Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / b856d555b0336234cb4d5e2d77c3d506efa17550 / . / docs / ProgrammersManual.html
blob: e00dd4a66d0dc0868910268ad6b2461b94d37b63 [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>
Owen Anderson5e8c50e2009-06-16 17:40:28 +0000[diff] [blame]5 <meta http-equiv="Content-type" content="text/html;charset=UTF-8">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]6 <title>LLVM Programmer's Manual</title>
Daniel Dunbaradea4972012-04-19 20:20:34 +0000[diff] [blame]7 <link rel="stylesheet" href="_static/llvm.css" type="text/css">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]8</head>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]9<body>
10
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]11<h1>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]12 LLVM Programmer's Manual
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]13</h1>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]14
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]15<ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]16 <li><a href="#introduction">Introduction</a></li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]17 <li><a href="#general">General Information</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]18 <ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]19 <li><a href="#stl">The C++ Standard Template Library</a></li>
20<!--
21 <li>The <tt>-time-passes</tt> option</li>
22 <li>How to use the LLVM Makefile system</li>
23 <li>How to write a regression test</li>
Chris Lattner61db4652004-12-08 19:05:44 +0000[diff] [blame]24
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]25-->
Chris Lattner84b7f8d2003-08-01 22:20:59 +0000[diff] [blame]26 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]27 </li>
28 <li><a href="#apis">Important and useful LLVM APIs</a>
29 <ul>
30 <li><a href="#isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt>
31and <tt>dyn_cast&lt;&gt;</tt> templates</a> </li>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]32 <li><a href="#string_apis">Passing strings (the <tt>StringRef</tt>
Benjamin Kramere15192b2009-08-05 15:42:44 +0000[diff] [blame]33and <tt>Twine</tt> classes)</a>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]34 <ul>
35 <li><a href="#StringRef">The <tt>StringRef</tt> class</a> </li>
36 <li><a href="#Twine">The <tt>Twine</tt> class</a> </li>
37 </ul>
Benjamin Kramere15192b2009-08-05 15:42:44 +0000[diff] [blame]38 </li>
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]39 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]40option</a>
41 <ul>
42 <li><a href="#DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt>
43and the <tt>-debug-only</tt> option</a> </li>
44 </ul>
45 </li>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]46 <li><a href="#Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]47option</a></li>
48<!--
49 <li>The <tt>InstVisitor</tt> template
50 <li>The general graph API
51-->
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]52 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]53 </ul>
54 </li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]55 <li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
56 <ul>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]57 <li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
58 <ul>
Chris Lattner8ae42612011-04-05 23:18:20 +0000[diff] [blame]59 <li><a href="#dss_arrayref">llvm/ADT/ArrayRef.h</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]60 <li><a href="#dss_fixedarrays">Fixed Size Arrays</a></li>
61 <li><a href="#dss_heaparrays">Heap Allocated Arrays</a></li>
Chris Lattner9d69d4a2011-07-18 01:40:02 +0000[diff] [blame]62 <li><a href="#dss_tinyptrvector">"llvm/ADT/TinyPtrVector.h"</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]63 <li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
64 <li><a href="#dss_vector">&lt;vector&gt;</a></li>
65 <li><a href="#dss_deque">&lt;deque&gt;</a></li>
66 <li><a href="#dss_list">&lt;list&gt;</a></li>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]67 <li><a href="#dss_ilist">llvm/ADT/ilist.h</a></li>
Argyrios Kyrtzidis81df0892011-06-15 19:56:01 +0000[diff] [blame]68 <li><a href="#dss_packedvector">llvm/ADT/PackedVector.h</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]69 <li><a href="#dss_other">Other Sequential Container Options</a></li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]70 </ul></li>
Chris Lattner7314a202011-07-22 20:46:49 +0000[diff] [blame]71 <li><a href="#ds_string">String-like containers</a>
Chris Lattner66827462011-07-22 21:36:29 +0000[diff] [blame]72 <ul>
73 <li><a href="#dss_stringref">llvm/ADT/StringRef.h</a></li>
74 <li><a href="#dss_twine">llvm/ADT/Twine.h</a></li>
75 <li><a href="#dss_smallstring">llvm/ADT/SmallString.h</a></li>
76 <li><a href="#dss_stdstring">std::string</a></li>
77 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]78 <li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
79 <ul>
80 <li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
81 <li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
82 <li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]83 <li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
Jakob Stoklund Olesen62588622012-02-22 00:56:08 +0000[diff] [blame]84 <li><a href="#dss_sparseset">"llvm/ADT/SparseSet.h"</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]85 <li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
86 <li><a href="#dss_set">&lt;set&gt;</a></li>
87 <li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]88 <li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
Chris Lattner2fdd0052011-11-15 22:40:14 +0000[diff] [blame]89 <li><a href="#dss_immutableset">"llvm/ADT/ImmutableSet.h"</a></li>
90 <li><a href="#dss_otherset">Other Set-Like Container Options</a></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]91 </ul></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]92 <li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
93 <ul>
94 <li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]95 <li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]96 <li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
97 <li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
Benjamin Kramerb856d552012-04-25 18:01:58 +0000[diff] [blame^]98 <li><a href="#dss_multiimplmap">"llvm/ADT/MultiImplMap.h"</a></li>
99 <li><a href="#dss_flatarraymap">"llvm/ADT/FlatArrayMap.h"</a></li>
100 <li><a href="#dss_smallmap">"llvm/ADT/SmallMap.h"</a></li>
Jeffrey Yasskin71a5c222009-10-22 22:11:22 +0000[diff] [blame]101 <li><a href="#dss_valuemap">"llvm/ADT/ValueMap.h"</a></li>
Jakob Stoklund Olesenaca0da62010-12-14 00:55:51 +0000[diff] [blame]102 <li><a href="#dss_intervalmap">"llvm/ADT/IntervalMap.h"</a></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]103 <li><a href="#dss_map">&lt;map&gt;</a></li>
Jakob Stoklund Olesen4359e5e2011-04-05 20:56:08 +0000[diff] [blame]104 <li><a href="#dss_inteqclasses">"llvm/ADT/IntEqClasses.h"</a></li>
Chris Lattner2fdd0052011-11-15 22:40:14 +0000[diff] [blame]105 <li><a href="#dss_immutablemap">"llvm/ADT/ImmutableMap.h"</a></li>
Chris Lattnerf3692522007-02-03 19:51:56 +0000[diff] [blame]106 <li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
107 </ul></li>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]108 <li><a href="#ds_bit">BitVector-like containers</a>
109 <ul>
110 <li><a href="#dss_bitvector">A dense bitvector</a></li>
Dan Gohman5f7775c2010-01-05 18:24:00 +0000[diff] [blame]111 <li><a href="#dss_smallbitvector">A "small" dense bitvector</a></li>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]112 <li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
113 </ul></li>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]114 </ul>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]115 </li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]116 <li><a href="#common">Helpful Hints for Common Operations</a>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]117 <ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]118 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
119 <ul>
120 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
121in a <tt>Function</tt></a> </li>
122 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
123in a <tt>BasicBlock</tt></a> </li>
124 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
125in a <tt>Function</tt></a> </li>
126 <li><a href="#iterate_convert">Turning an iterator into a
127class pointer</a> </li>
128 <li><a href="#iterate_complex">Finding call sites: a more
129complex example</a> </li>
130 <li><a href="#calls_and_invokes">Treating calls and invokes
131the same way</a> </li>
132 <li><a href="#iterate_chains">Iterating over def-use &amp;
133use-def chains</a> </li>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]134 <li><a href="#iterate_preds">Iterating over predecessors &amp;
135successors of blocks</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]136 </ul>
137 </li>
138 <li><a href="#simplechanges">Making simple changes</a>
139 <ul>
140 <li><a href="#schanges_creating">Creating and inserting new
141 <tt>Instruction</tt>s</a> </li>
142 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
143 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
144with another <tt>Value</tt></a> </li>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]145 <li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]146 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]147 </li>
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]148 <li><a href="#create_types">How to Create Types</a></li>
Chris Lattnerae7f7592002-09-06 18:31:18 +0000[diff] [blame]149<!--
150 <li>Working with the Control Flow Graph
151 <ul>
152 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
