre>
2373</div>
2374<p>
2375Printing &lt;test&gt; gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
2376<p>
2377The reverse algorithm computes the length of the string just by examining
2378a certain prefix:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2379
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2380<div class="doc_code">
2381<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2382> pref :: [Char] -> Int
2383> pref "S" = 1
2384> pref ('s':'1':rest) = decode 2 1 rest
2385> pref (_:rest) = 1 + pref rest
2386>
2387> decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
2388> decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
2389> decode walk acc _ = walk + acc
2390>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2391</pre>
2392</div>
2393<p>
2394Now, as expected, printing &lt;pref test&gt; gives <tt>40</tt>.</p>
2395<p>
2396We can <i>quickCheck</i> this with following property:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2397
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2398<div class="doc_code">
2399<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2400> testcase = dist 2000 []
2401> testcaseLength = length testcase
2402>
2403> identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
2404> where arr = takeLast n testcase
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2405>
2406</pre>
2407</div>
2408<p>
2409As expected &lt;quickCheck identityProp&gt; gives:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2410
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2411<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2412*Main> quickCheck identityProp
2413OK, passed 100 tests.
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2414</pre>
2415<p>
2416Let's be a bit more exhaustive:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2417
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2418<div class="doc_code">
2419<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2420>
2421> deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
2422>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2423</pre>
2424</div>
2425<p>
2426And here is the result of &lt;deepCheck identityProp&gt;:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2427
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2428<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2429*Main> deepCheck identityProp
2430OK, passed 500 tests.
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2431</pre>
2432
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]2433<!-- ______________________________________________________________________ -->
2434<div class="doc_subsubsection">
2435 <a name="PATypeHolder">Tagging considerations</a>
2436</div>
2437
2438<p>
2439To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
2440never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
2441new <tt>Use**</tt> on every modification. Accordingly getters must strip the
2442tag bits.</p>
2443<p>
2444For layout b) instead of the <tt>User</tt> we will find a pointer (<tt>User*</tt> with LSBit set).
2445Following this pointer brings us to the <tt>User</tt>. A portable trick will ensure
2446that the first bytes of <tt>User</tt> (if interpreted as a pointer) will never have
2447the LSBit set.</p>
2448
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]2449</div>
2450
2451 <!-- *********************************************************************** -->
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2452<div class="doc_section">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2453 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
2454</div>
2455<!-- *********************************************************************** -->
2456
2457<div class="doc_text">
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2458<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
2459<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2460
2461<p>The Core LLVM classes are the primary means of representing the program
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2462being inspected or transformed. The core LLVM classes are defined in
2463header files in the <tt>include/llvm/</tt> directory, and implemented in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2464the <tt>lib/VMCore</tt> directory.</p>
2465
2466</div>
2467
2468<!-- ======================================================================= -->
2469<div class="doc_subsection">
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2470 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
2471</div>
2472
2473<div class="doc_text">
2474
2475 <p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
2476 a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
2477 through its subclasses. Certain primitive types (<tt>VoidType</tt>,
2478 <tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
2479 subclasses. They are hidden because they offer no useful functionality beyond
2480 what the <tt>Type</tt> class offers except to distinguish themselves from
2481 other subclasses of <tt>Type</tt>.</p>
2482 <p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
2483 named, but this is not a requirement. There exists exactly
2484 one instance of a given shape at any one time. This allows type equality to
2485 be performed with address equality of the Type Instance. That is, given two
2486 <tt>Type*</tt> values, the types are identical if the pointers are identical.
2487 </p>
2488</div>
2489
2490<!-- _______________________________________________________________________ -->
2491<div class="doc_subsubsection">
2492 <a name="m_Value">Important Public Methods</a>
2493</div>
2494
2495<div class="doc_text">
2496
2497<ul>
Chris Lattner8f79df32007-01-15 01:55:32 +0000[diff] [blame]2498 <li><tt>bool isInteger() const</tt>: Returns true for any integer type.</li>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2499
2500 <li><tt>bool isFloatingPoint()</tt>: Return true if this is one of the two
2501 floating point types.</li>
2502
2503 <li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
2504 an OpaqueType anywhere in its definition).</li>
2505
2506 <li><tt>bool isSized()</tt>: Return true if the type has known size. Things
2507 that don't have a size are abstract types, labels and void.</li>
2508
2509</ul>
2510</div>
2511
2512<!-- _______________________________________________________________________ -->
2513<div class="doc_subsubsection">
2514 <a name="m_Value">Important Derived Types</a>
2515</div>
2516<div class="doc_text">
2517<dl>
2518 <dt><tt>IntegerType</tt></dt>
2519 <dd>Subclass of DerivedType that represents integer types of any bit width.