153 <li>
154 <li>
155 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]156-->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]157 </ul>
158 </li>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]159
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]160 <li><a href="#threading">Threads and LLVM</a>
161 <ul>
Owen Anderson1ad70e32009-06-16 18:04:19 +0000[diff] [blame]162 <li><a href="#startmultithreaded">Entering and Exiting Multithreaded Mode
163 </a></li>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]164 <li><a href="#shutdown">Ending execution with <tt>llvm_shutdown()</tt></a></li>
165 <li><a href="#managedstatic">Lazy initialization with <tt>ManagedStatic</tt></a></li>
Owen Andersone0c951a2009-08-19 17:58:52 +0000[diff] [blame]166 <li><a href="#llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a></li>
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]167 <li><a href="#jitthreading">Threads and the JIT</a></li>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]168 </ul>
169 </li>
170
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]171 <li><a href="#advanced">Advanced Topics</a>
172 <ul>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]173
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]174 <li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> class</a></li>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]175 <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]176 </ul></li>
177
Joel Stanley9b96c442002-09-06 21:55:13 +0000[diff] [blame]178 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]179 <ul>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]180 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]181 <li><a href="#Module">The <tt>Module</tt> class</a></li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]182 <li><a href="#Value">The <tt>Value</tt> class</a>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]183 <ul>
184 <li><a href="#User">The <tt>User</tt> class</a>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]185 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]186 <li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
187 <li><a href="#Constant">The <tt>Constant</tt> class</a>
188 <ul>
189 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]190 <ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]191 <li><a href="#Function">The <tt>Function</tt> class</a></li>
192 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
193 </ul>
194 </li>
195 </ul>
196 </li>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]197 </ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]198 </li>
199 <li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
200 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
201 </ul>
Reid Spencerfe8f4ff2004-11-01 09:02:53 +0000[diff] [blame]202 </li>
203 </ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]204 </li>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]205</ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]206
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]207<div class="doc_author">
208 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
Chris Lattner94c43592004-05-26 16:52:55 +0000[diff] [blame]209 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]210 <a href="mailto:ggreif@gmail.com">Gabor Greif</a>,
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]211 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>,
212 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a> and
213 <a href="mailto:owen@apple.com">Owen Anderson</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]214</div>
215
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]216<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]217<h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]218 <a name="introduction">Introduction </a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]219</h2>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]220<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]221
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]222<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]223
224<p>This document is meant to highlight some of the important classes and
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]225interfaces available in the LLVM source-base. This manual is not
226intended to explain what LLVM is, how it works, and what LLVM code looks
227like. It assumes that you know the basics of LLVM and are interested
228in writing transformations or otherwise analyzing or manipulating the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]229code.</p>
230
231<p>This document should get you oriented so that you can find your
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]232way in the continuously growing source code that makes up the LLVM
233infrastructure. Note that this manual is not intended to serve as a
234replacement for reading the source code, so if you think there should be
235a method in one of these classes to do something, but it's not listed,
236check the source. Links to the <a href="/doxygen/">doxygen</a> sources
237are provided to make this as easy as possible.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]238
239<p>The first section of this document describes general information that is
240useful to know when working in the LLVM infrastructure, and the second describes
241the Core LLVM classes. In the future this manual will be extended with
242information describing how to use extension libraries, such as dominator
243information, CFG traversal routines, and useful utilities like the <tt><a
244href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
245
246</div>
247
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]248<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]249<h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]250 <a name="general">General Information</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]251</h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]252<!-- *********************************************************************** -->
253
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]254<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]255
256<p>This section contains general information that is useful if you are working
257in the LLVM source-base, but that isn't specific to any particular API.</p>
258
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]259<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]260<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]261 <a name="stl">The C++ Standard Template Library</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]262</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]263
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]264<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]265
266<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]267perhaps much more than you are used to, or have seen before. Because of
268this, you might want to do a little background reading in the
269techniques used and capabilities of the library. There are many good
270pages that discuss the STL, and several books on the subject that you
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]271can get, so it will not be discussed in this document.</p>
272
273<p>Here are some useful links:</p>
274
275<ol>
276
Nick Lewyckyea1fe2c2010-10-09 21:12:29 +0000[diff] [blame]277<li><a href="http://www.dinkumware.com/manuals/#Standard C++ Library">Dinkumware
278C++ Library reference</a> - an excellent reference for the STL and other parts
279of the standard C++ library.</li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]280
281<li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
Gabor Greif0cbcabe2009-03-12 09:47:03 +0000[diff] [blame]282O'Reilly book in the making. It has a decent Standard Library
283Reference that rivals Dinkumware's, and is unfortunately no longer free since the
284book has been published.</li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]285
286<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
287Questions</a></li>
288
289<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
290Contains a useful <a
291href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
292STL</a>.</li>
293
294<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
295Page</a></li>
296
Tanya Lattner79445ba2004-12-08 18:34:56 +0000[diff] [blame]297<li><a href="http://64.78.49.204/">
Reid Spencer096603a2004-05-26 08:41:35 +0000[diff] [blame]298Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
299the book).</a></li>
300
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]301</ol>
302
303<p>You are also encouraged to take a look at the <a
304href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
305to write maintainable code more than where to put your curly braces.</p>
306
307</div>
308
309<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]310<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]311 <a name="stl">Other useful references</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]312</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]313
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]314<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]315
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]316<ol>
Misha Brukmana0f71e42004-06-18 18:39:00 +0000[diff] [blame]317<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
318static and shared libraries across platforms</a></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]319</ol>
320
321</div>
322
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]323</div>
324