2520 Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
2521 <tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
2522 <ul>
2523 <li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
2524 type of a specific bit width.</li>
2525 <li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
2526 type.</li>
2527 </ul>
2528 </dd>
2529 <dt><tt>SequentialType</tt></dt>
2530 <dd>This is subclassed by ArrayType and PointerType
2531 <ul>
2532 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
2533 of the elements in the sequential type. </li>
2534 </ul>
2535 </dd>
2536 <dt><tt>ArrayType</tt></dt>
2537 <dd>This is a subclass of SequentialType and defines the interface for array
2538 types.
2539 <ul>
2540 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
2541 elements in the array. </li>
2542 </ul>
2543 </dd>
2544 <dt><tt>PointerType</tt></dt>
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]2545 <dd>Subclass of SequentialType for pointer types.</dd>
Reid Spencer9d6565a2007-02-15 02:26:10 +0000[diff] [blame]2546 <dt><tt>VectorType</tt></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2547 <dd>Subclass of SequentialType for vector types. A
2548 vector type is similar to an ArrayType but is distinguished because it is
2549 a first class type wherease ArrayType is not. Vector types are used for
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2550 vector operations and are usually small vectors of of an integer or floating
2551 point type.</dd>
2552 <dt><tt>StructType</tt></dt>
2553 <dd>Subclass of DerivedTypes for struct types.</dd>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2554 <dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2555 <dd>Subclass of DerivedTypes for function types.
2556 <ul>
2557 <li><tt>bool isVarArg() const</tt>: Returns true if its a vararg
2558 function</li>
2559 <li><tt> const Type * getReturnType() const</tt>: Returns the
2560 return type of the function.</li>
2561 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
2562 the type of the ith parameter.</li>
2563 <li><tt> const unsigned getNumParams() const</tt>: Returns the
2564 number of formal parameters.</li>
2565 </ul>
2566 </dd>
2567 <dt><tt>OpaqueType</tt></dt>
2568 <dd>Sublcass of DerivedType for abstract types. This class
2569 defines no content and is used as a placeholder for some other type. Note
2570 that OpaqueType is used (temporarily) during type resolution for forward
2571 references of types. Once the referenced type is resolved, the OpaqueType
2572 is replaced with the actual type. OpaqueType can also be used for data
2573 abstraction. At link time opaque types can be resolved to actual types
2574 of the same name.</dd>
2575</dl>
2576</div>
2577
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2578
2579
2580<!-- ======================================================================= -->
2581<div class="doc_subsection">
2582 <a name="Module">The <tt>Module</tt> class</a>
2583</div>
2584
2585<div class="doc_text">
2586
2587<p><tt>#include "<a
2588href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
2589<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
2590
2591<p>The <tt>Module</tt> class represents the top level structure present in LLVM
2592programs. An LLVM module is effectively either a translation unit of the
2593original program or a combination of several translation units merged by the
2594linker. The <tt>Module</tt> class keeps track of a list of <a
2595href="#Function"><tt>Function</tt></a>s, a list of <a
2596href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
2597href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
2598helpful member functions that try to make common operations easy.</p>
2599
2600</div>
2601
2602<!-- _______________________________________________________________________ -->
2603<div class="doc_subsubsection">
2604 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
2605</div>
2606
2607<div class="doc_text">
2608
2609<ul>
2610 <li><tt>Module::Module(std::string name = "")</tt></li>
2611</ul>
2612
2613<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
2614provide a name for it (probably based on the name of the translation unit).</p>
2615
2616<ul>
2617 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
2618 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
2619
2620 <tt>begin()</tt>, <tt>end()</tt>
2621 <tt>size()</tt>, <tt>empty()</tt>
2622
2623 <p>These are forwarding methods that make it easy to access the contents of
2624 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
2625 list.</p></li>
2626
2627 <li><tt>Module::FunctionListType &amp;getFunctionList()</tt>
2628
2629 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
2630 necessary to use when you need to update the list or perform a complex
2631 action that doesn't have a forwarding method.</p>
2632
2633 <p><!-- Global Variable --></p></li>
2634</ul>
2635
2636<hr>
2637
2638<ul>
2639 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
2640
2641 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
2642
2643 <tt>global_begin()</tt>, <tt>global_end()</tt>
2644 <tt>global_size()</tt>, <tt>global_empty()</tt>
2645
2646 <p> These are forwarding methods that make it easy to access the contents of
2647 a <tt>Module</tt> object's <a
2648 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
2649
2650 <li><tt>Module::GlobalListType &amp;getGlobalList()</tt>
2651
2652 <p>Returns the list of <a
2653 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
2654 use when you need to update the list or perform a complex action that
2655 doesn't have a forwarding method.</p>
2656
2657 <p><!-- Symbol table stuff --> </p></li>
2658</ul>
2659
2660<hr>
2661
2662<ul>
2663 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