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]325<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]326<h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]327 <a name="apis">Important and useful LLVM APIs</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]328</h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]329<!-- *********************************************************************** -->
330
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]331<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]332
333<p>Here we highlight some LLVM APIs that are generally useful and good to
334know about when writing transformations.</p>
335
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]336<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]337<h3>
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]338 <a name="isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt> and
339 <tt>dyn_cast&lt;&gt;</tt> templates</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]340</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]341
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]342<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]343
344<p>The LLVM source-base makes extensive use of a custom form of RTTI.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]345These templates have many similarities to the C++ <tt>dynamic_cast&lt;&gt;</tt>
346operator, but they don't have some drawbacks (primarily stemming from
347the fact that <tt>dynamic_cast&lt;&gt;</tt> only works on classes that
348have a v-table). Because they are used so often, you must know what they
349do and how they work. All of these templates are defined in the <a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]350 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]351file (note that you very rarely have to include this file directly).</p>
352
353<dl>
354 <dt><tt>isa&lt;&gt;</tt>: </dt>
355
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]356 <dd><p>The <tt>isa&lt;&gt;</tt> operator works exactly like the Java
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]357 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
358 a reference or pointer points to an instance of the specified class. This can
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]359 be very useful for constraint checking of various sorts (example below).</p>
360 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]361
362 <dt><tt>cast&lt;&gt;</tt>: </dt>
363
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]364 <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]365 converts a pointer or reference from a base class to a derived class, causing
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]366 an assertion failure if it is not really an instance of the right type. This
367 should be used in cases where you have some information that makes you believe
368 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]369 and <tt>cast&lt;&gt;</tt> template is:</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>
373static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
374 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))
375 return true;
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]376
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]377 // <i>Otherwise, it must be an instruction...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]378 return !L-&gt;contains(cast&lt;<a href="#Instruction">Instruction</a>&gt;(V)-&gt;getParent());
379}
380</pre>
381</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]382
383 <p>Note that you should <b>not</b> use an <tt>isa&lt;&gt;</tt> test followed
384 by a <tt>cast&lt;&gt;</tt>, for that use the <tt>dyn_cast&lt;&gt;</tt>
385 operator.</p>
386
387 </dd>
388
389 <dt><tt>dyn_cast&lt;&gt;</tt>:</dt>
390
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]391 <dd><p>The <tt>dyn_cast&lt;&gt;</tt> operator is a "checking cast" operation.
392 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]393 pointer to it (this operator does not work with references). If the operand is
394 not of the correct type, a null pointer is returned. Thus, this works very
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]395 much like the <tt>dynamic_cast&lt;&gt;</tt> operator in C++, and should be
396 used in the same circumstances. Typically, the <tt>dyn_cast&lt;&gt;</tt>
397 operator is used in an <tt>if</tt> statement or some other flow control
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]398 statement like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]399
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]400<div class="doc_code">
401<pre>
402if (<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]403 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]404}
405</pre>
406</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]407
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]408 <p>This form of the <tt>if</tt> statement effectively combines together a call
409 to <tt>isa&lt;&gt;</tt> and a call to <tt>cast&lt;&gt;</tt> into one
410 statement, which is very convenient.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]411
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]412 <p>Note that the <tt>dyn_cast&lt;&gt;</tt> operator, like C++'s
413 <tt>dynamic_cast&lt;&gt;</tt> or Java's <tt>instanceof</tt> operator, can be
414 abused. In particular, you should not use big chained <tt>if/then/else</tt>
415 blocks to check for lots of different variants of classes. If you find
416 yourself wanting to do this, it is much cleaner and more efficient to use the
417 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]418
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]419 </dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]420
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]421 <dt><tt>cast_or_null&lt;&gt;</tt>: </dt>
422
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]423 <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]424 <tt>cast&lt;&gt;</tt> operator, except that it allows for a null pointer as an
425 argument (which it then propagates). This can sometimes be useful, allowing
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]426 you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]427
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]428 <dt><tt>dyn_cast_or_null&lt;&gt;</tt>: </dt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]429
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]430 <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]431 <tt>dyn_cast&lt;&gt;</tt> operator, except that it allows for a null pointer
432 as an argument (which it then propagates). This can sometimes be useful,
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]433 allowing you to combine several null checks into one.</p></dd>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]434
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]435</dl>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]436
437<p>These five templates can be used with any classes, whether they have a
438v-table or not. To add support for these templates, you simply need to add
439<tt>classof</tt> static methods to the class you are interested casting
440to. Describing this is currently outside the scope of this document, but there
441are lots of examples in the LLVM source base.</p>
442
443</div>
444
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]445
446<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]447<h3>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]448 <a name="string_apis">Passing strings (the <tt>StringRef</tt>
449and <tt>Twine</tt> classes)</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]450</h3>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]451
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]452<div>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]453
454<p>Although LLVM generally does not do much string manipulation, we do have
Chris Lattner81187ae2009-07-25 07:16:59 +0000[diff] [blame]455several important APIs which take strings. Two important examples are the
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]456Value class -- which has names for instructions, functions, etc. -- and the
457StringMap class which is used extensively in LLVM and Clang.</p>
458
459<p>These are generic classes, and they need to be able to accept strings which
460may have embedded null characters. Therefore, they cannot simply take
Chris Lattner81187ae2009-07-25 07:16:59 +0000[diff] [blame]461a <tt>const char *</tt>, and taking a <tt>const std::string&amp;</tt> requires
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]462clients to perform a heap allocation which is usually unnecessary. Instead,
Benjamin Kramer38e59892010-07-14 22:38:02 +0000[diff] [blame]463many LLVM APIs use a <tt>StringRef</tt> or a <tt>const Twine&amp;</tt> for
464passing strings efficiently.</p>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]465
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]466<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]467<h4>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]468 <a name="StringRef">The <tt>StringRef</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]469</h4>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]470
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]471<div>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]472
473<p>The <tt>StringRef</tt> data type represents a reference to a constant string
474(a character array and a length) and supports the common operations available
475on <tt>std:string</tt>, but does not require heap allocation.</p>
476
Chris Lattner81187ae2009-07-25 07:16:59 +0000[diff] [blame]477<p>It can be implicitly constructed using a C style null-terminated string,
478an <tt>std::string</tt>, or explicitly with a character pointer and length.