2664
2665 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2666 for this <tt>Module</tt>.</p>
2667
2668 <p><!-- Convenience methods --></p></li>
2669</ul>
2670
2671<hr>
2672
2673<ul>
2674 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
2675 &amp;Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
2676
2677 <p>Look up the specified function in the <tt>Module</tt> <a
2678 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
2679 <tt>null</tt>.</p></li>
2680
2681 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
2682 std::string &amp;Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
2683
2684 <p>Look up the specified function in the <tt>Module</tt> <a
2685 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
2686 external declaration for the function and return it.</p></li>
2687
2688 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
2689
2690 <p>If there is at least one entry in the <a
2691 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
2692 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
2693 string.</p></li>
2694
2695 <li><tt>bool addTypeName(const std::string &amp;Name, const <a
2696 href="#Type">Type</a> *Ty)</tt>
2697
2698 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
2699 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
2700 name, true is returned and the <a
2701 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
2702</ul>
2703
2704</div>
2705
2706
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]2707<!-- ======================================================================= -->
2708<div class="doc_subsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2709 <a name="Value">The <tt>Value</tt> class</a>
2710</div>
2711
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2712<div class="doc_text">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2713
2714<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
2715<br>
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]2716doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2717
2718<p>The <tt>Value</tt> class is the most important class in the LLVM Source
2719base. It represents a typed value that may be used (among other things) as an
2720operand to an instruction. There are many different types of <tt>Value</tt>s,
2721such as <a href="#Constant"><tt>Constant</tt></a>s,<a
2722href="#Argument"><tt>Argument</tt></a>s. Even <a
2723href="#Instruction"><tt>Instruction</tt></a>s and <a
2724href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
2725
2726<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
2727for a program. For example, an incoming argument to a function (represented
2728with an instance of the <a href="#Argument">Argument</a> class) is "used" by
2729every instruction in the function that references the argument. To keep track
2730of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
2731href="#User"><tt>User</tt></a>s that is using it (the <a
2732href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
2733graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
2734def-use information in the program, and is accessible through the <tt>use_</tt>*
2735methods, shown below.</p>
2736
2737<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
2738and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
2739method. In addition, all LLVM values can be named. The "name" of the
2740<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
2741
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2742<div class="doc_code">
2743<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]2744%<b>foo</b> = add i32 1, 2
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2745</pre>
2746</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2747
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]2748<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]2749that the name of any value may be missing (an empty string), so names should
2750<b>ONLY</b> be used for debugging (making the source code easier to read,
2751debugging printouts), they should not be used to keep track of values or map
2752between them. For this purpose, use a <tt>std::map</tt> of pointers to the
2753<tt>Value</tt> itself instead.</p>
2754
2755<p>One important aspect of LLVM is that there is no distinction between an SSA
2756variable and the operation that produces it. Because of this, any reference to
2757the value produced by an instruction (or the value available as an incoming
Chris Lattnerd5fc4fc2004-03-18 14:58:55 +0000[diff] [blame]2758argument, for example) is represented as a direct pointer to the instance of
2759the class that
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2760represents this value. Although this may take some getting used to, it
2761simplifies the representation and makes it easier to manipulate.</p>
2762
2763</div>
2764
2765<!-- _______________________________________________________________________ -->
2766<div class="doc_subsubsection">
2767 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
2768</div>
2769
2770<div class="doc_text">
2771
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2772<ul>
2773 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
2774use-list<br>
2775 <tt>Value::use_const_iterator</tt> - Typedef for const_iterator over
2776the use-list<br>
2777 <tt>unsigned use_size()</tt> - Returns the number of users of the
2778value.<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2779 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2780 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
2781the use-list.<br>
2782 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
2783use-list.<br>
2784 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
2785element in the list.