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]479For example, the <tt>StringRef</tt> find function is declared as:</p>
Chris Lattner81187ae2009-07-25 07:16:59 +0000[diff] [blame]480
Benjamin Kramer38e59892010-07-14 22:38:02 +0000[diff] [blame]481<pre class="doc_code">
482 iterator find(StringRef Key);
483</pre>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]484
485<p>and clients can call it using any one of:</p>
486
Benjamin Kramer38e59892010-07-14 22:38:02 +0000[diff] [blame]487<pre class="doc_code">
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]488 Map.find("foo"); <i>// Lookup "foo"</i>
489 Map.find(std::string("bar")); <i>// Lookup "bar"</i>
490 Map.find(StringRef("\0baz", 4)); <i>// Lookup "\0baz"</i>
491</pre>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]492
493<p>Similarly, APIs which need to return a string may return a <tt>StringRef</tt>
494instance, which can be used directly or converted to an <tt>std::string</tt>
495using the <tt>str</tt> member function. See
496"<tt><a href="/doxygen/classllvm_1_1StringRef_8h-source.html">llvm/ADT/StringRef.h</a></tt>"
497for more information.</p>
498
499<p>You should rarely use the <tt>StringRef</tt> class directly, because it contains
500pointers to external memory it is not generally safe to store an instance of the
Benjamin Kramer38e59892010-07-14 22:38:02 +0000[diff] [blame]501class (unless you know that the external storage will not be freed). StringRef is
502small and pervasive enough in LLVM that it should always be passed by value.</p>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]503
504</div>
505
506<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]507<h4>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]508 <a name="Twine">The <tt>Twine</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]509</h4>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]510
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]511<div>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]512
513<p>The <tt>Twine</tt> class is an efficient way for APIs to accept concatenated
514strings. For example, a common LLVM paradigm is to name one instruction based on
515the name of another instruction with a suffix, for example:</p>
516
517<div class="doc_code">
518<pre>
519 New = CmpInst::Create(<i>...</i>, SO->getName() + ".cmp");
520</pre>
521</div>
522
523<p>The <tt>Twine</tt> class is effectively a
524lightweight <a href="http://en.wikipedia.org/wiki/Rope_(computer_science)">rope</a>
525which points to temporary (stack allocated) objects. Twines can be implicitly
526constructed as the result of the plus operator applied to strings (i.e., a C
527strings, an <tt>std::string</tt>, or a <tt>StringRef</tt>). The twine delays the
Dan Gohmancf0c9bc2010-02-25 23:51:27 +0000[diff] [blame]528actual concatenation of strings until it is actually required, at which point
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]529it can be efficiently rendered directly into a character array. This avoids
530unnecessary heap allocation involved in constructing the temporary results of
531string concatenation. See
532"<tt><a href="/doxygen/classllvm_1_1Twine_8h-source.html">llvm/ADT/Twine.h</a></tt>"
Benjamin Kramere15192b2009-08-05 15:42:44 +0000[diff] [blame]533for more information.</p>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]534
535<p>As with a <tt>StringRef</tt>, <tt>Twine</tt> objects point to external memory
536and should almost never be stored or mentioned directly. They are intended
537solely for use when defining a function which should be able to efficiently
538accept concatenated strings.</p>
539
540</div>
541
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]542</div>
Daniel Dunbar6e0d1cb2009-07-25 04:41:11 +0000[diff] [blame]543
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]544<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]545<h3>
Misha Brukman2c122ce2005-11-01 21:12:49 +0000[diff] [blame]546 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]547</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]548
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]549<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]550
551<p>Often when working on your pass you will put a bunch of debugging printouts
552and other code into your pass. After you get it working, you want to remove
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]553it, 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]554across).</p>
555
556<p> Naturally, because of this, you don't want to delete the debug printouts,
557but you don't want them to always be noisy. A standard compromise is to comment
558them out, allowing you to enable them if you need them in the future.</p>
559
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]560<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]561file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
562this problem. Basically, you can put arbitrary code into the argument of the
563<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
564tool) is run with the '<tt>-debug</tt>' command line argument:</p>
565
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]566<div class="doc_code">
567<pre>
Daniel Dunbar06388ae2009-07-25 01:55:32 +0000[diff] [blame]568DEBUG(errs() &lt;&lt; "I am here!\n");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]569</pre>
570</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]571
572<p>Then you can run your pass like this:</p>
573
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]574<div class="doc_code">
575<pre>
576$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]577<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]578$ opt &lt; a.bc &gt; /dev/null -mypass -debug
579I am here!
580</pre>
581</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]582
583<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
584to not have to create "yet another" command line option for the debug output for
585your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
586so they do not cause a performance impact at all (for the same reason, they
587should also not contain side-effects!).</p>
588
589<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
590enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
591"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
592program hasn't been started yet, you can always just run it with
593<tt>-debug</tt>.</p>
594
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]595<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]596<h4>
Chris Lattnerc9151082005-04-26 22:57:07 +0000[diff] [blame]597 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]598 the <tt>-debug-only</tt> option</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]599</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]600
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]601<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]602
603<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
604just turns on <b>too much</b> information (such as when working on the code
605generator). If you want to enable debug information with more fine-grained
606control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
607option as follows:</p>
608
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]609<div class="doc_code">
610<pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]611#undef DEBUG_TYPE
Daniel Dunbar06388ae2009-07-25 01:55:32 +0000[diff] [blame]612DEBUG(errs() &lt;&lt; "No debug type\n");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]613#define DEBUG_TYPE "foo"
Daniel Dunbar06388ae2009-07-25 01:55:32 +0000[diff] [blame]614DEBUG(errs() &lt;&lt; "'foo' debug type\n");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]615#undef DEBUG_TYPE
616#define DEBUG_TYPE "bar"
Daniel Dunbar06388ae2009-07-25 01:55:32 +0000[diff] [blame]617DEBUG(errs() &lt;&lt; "'bar' debug type\n"));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]618#undef DEBUG_TYPE
619#define DEBUG_TYPE ""
Daniel Dunbar06388ae2009-07-25 01:55:32 +0000[diff] [blame]620DEBUG(errs() &lt;&lt; "No debug type (2)\n");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]621</pre>
622</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]623
624<p>Then you can run your pass like this:</p>
625
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]626<div class="doc_code">
627<pre>
628$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]629<i>&lt;no output&gt;</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]630$ opt &lt; a.bc &gt; /dev/null -mypass -debug
631No debug type
632'foo' debug type
633'bar' debug type
634No debug type (2)
635$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=foo
636'foo' debug type
637$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=bar
638'bar' debug type
639</pre>
640</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]641
642<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
643a 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]644you <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]645<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
646"bar", because there is no system in place to ensure that names do not
647conflict. If two different modules use the same string, they will all be turned
648on when the name is specified. This allows, for example, all debug information
649for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]650even if the source lives in multiple files.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]651
Daniel Dunbarc3c92392009-08-07 23:48:59 +0000[diff] [blame]652<p>The <tt>DEBUG_WITH_TYPE</tt> macro is also available for situations where you
653would like to set <tt>DEBUG_TYPE</tt>, but only for one specific <tt>DEBUG</tt>
654statement. It takes an additional first parameter, which is the type to use. For
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]655example, the preceding example could be written as:</p>
Daniel Dunbarc3c92392009-08-07 23:48:59 +0000[diff] [blame]656
657
658<div class="doc_code">
659<pre>
660DEBUG_WITH_TYPE("", errs() &lt;&lt; "No debug type\n");
661DEBUG_WITH_TYPE("foo", errs() &lt;&lt; "'foo' debug type\n");
662DEBUG_WITH_TYPE("bar", errs() &lt;&lt; "'bar' debug type\n"));
663DEBUG_WITH_TYPE("", errs() &lt;&lt; "No debug type (2)\n");
664</pre>
665</div>
666
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]667</div>
668
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]669</div>
670
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]671<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]672<h3>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]673 <a name="Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]674 option</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]675</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]676