2786 <p> These methods are the interface to access the def-use
2787information in LLVM. As with all other iterators in LLVM, the naming
2788conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2789 </li>
2790 <li><tt><a href="#Type">Type</a> *getType() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2791 <p>This method returns the Type of the Value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2792 </li>
2793 <li><tt>bool hasName() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2794 <tt>std::string getName() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2795 <tt>void setName(const std::string &amp;Name)</tt>
2796 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
2797be aware of the <a href="#nameWarning">precaution above</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2798 </li>
2799 <li><tt>void replaceAllUsesWith(Value *V)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2800
2801 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
2802 href="#User"><tt>User</tt>s</a> of the current value to refer to
2803 "<tt>V</tt>" instead. For example, if you detect that an instruction always
2804 produces a constant value (for example through constant folding), you can
2805 replace all uses of the instruction with the constant like this:</p>
2806
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2807<div class="doc_code">
2808<pre>
2809Inst-&gt;replaceAllUsesWith(ConstVal);
2810</pre>
2811</div>
2812
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2813</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2814
2815</div>
2816
2817<!-- ======================================================================= -->
2818<div class="doc_subsection">
2819 <a name="User">The <tt>User</tt> class</a>
2820</div>
2821
2822<div class="doc_text">
2823
2824<p>
2825<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]2826doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2827Superclass: <a href="#Value"><tt>Value</tt></a></p>
2828
2829<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
2830refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
2831that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
2832referring to. The <tt>User</tt> class itself is a subclass of
2833<tt>Value</tt>.</p>
2834
2835<p>The operands of a <tt>User</tt> point directly to the LLVM <a
2836href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
2837Single Assignment (SSA) form, there can only be one definition referred to,
2838allowing this direct connection. This connection provides the use-def
2839information in LLVM.</p>
2840
2841</div>
2842
2843<!-- _______________________________________________________________________ -->
2844<div class="doc_subsubsection">
2845 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
2846</div>
2847
2848<div class="doc_text">
2849
2850<p>The <tt>User</tt> class exposes the operand list in two ways: through
2851an index access interface and through an iterator based interface.</p>
2852
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2853<ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2854 <li><tt>Value *getOperand(unsigned i)</tt><br>
2855 <tt>unsigned getNumOperands()</tt>
2856 <p> These two methods expose the operands of the <tt>User</tt> in a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2857convenient form for direct access.</p></li>
2858
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2859 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
2860list<br>
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]2861 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
2862the operand list.<br>
2863 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2864operand list.