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]677<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]678
679<p>The "<tt><a
Chris Lattner695b78b2005-04-26 22:56:16 +0000[diff] [blame]680href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]681provides a class named <tt>Statistic</tt> that is used as a unified way to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]682keep track of what the LLVM compiler is doing and how effective various
683optimizations are. It is useful to see what optimizations are contributing to
684making a particular program run faster.</p>
685
686<p>Often you may run your pass on some big program, and you're interested to see
687how many times it makes a certain transformation. Although you can do this with
688hand 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]689for big programs. Using the <tt>Statistic</tt> class makes it very easy to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]690keep track of this information, and the calculated information is presented in a
691uniform manner with the rest of the passes being executed.</p>
692
693<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
694it are as follows:</p>
695
696<ol>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]697 <li><p>Define your statistic like this:</p>
698
699<div class="doc_code">
700<pre>
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]701#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
702STATISTIC(NumXForms, "The # of times I did stuff");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]703</pre>
704</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]705
Chris Lattner0be6fdf2006-12-19 21:46:21 +0000[diff] [blame]706 <p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
707 specified by the first argument. The pass name is taken from the DEBUG_TYPE
708 macro, and the description is taken from the second argument. The variable
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]709 defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]710
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]711 <li><p>Whenever you make a transformation, bump the counter:</p>
712
713<div class="doc_code">
714<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]715++NumXForms; // <i>I did stuff!</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]716</pre>
717</div>
718
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]719 </li>
720 </ol>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]721
722 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
723 statistics gathered, use the '<tt>-stats</tt>' option:</p>
724
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]725<div class="doc_code">
726<pre>
727$ opt -stats -mypassname &lt; program.bc &gt; /dev/null
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]728<i>... statistics output ...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]729</pre>
730</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]731
Reid Spencer6b6c73e2007-02-09 16:00:28 +0000[diff] [blame]732 <p> When running <tt>opt</tt> on a C file from the SPEC benchmark
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]733suite, it gives a report that looks like this:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]734
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]735<div class="doc_code">
736<pre>
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]737 7646 bitcodewriter - Number of normal instructions
738 725 bitcodewriter - Number of oversized instructions
739 129996 bitcodewriter - Number of bitcode bytes written
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]740 2817 raise - Number of insts DCEd or constprop'd
741 3213 raise - Number of cast-of-self removed
742 5046 raise - Number of expression trees converted
743 75 raise - Number of other getelementptr's formed
744 138 raise - Number of load/store peepholes
745 42 deadtypeelim - Number of unused typenames removed from symtab
746 392 funcresolve - Number of varargs functions resolved
747 27 globaldce - Number of global variables removed
748 2 adce - Number of basic blocks removed
749 134 cee - Number of branches revectored
750 49 cee - Number of setcc instruction eliminated
751 532 gcse - Number of loads removed
752 2919 gcse - Number of instructions removed
753 86 indvars - Number of canonical indvars added
754 87 indvars - Number of aux indvars removed
755 25 instcombine - Number of dead inst eliminate
756 434 instcombine - Number of insts combined
757 248 licm - Number of load insts hoisted
758 1298 licm - Number of insts hoisted to a loop pre-header
759 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
760 75 mem2reg - Number of alloca's promoted
761 1444 cfgsimplify - Number of blocks simplified
762</pre>
763</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]764
765<p>Obviously, with so many optimizations, having a unified framework for this
766stuff is very nice. Making your pass fit well into the framework makes it more
767maintainable and useful.</p>
768
769</div>
770
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]771<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]772<h3>
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]773 <a name="ViewGraph">Viewing graphs while debugging code</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]774</h3>
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]775
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]776<div>
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]777
778<p>Several of the important data structures in LLVM are graphs: for example
779CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
780LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
781<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
782DAGs</a>. In many cases, while debugging various parts of the compiler, it is
783nice to instantly visualize these graphs.</p>
784
785<p>LLVM provides several callbacks that are available in a debug build to do
786exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
787the current LLVM tool will pop up a window containing the CFG for the function
788where each basic block is a node in the graph, and each node contains the
789instructions in the block. Similarly, there also exists
790<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
791<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
792and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]793you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]794up a window. Alternatively, you can sprinkle calls to these functions in your
795code in places you want to debug.</p>
796
797<p>Getting this to work requires a small amount of configuration. On Unix
798systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
799toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
800Mac OS/X, download and install the Mac OS/X <a
801href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]802<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]803it) to your path. Once in your system and path are set up, rerun the LLVM
804configure script and rebuild LLVM to enable this functionality.</p>
805
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]806<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
807<i>interesting</i> nodes in large complex graphs. From gdb, if you
808<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]809next <tt>call DAG.viewGraph()</tt> would highlight the node in the
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]810specified color (choices of colors can be found at <a
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]811href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
Jim Laskey543a0ee2006-10-02 12:28:07 +0000[diff] [blame]812complex node attributes can be provided with <tt>call
813DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
814found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
815Attributes</a>.) If you want to restart and clear all the current graph
816attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
817
Chris Lattner83f94672011-06-13 15:59:35 +0000[diff] [blame]818<p>Note that graph visualization features are compiled out of Release builds
819to reduce file size. This means that you need a Debug+Asserts or
820Release+Asserts build to use these features.</p>
821
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]822</div>
823
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]824</div>
825
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]826<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]827<h2>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]828 <a name="datastructure">Picking the Right Data Structure for a Task</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]829</h2>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]830<!-- *********************************************************************** -->
831
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]832<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]833
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]834<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
835 and we commonly use STL data structures. This section describes the trade-offs
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]836 you should consider when you pick one.</p>
837
838<p>
839The first step is a choose your own adventure: do you want a sequential
840container, a set-like container, or a map-like container? The most important
841thing when choosing a container is the algorithmic properties of how you plan to
842access the container. Based on that, you should use:</p>
843
844<ul>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]845<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]846 of an value based on another value. Map-like containers also support
847 efficient queries for containment (whether a key is in the map). Map-like
848 containers generally do not support efficient reverse mapping (values to
849 keys). If you need that, use two maps. Some map-like containers also
850 support efficient iteration through the keys in sorted order. Map-like
851 containers are the most expensive sort, only use them if you need one of
852 these capabilities.</li>
853
854<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
855 stuff into a container that automatically eliminates duplicates. Some
856 set-like containers support efficient iteration through the elements in
857 sorted order. Set-like containers are more expensive than sequential
858 containers.