2865 <p> Together, these methods make up the iterator based interface to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2866the operands of a <tt>User</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2867</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2868
2869</div>
2870
2871<!-- ======================================================================= -->
2872<div class="doc_subsection">
2873 <a name="Instruction">The <tt>Instruction</tt> class</a>
2874</div>
2875
2876<div class="doc_text">
2877
2878<p><tt>#include "</tt><tt><a
2879href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2880doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2881Superclasses: <a href="#User"><tt>User</tt></a>, <a
2882href="#Value"><tt>Value</tt></a></p>
2883
2884<p>The <tt>Instruction</tt> class is the common base class for all LLVM
2885instructions. It provides only a few methods, but is a very commonly used
2886class. The primary data tracked by the <tt>Instruction</tt> class itself is the
2887opcode (instruction type) and the parent <a
2888href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
2889into. To represent a specific type of instruction, one of many subclasses of
2890<tt>Instruction</tt> are used.</p>
2891
2892<p> Because the <tt>Instruction</tt> class subclasses the <a
2893href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
2894way as for other <a href="#User"><tt>User</tt></a>s (with the
2895<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
2896<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
2897the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
2898file contains some meta-data about the various different types of instructions
2899in LLVM. It describes the enum values that are used as opcodes (for example
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2900<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2901concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
2902example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2903href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2904this file confuses doxygen, so these enum values don't show up correctly in the
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2905<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2906
2907</div>
2908
2909<!-- _______________________________________________________________________ -->
2910<div class="doc_subsubsection">
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]2911 <a name="s_Instruction">Important Subclasses of the <tt>Instruction</tt>
2912 class</a>
2913</div>
2914<div class="doc_text">
2915 <ul>
2916 <li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
2917 <p>This subclasses represents all two operand instructions whose operands
2918 must be the same type, except for the comparison instructions.</p></li>
2919 <li><tt><a name="CastInst">CastInst</a></tt>
2920 <p>This subclass is the parent of the 12 casting instructions. It provides
2921 common operations on cast instructions.</p>
2922 <li><tt><a name="CmpInst">CmpInst</a></tt>
2923 <p>This subclass respresents the two comparison instructions,
2924 <a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
2925 <a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
2926 <li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
2927 <p>This subclass is the parent of all terminator instructions (those which
2928 can terminate a block).</p>
2929 </ul>
2930 </div>
2931
2932<!-- _______________________________________________________________________ -->
2933<div class="doc_subsubsection">
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2934 <a name="m_Instruction">Important Public Members of the <tt>Instruction</tt>
2935 class</a>
2936</div>
2937
2938<div class="doc_text">
2939
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2940<ul>
2941 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2942 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
2943this <tt>Instruction</tt> is embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2944 <li><tt>bool mayWriteToMemory()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2945 <p>Returns true if the instruction writes to memory, i.e. it is a
2946 <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]2947 <li><tt>unsigned getOpcode()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2948 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2949 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2950 <p>Returns another instance of the specified instruction, identical
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2951in all ways to the original except that the instruction has no parent
2952(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2953and it has no name</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2954</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2955
2956</div>
2957
2958<!-- ======================================================================= -->
2959<div class="doc_subsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2960 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2961</div>
2962
2963<div class="doc_text">
2964
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2965<p>Constant represents a base class for different types of constants. It
2966is subclassed by ConstantInt, ConstantArray, etc. for representing
2967the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
2968a subclass, which represents the address of a global variable or function.
2969</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2970
2971</div>
2972
2973<!-- _______________________________________________________________________ -->
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2974<div class="doc_subsubsection">Important Subclasses of Constant </div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2975<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2976<ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2977 <li>ConstantInt : This subclass of Constant represents an integer constant of
2978 any width.
2979 <ul>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]2980 <li><tt>const APInt&amp; getValue() const</tt>: Returns the underlying
2981 value of this constant, an APInt value.</li>
2982 <li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
2983 value to an int64_t via sign extension. If the value (not the bit width)
2984 of the APInt is too large to fit in an int64_t, an assertion will result.
2985 For this reason, use of this method is discouraged.</li>
2986 <li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
2987 value to a uint64_t via zero extension. IF the value (not the bit width)
2988 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]2989 For this reason, use of this method is discouraged.</li>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]2990 <li><tt>static ConstantInt* get(const APInt&amp; Val)</tt>: Returns the
2991 ConstantInt object that represents the value provided by <tt>Val</tt>.
2992 The type is implied as the IntegerType that corresponds to the bit width
2993 of <tt>Val</tt>.</li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]2994 <li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
2995 Returns the ConstantInt object that represents the value provided by
2996 <tt>Val</tt> for integer type <tt>Ty</tt>.</li>
2997 </ul>
2998 </li>
2999 <li>ConstantFP : This class represents a floating point constant.
3000 <ul>
3001 <li><tt>double getValue() const</tt>: Returns the underlying value of
3002 this constant. </li>
3003 </ul>
3004 </li>
3005 <li>ConstantArray : This represents a constant array.
3006 <ul>
3007 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
3008 a vector of component constants that makeup this array. </li>
3009 </ul>
3010 </li>
3011 <li>ConstantStruct : This represents a constant struct.
3012 <ul>
3013 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
3014 a vector of component constants that makeup this array. </li>
3015 </ul>
3016 </li>
3017 <li>GlobalValue : This represents either a global variable or a function. In
3018 either case, the value is a constant fixed address (after linking).
3019 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3020</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3021</div>
3022
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3023
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3024<!-- ======================================================================= -->
3025<div class="doc_subsection">
3026 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
3027</div>
3028
3029<div class="doc_text">
3030
3031<p><tt>#include "<a
3032href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]3033doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
3034Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3035Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
3036<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]3037
3038<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
3039href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
3040visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
3041Because they are visible at global scope, they are also subject to linking with
3042other globals defined in different translation units. To control the linking
3043process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
3044<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]3045defined by the <tt>LinkageTypes</tt> enumeration.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3046
3047<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
3048<tt>static</tt> in C), it is not visible to code outside the current translation
3049unit, and does not participate in linking. If it has external linkage, it is
3050visible to external code, and does participate in linking. In addition to
3051linkage information, <tt>GlobalValue</tt>s keep track of which <a
3052href="#Module"><tt>Module</tt></a> they are currently part of.</p>
3053
3054<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
3055by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
3056global is always a pointer to its contents. It is important to remember this
3057when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
3058be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
3059subclass 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]3060i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3061the address of the first element of this array and the value of the
3062<tt>GlobalVariable</tt> are the same, they have different types. The
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]3063<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
3064is <tt>i32.</tt> Because of this, accessing a global value requires you to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3065dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
3066can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
3067Language Reference Manual</a>.</p>
3068
3069</div>
3070
3071<!-- _______________________________________________________________________ -->
3072<div class="doc_subsubsection">
3073 <a name="m_GlobalValue">Important Public Members of the <tt>GlobalValue</tt>
3074 class</a>
3075</div>
3076
3077<div class="doc_text">
3078
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3079<ul>
3080 <li><tt>bool hasInternalLinkage() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3081 <tt>bool hasExternalLinkage() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3082 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
3083 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
3084 <p> </p>
3085 </li>
3086 <li><tt><a href="#Module">Module</a> *getParent()</tt>
3087 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3088GlobalValue is currently embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3089</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3090
3091</div>
3092
3093<!-- ======================================================================= -->
3094<div class="doc_subsection">
3095 <a name="Function">The <tt>Function</tt> class</a>
3096</div>
3097
3098<div class="doc_text">
3099
3100<p><tt>#include "<a
3101href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]3102info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3103Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3104<a href="#Constant"><tt>Constant</tt></a>,
3105<a href="#User"><tt>User</tt></a>,
3106<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3107
3108<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
3109actually one of the more complex classes in the LLVM heirarchy because it must
3110keep track of a large amount of data. The <tt>Function</tt> class keeps track
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3111of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
3112<a href="#Argument"><tt>Argument</tt></a>s, and a
3113<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3114
3115<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
3116commonly used part of <tt>Function</tt> objects. The list imposes an implicit
3117ordering of the blocks in the function, which indicate how the code will be
3118layed out by the backend. Additionally, the first <a
3119href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
3120<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
3121block. There are no implicit exit nodes, and in fact there may be multiple exit
3122nodes from a single <tt>Function</tt>. If the <a
3123href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
3124the <tt>Function</tt> is actually a function declaration: the actual body of the
3125function hasn't been linked in yet.</p>
3126
3127<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
3128<tt>Function</tt> class also keeps track of the list of formal <a
3129href="#Argument"><tt>Argument</tt></a>s that the function receives. This
3130container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
3131nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
3132the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
3133
3134<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
3135LLVM feature that is only used when you have to look up a value by name. Aside
3136from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
3137internally to make sure that there are not conflicts between the names of <a
3138href="#Instruction"><tt>Instruction</tt></a>s, <a
3139href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
3140href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
3141
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]3142<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
3143and therefore also a <a href="#Constant">Constant</a>. The value of the function
3144is its address (after linking) which is guaranteed to be constant.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3145</div>
3146
3147<!-- _______________________________________________________________________ -->
3148<div class="doc_subsubsection">
3149 <a name="m_Function">Important Public Members of the <tt>Function</tt>
3150 class</a>
3151</div>
3152
3153<div class="doc_text">
3154
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3155<ul>