859</li>
860
861<li>a <a href="#ds_sequential">sequential</a> container provides
862 the most efficient way to add elements and keeps track of the order they are
863 added to the collection. They permit duplicates and support efficient
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]864 iteration, but do not support efficient look-up based on a key.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]865</li>
866
Chris Lattnerdced9fb2009-07-25 07:22:20 +0000[diff] [blame]867<li>a <a href="#ds_string">string</a> container is a specialized sequential
868 container or reference structure that is used for character or byte
869 arrays.</li>
870
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]871<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
872 perform set operations on sets of numeric id's, while automatically
873 eliminating duplicates. Bit containers require a maximum of 1 bit for each
874 identifier you want to store.
875</li>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]876</ul>
877
878<p>
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]879Once the proper category of container is determined, you can fine tune the
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]880memory use, constant factors, and cache behaviors of access by intelligently
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]881picking a member of the category. Note that constant factors and cache behavior
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]882can be a big deal. If you have a vector that usually only contains a few
883elements (but could contain many), for example, it's much better to use
884<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
885. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
886cost of adding the elements to the container. </p>
887
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]888<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]889<h3>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]890 <a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]891</h3>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]892
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]893<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]894There are a variety of sequential containers available for you, based on your
895needs. Pick the first in this section that will do what you want.
Chris Lattner3b4f4172011-07-22 21:34:12 +0000[diff] [blame]896
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]897<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]898<h4>
Chris Lattner8ae42612011-04-05 23:18:20 +0000[diff] [blame]899 <a name="dss_arrayref">llvm/ADT/ArrayRef.h</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]900</h4>
Chris Lattner8ae42612011-04-05 23:18:20 +0000[diff] [blame]901
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]902<div>
Chris Lattner8ae42612011-04-05 23:18:20 +0000[diff] [blame]903<p>The llvm::ArrayRef class is the preferred class to use in an interface that
904 accepts a sequential list of elements in memory and just reads from them. By
905 taking an ArrayRef, the API can be passed a fixed size array, an std::vector,
906 an llvm::SmallVector and anything else that is contiguous in memory.
907</p>
908</div>
909
910
911
912<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]913<h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]914 <a name="dss_fixedarrays">Fixed Size Arrays</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]915</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]916
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]917<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]918<p>Fixed size arrays are very simple and very fast. They are good if you know
919exactly how many elements you have, or you have a (low) upper bound on how many
920you have.</p>
921</div>
922
923<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]924<h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]925 <a name="dss_heaparrays">Heap Allocated Arrays</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]926</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]927
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]928<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]929<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
930the number of elements is variable, if you know how many elements you will need
931before the array is allocated, and if the array is usually large (if not,
932consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
933allocated array is the cost of the new/delete (aka malloc/free). Also note that
934if you are allocating an array of a type with a constructor, the constructor and
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]935destructors will be run for every element in the array (re-sizable vectors only
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]936construct those elements actually used).</p>
937</div>
938
939<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]940<h4>
Chris Lattner9d69d4a2011-07-18 01:40:02 +0000[diff] [blame]941 <a name="dss_tinyptrvector">"llvm/ADT/TinyPtrVector.h"</a>
942</h4>
943
944
945<div>
946<p><tt>TinyPtrVector&lt;Type&gt;</tt> is a highly specialized collection class
947that is optimized to avoid allocation in the case when a vector has zero or one
948elements. It has two major restrictions: 1) it can only hold values of pointer
949type, and 2) it cannot hold a null pointer.</p>
950
951<p>Since this container is highly specialized, it is rarely used.</p>
952
953</div>
954
Chris Lattner9d69d4a2011-07-18 01:40:02 +0000[diff] [blame]955<!-- _______________________________________________________________________ -->
956<h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]957 <a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]958</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]959
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]960<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]961<p><tt>SmallVector&lt;Type, N&gt;</tt> is a simple class that looks and smells
962just like <tt>vector&lt;Type&gt;</tt>:
963it supports efficient iteration, lays out elements in memory order (so you can
964do pointer arithmetic between elements), supports efficient push_back/pop_back
965operations, supports efficient random access to its elements, etc.</p>
966
967<p>The advantage of SmallVector is that it allocates space for
968some number of elements (N) <b>in the object itself</b>. Because of this, if
969the SmallVector is dynamically smaller than N, no malloc is performed. This can
970be a big win in cases where the malloc/free call is far more expensive than the
971code that fiddles around with the elements.</p>
972
973<p>This is good for vectors that are "usually small" (e.g. the number of
974predecessors/successors of a block is usually less than 8). On the other hand,
975this makes the size of the SmallVector itself large, so you don't want to
976allocate lots of them (doing so will waste a lot of space). As such,
977SmallVectors are most useful when on the stack.</p>
978
979<p>SmallVector also provides a nice portable and efficient replacement for
980<tt>alloca</tt>.</p>
981
982</div>
983
984<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]985<h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]986 <a name="dss_vector">&lt;vector&gt;</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]987</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]988
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]989<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]990<p>
991std::vector is well loved and respected. It is useful when SmallVector isn't:
992when the size of the vector is often large (thus the small optimization will
993rarely be a benefit) or if you will be allocating many instances of the vector
994itself (which would waste space for elements that aren't in the container).
995vector is also useful when interfacing with code that expects vectors :).
996</p>
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]997
998<p>One worthwhile note about std::vector: avoid code like this:</p>
999
1000<div class="doc_code">
1001<pre>
1002for ( ... ) {
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]1003 std::vector&lt;foo&gt; V;
Jim Grosbach087f0502011-10-28 20:52:20 +0000[diff] [blame]1004 // make use of V.
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]1005}
1006</pre>
1007</div>
1008
1009<p>Instead, write this as:</p>
1010
1011<div class="doc_code">
1012<pre>
Chris Lattner9bb3dbb2007-03-28 18:27:57 +0000[diff] [blame]1013std::vector&lt;foo&gt; V;
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]1014for ( ... ) {
Jim Grosbach087f0502011-10-28 20:52:20 +0000[diff] [blame]1015 // make use of V.
Chris Lattner32d84762007-02-05 06:30:51 +0000[diff] [blame]1016 V.clear();
1017}
1018</pre>
1019</div>
1020
1021<p>Doing so will save (at least) one heap allocation and free per iteration of
1022the loop.</p>
1023
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1024</div>
1025
1026<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1027<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1028 <a name="dss_deque">&lt;deque&gt;</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1029</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1030
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1031<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1032<p>std::deque is, in some senses, a generalized version of std::vector. Like
1033std::vector, it provides constant time random access and other similar
1034properties, but it also provides efficient access to the front of the list. It
1035does not guarantee continuity of elements within memory.</p>
1036
1037<p>In exchange for this extra flexibility, std::deque has significantly higher
1038constant factor costs than std::vector. If possible, use std::vector or
1039something cheaper.</p>
1040</div>
1041
1042<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1043<h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1044 <a name="dss_list">&lt;list&gt;</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1045</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1046
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1047<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1048<p>std::list is an extremely inefficient class that is rarely useful.
1049It performs a heap allocation for every element inserted into it, thus having an
1050extremely high constant factor, particularly for small data types. std::list
1051also only supports bidirectional iteration, not random access iteration.</p>
1052
1053<p>In exchange for this high cost, std::list supports efficient access to both
1054ends of the list (like std::deque, but unlike std::vector or SmallVector). In
1055addition, the iterator invalidation characteristics of std::list are stronger
1056than that of a vector class: inserting or removing an element into the list does
1057not invalidate iterator or pointers to other elements in the list.</p>
1058</div>
1059
1060<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1061<h4>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1062 <a name="dss_ilist">llvm/ADT/ilist.h</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1063</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1064
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1065<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1066<p><tt>ilist&lt;T&gt;</tt> implements an 'intrusive' doubly-linked list. It is
1067intrusive, because it requires the element to store and provide access to the
1068prev/next pointers for the list.</p>
1069
Gabor Greif2946d1c2009-02-27 12:02:19 +0000[diff] [blame]1070<p><tt>ilist</tt> has the same drawbacks as <tt>std::list</tt>, and additionally
1071requires an <tt>ilist_traits</tt> implementation for the element type, but it
1072provides some novel characteristics. In particular, it can efficiently store
1073polymorphic objects, the traits class is informed when an element is inserted or
Gabor Greif0cbcabe2009-03-12 09:47:03 +0000[diff] [blame]1074removed from the list, and <tt>ilist</tt>s are guaranteed to support a
1075constant-time splice operation.</p>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1076
Gabor Greif0cbcabe2009-03-12 09:47:03 +0000[diff] [blame]1077<p>These properties are exactly what we want for things like
1078<tt>Instruction</tt>s and basic blocks, which is why these are implemented with
1079<tt>ilist</tt>s.</p>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1080
1081Related classes of interest are explained in the following subsections:
1082 <ul>
Gabor Greif01862502009-02-27 13:28:07 +0000[diff] [blame]1083 <li><a href="#dss_ilist_traits">ilist_traits</a></li>
Gabor Greif2946d1c2009-02-27 12:02:19 +0000[diff] [blame]1084 <li><a href="#dss_iplist">iplist</a></li>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1085 <li><a href="#dss_ilist_node">llvm/ADT/ilist_node.h</a></li>
Gabor Greif6a65f422009-03-12 10:30:31 +0000[diff] [blame]1086 <li><a href="#dss_ilist_sentinel">Sentinels</a></li>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1087 </ul>
1088</div>
1089
1090<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1091<h4>
Argyrios Kyrtzidis81df0892011-06-15 19:56:01 +0000[diff] [blame]1092 <a name="dss_packedvector">llvm/ADT/PackedVector.h</a>
1093</h4>
1094
1095<div>
1096<p>
1097Useful for storing a vector of values using only a few number of bits for each
1098value. Apart from the standard operations of a vector-like container, it can
1099also perform an 'or' set operation.
1100</p>
1101
1102<p>For example:</p>
1103
1104<div class="doc_code">
1105<pre>
1106enum State {
1107 None = 0x0,
1108 FirstCondition = 0x1,
1109 SecondCondition = 0x2,
1110 Both = 0x3
1111};
1112
1113State get() {
1114 PackedVector&lt;State, 2&gt; Vec1;
1115 Vec1.push_back(FirstCondition);
1116
1117 PackedVector&lt;State, 2&gt; Vec2;
1118 Vec2.push_back(SecondCondition);
1119
1120 Vec1 |= Vec2;
1121 return Vec1[0]; // returns 'Both'.