3156 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]3157 *Ty, LinkageTypes Linkage, const std::string &amp;N = "", Module* Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3158
3159 <p>Constructor used when you need to create new <tt>Function</tt>s to add
3160 the the program. The constructor must specify the type of the function to
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]3161 create and what type of linkage the function should have. The <a
3162 href="#FunctionType"><tt>FunctionType</tt></a> argument
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3163 specifies the formal arguments and return value for the function. The same
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3164 <a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3165 create multiple functions. The <tt>Parent</tt> argument specifies the Module
3166 in which the function is defined. If this argument is provided, the function
3167 will automatically be inserted into that module's list of
3168 functions.</p></li>
3169
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3170 <li><tt>bool isExternal()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3171
3172 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
3173 function is "external", it does not have a body, and thus must be resolved
3174 by linking with a function defined in a different translation unit.</p></li>
3175
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3176 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3177 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3178
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]3179 <tt>begin()</tt>, <tt>end()</tt>
3180 <tt>size()</tt>, <tt>empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3181
3182 <p>These are forwarding methods that make it easy to access the contents of
3183 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
3184 list.</p></li>
3185
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3186 <li><tt>Function::BasicBlockListType &amp;getBasicBlockList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3187
3188 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
3189 is necessary to use when you need to update the list or perform a complex
3190 action that doesn't have a forwarding method.</p></li>
3191
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]3192 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3193iterator<br>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]3194 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3195
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]3196 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]3197 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3198
3199 <p>These are forwarding methods that make it easy to access the contents of
3200 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
3201 list.</p></li>
3202
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3203 <li><tt>Function::ArgumentListType &amp;getArgumentList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3204
3205 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
3206 necessary to use when you need to update the list or perform a complex
3207 action that doesn't have a forwarding method.</p></li>
3208
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3209 <li><tt><a href="#BasicBlock">BasicBlock</a> &amp;getEntryBlock()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3210
3211 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
3212 function. Because the entry block for the function is always the first
3213 block, this returns the first block of the <tt>Function</tt>.</p></li>
3214
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3215 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
3216 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3217
3218 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
3219 <tt>Function</tt> and returns the return type of the function, or the <a
3220 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
3221 function.</p></li>
3222
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3223 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3224
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3225 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3226 for this <tt>Function</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3227</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3228
3229</div>
3230
3231<!-- ======================================================================= -->
3232<div class="doc_subsection">
3233 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
3234</div>
3235
3236<div class="doc_text">
3237
3238<p><tt>#include "<a
3239href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
3240<br>
Tanya Lattnera3da7772004-06-22 08:02:25 +0000[diff] [blame]3241doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3242 Class</a><br>
3243Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3244<a href="#Constant"><tt>Constant</tt></a>,
3245<a href="#User"><tt>User</tt></a>,
3246<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3247
3248<p>Global variables are represented with the (suprise suprise)
3249<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
3250subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
3251always referenced by their address (global values must live in memory, so their
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3252"name" refers to their constant address). See
3253<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
3254variables may have an initial value (which must be a
3255<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
3256they may be marked as "constant" themselves (indicating that their contents
3257never change at runtime).</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3258</div>
3259
3260<!-- _______________________________________________________________________ -->
3261<div class="doc_subsubsection">
3262 <a name="m_GlobalVariable">Important Public Members of the
3263 <tt>GlobalVariable</tt> class</a>
3264</div>
3265
3266<div class="doc_text">
3267
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3268<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3269 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
3270 isConstant, LinkageTypes&amp; Linkage, <a href="#Constant">Constant</a>
3271 *Initializer = 0, const std::string &amp;Name = "", Module* Parent = 0)</tt>
3272
3273 <p>Create a new global variable of the specified type. If
3274 <tt>isConstant</tt> is true then the global variable will be marked as
3275 unchanging for the program. The Linkage parameter specifies the type of
3276 linkage (internal, external, weak, linkonce, appending) for the variable. If
3277 the linkage is InternalLinkage, WeakLinkage, or LinkOnceLinkage,&nbsp; then
3278 the resultant global variable will have internal linkage. AppendingLinkage
3279 concatenates together all instances (in different translation units) of the
3280 variable into a single variable but is only applicable to arrays. &nbsp;See
3281 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
3282 further details on linkage types. Optionally an initializer, a name, and the
3283 module to put the variable into may be specified for the global variable as
3284 well.</p></li>
3285
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3286 <li><tt>bool isConstant() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3287
3288 <p>Returns true if this is a global variable that is known not to
3289 be modified at runtime.</p></li>
3290
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3291 <li><tt>bool hasInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3292
3293 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
3294
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3295 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3296
3297 <p>Returns the intial value for a <tt>GlobalVariable</tt>. It is not legal
3298 to call this method if there is no initializer.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3299</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3300
3301</div>
3302
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3303
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3304<!-- ======================================================================= -->
3305<div class="doc_subsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3306 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3307</div>
3308
3309<div class="doc_text">
3310
3311<p><tt>#include "<a
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3312href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
3313doxygen info: <a href="/doxygen/structllvm_1_1BasicBlock.html">BasicBlock
3314Class</a><br>
3315Superclass: <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3316
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3317<p>This class represents a single entry multiple exit section of the code,
3318commonly known as a basic block by the compiler community. The
3319<tt>BasicBlock</tt> class maintains a list of <a
3320href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
3321Matching the language definition, the last element of this list of instructions
3322is always a terminator instruction (a subclass of the <a
3323href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
3324
3325<p>In addition to tracking the list of instructions that make up the block, the
3326<tt>BasicBlock</tt> class also keeps track of the <a
3327href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
3328
3329<p>Note that <tt>BasicBlock</tt>s themselves are <a
3330href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
3331like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
3332<tt>label</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3333
3334</div>
3335
3336<!-- _______________________________________________________________________ -->
3337<div class="doc_subsubsection">
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3338 <a name="m_BasicBlock">Important Public Members of the <tt>BasicBlock</tt>
3339 class</a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3340</div>
3341
3342<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3343<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3344
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3345<li><tt>BasicBlock(const std::string &amp;Name = "", </tt><tt><a
3346 href="#Function">Function</a> *Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3347
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3348<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
3349insertion into a function. The constructor optionally takes a name for the new
3350block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
3351the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
3352automatically inserted at the end of the specified <a
3353href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
3354manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3355
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3356<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
3357<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
3358<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
3359<tt>size()</tt>, <tt>empty()</tt>
3360STL-style functions for accessing the instruction list.
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3361
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3362<p>These methods and typedefs are forwarding functions that have the same
3363semantics as the standard library methods of the same names. These methods
3364expose the underlying instruction list of a basic block in a way that is easy to
3365manipulate. To get the full complement of container operations (including
3366operations to update the list), you must use the <tt>getInstList()</tt>
3367method.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3368
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3369<li><tt>BasicBlock::InstListType &amp;getInstList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3370
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3371<p>This method is used to get access to the underlying container that actually
3372holds the Instructions. This method must be used when there isn't a forwarding
3373function in the <tt>BasicBlock</tt> class for the operation that you would like
3374to perform. Because there are no forwarding functions for "updating"
3375operations, you need to use this if you want to update the contents of a
3376<tt>BasicBlock</tt>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3377
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3378<li><tt><a href="#Function">Function</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3379
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3380<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
3381embedded into, or a null pointer if it is homeless.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3382
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3383<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3384
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3385<p> Returns a pointer to the terminator instruction that appears at the end of
3386the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
3387instruction in the block is not a terminator, then a null pointer is
3388returned.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3389
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3390</ul>
3391
3392</div>
3393
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3394
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3395<!-- ======================================================================= -->
3396<div class="doc_subsection">
3397 <a name="Argument">The <tt>Argument</tt> class</a>
3398</div>
3399
3400<div class="doc_text">
3401
3402<p>This subclass of Value defines the interface for incoming formal
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]3403arguments to a function. A Function maintains a list of its formal
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3404arguments. An argument has a pointer to the parent Function.</p>
3405
3406</div>
3407
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3408<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3409<hr>
3410<address>
3411 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3412 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3413 <a href="http://validator.w3.org/check/referer"><img
Gabor Greifa9c0f2b2008-06-18 14:05:31 +0000[diff] [blame]3414 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01 Strict"></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3415
3416 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
3417 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]3418 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3419 Last modified: $Date$
3420</address>
3421
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3422</body>
3423</html>