1122}
1123</pre>
1124</div>
1125
1126</div>
1127
1128<!-- _______________________________________________________________________ -->
1129<h4>
Gabor Greif01862502009-02-27 13:28:07 +0000[diff] [blame]1130 <a name="dss_ilist_traits">ilist_traits</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1131</h4>
Gabor Greif01862502009-02-27 13:28:07 +0000[diff] [blame]1132
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1133<div>
Gabor Greif01862502009-02-27 13:28:07 +0000[diff] [blame]1134<p><tt>ilist_traits&lt;T&gt;</tt> is <tt>ilist&lt;T&gt;</tt>'s customization
1135mechanism. <tt>iplist&lt;T&gt;</tt> (and consequently <tt>ilist&lt;T&gt;</tt>)
1136publicly derive from this traits class.</p>
1137</div>
1138
1139<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1140<h4>
Gabor Greif2946d1c2009-02-27 12:02:19 +0000[diff] [blame]1141 <a name="dss_iplist">iplist</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1142</h4>
Gabor Greif2946d1c2009-02-27 12:02:19 +0000[diff] [blame]1143
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1144<div>
Gabor Greif2946d1c2009-02-27 12:02:19 +0000[diff] [blame]1145<p><tt>iplist&lt;T&gt;</tt> is <tt>ilist&lt;T&gt;</tt>'s base and as such
Gabor Greif0cbcabe2009-03-12 09:47:03 +0000[diff] [blame]1146supports a slightly narrower interface. Notably, inserters from
1147<tt>T&amp;</tt> are absent.</p>
Gabor Greif01862502009-02-27 13:28:07 +0000[diff] [blame]1148
1149<p><tt>ilist_traits&lt;T&gt;</tt> is a public base of this class and can be
1150used for a wide variety of customizations.</p>
Gabor Greif2946d1c2009-02-27 12:02:19 +0000[diff] [blame]1151</div>
1152
1153<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1154<h4>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1155 <a name="dss_ilist_node">llvm/ADT/ilist_node.h</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1156</h4>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1157
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1158<div>
Gabor Greif3899e492009-02-27 11:37:41 +0000[diff] [blame]1159<p><tt>ilist_node&lt;T&gt;</tt> implements a the forward and backward links
1160that are expected by the <tt>ilist&lt;T&gt;</tt> (and analogous containers)
1161in the default manner.</p>
1162
1163<p><tt>ilist_node&lt;T&gt;</tt>s are meant to be embedded in the node type
Gabor Greif0cbcabe2009-03-12 09:47:03 +0000[diff] [blame]1164<tt>T</tt>, usually <tt>T</tt> publicly derives from
1165<tt>ilist_node&lt;T&gt;</tt>.</p>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1166</div>
1167
1168<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1169<h4>
Gabor Greif6a65f422009-03-12 10:30:31 +0000[diff] [blame]1170 <a name="dss_ilist_sentinel">Sentinels</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1171</h4>
Gabor Greif6a65f422009-03-12 10:30:31 +0000[diff] [blame]1172
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1173<div>
Dan Gohmancf0c9bc2010-02-25 23:51:27 +0000[diff] [blame]1174<p><tt>ilist</tt>s have another specialty that must be considered. To be a good
Gabor Greif6a65f422009-03-12 10:30:31 +0000[diff] [blame]1175citizen in the C++ ecosystem, it needs to support the standard container
1176operations, such as <tt>begin</tt> and <tt>end</tt> iterators, etc. Also, the
1177<tt>operator--</tt> must work correctly on the <tt>end</tt> iterator in the
1178case of non-empty <tt>ilist</tt>s.</p>
1179
1180<p>The only sensible solution to this problem is to allocate a so-called
1181<i>sentinel</i> along with the intrusive list, which serves as the <tt>end</tt>
1182iterator, providing the back-link to the last element. However conforming to the
1183C++ convention it is illegal to <tt>operator++</tt> beyond the sentinel and it
1184also must not be dereferenced.</p>
1185
1186<p>These constraints allow for some implementation freedom to the <tt>ilist</tt>
1187how to allocate and store the sentinel. The corresponding policy is dictated
1188by <tt>ilist_traits&lt;T&gt;</tt>. By default a <tt>T</tt> gets heap-allocated
1189whenever the need for a sentinel arises.</p>
1190
1191<p>While the default policy is sufficient in most cases, it may break down when
1192<tt>T</tt> does not provide a default constructor. Also, in the case of many
1193instances of <tt>ilist</tt>s, the memory overhead of the associated sentinels
1194is wasted. To alleviate the situation with numerous and voluminous
1195<tt>T</tt>-sentinels, sometimes a trick is employed, leading to <i>ghostly
1196sentinels</i>.</p>
1197
1198<p>Ghostly sentinels are obtained by specially-crafted <tt>ilist_traits&lt;T&gt;</tt>
1199which superpose the sentinel with the <tt>ilist</tt> instance in memory. Pointer
1200arithmetic is used to obtain the sentinel, which is relative to the
1201<tt>ilist</tt>'s <tt>this</tt> pointer. The <tt>ilist</tt> is augmented by an
1202extra pointer, which serves as the back-link of the sentinel. This is the only
1203field in the ghostly sentinel which can be legally accessed.</p>
1204</div>
1205
1206<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1207<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1208 <a name="dss_other">Other Sequential Container options</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1209</h4>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1210
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1211<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1212<p>Other STL containers are available, such as std::string.</p>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1213
1214<p>There are also various STL adapter classes such as std::queue,
1215std::priority_queue, std::stack, etc. These provide simplified access to an
1216underlying container but don't affect the cost of the container itself.</p>
1217
1218</div>
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1219</div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1220
1221<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1222<h3>
Chris Lattner7314a202011-07-22 20:46:49 +0000[diff] [blame]1223 <a name="ds_string">String-like containers</a>
1224</h3>
1225
1226<div>
1227
1228<p>
Chris Lattner3b4f4172011-07-22 21:34:12 +0000[diff] [blame]1229There are a variety of ways to pass around and use strings in C and C++, and
1230LLVM adds a few new options to choose from. Pick the first option on this list
1231that will do what you need, they are ordered according to their relative cost.
1232</p>
1233<p>
1234Note that is is generally preferred to <em>not</em> pass strings around as
1235"<tt>const char*</tt>"'s. These have a number of problems, including the fact
1236that they cannot represent embedded nul ("\0") characters, and do not have a
1237length available efficiently. The general replacement for '<tt>const
1238char*</tt>' is StringRef.
1239</p>
1240
1241<p>For more information on choosing string containers for APIs, please see
1242<a href="#string_apis">Passing strings</a>.</p>
1243
1244
1245<!-- _______________________________________________________________________ -->
1246<h4>
1247 <a name="dss_stringref">llvm/ADT/StringRef.h</a>
1248</h4>
1249
1250<div>
1251<p>
1252The StringRef class is a simple value class that contains a pointer to a
1253character and a length, and is quite related to the <a
1254href="#dss_arrayref">ArrayRef</a> class (but specialized for arrays of
1255characters). Because StringRef carries a length with it, it safely handles
1256strings with embedded nul characters in it, getting the length does not require
1257a strlen call, and it even has very convenient APIs for slicing and dicing the
1258character range that it represents.
1259</p>
1260
1261<p>
1262StringRef is ideal for passing simple strings around that are known to be live,
1263either because they are C string literals, std::string, a C array, or a
1264SmallVector. Each of these cases has an efficient implicit conversion to
1265StringRef, which doesn't result in a dynamic strlen being executed.
1266</p>
1267
1268<p>StringRef has a few major limitations which make more powerful string
1269containers useful:</p>
1270
1271<ol>
1272<li>You cannot directly convert a StringRef to a 'const char*' because there is
1273no way to add a trailing nul (unlike the .c_str() method on various stronger
1274classes).</li>
1275
1276
1277<li>StringRef doesn't own or keep alive the underlying string bytes.
1278As such it can easily lead to dangling pointers, and is not suitable for
1279embedding in datastructures in most cases (instead, use an std::string or
1280something like that).</li>
1281
1282<li>For the same reason, StringRef cannot be used as the return value of a
1283method if the method "computes" the result string. Instead, use
1284std::string.</li>
1285
Chris Lattnerec8f1ea2011-07-23 17:18:57 +0000[diff] [blame]1286<li>StringRef's do not allow you to mutate the pointed-to string bytes and it
1287doesn't allow you to insert or remove bytes from the range. For editing
1288operations like this, it interoperates with the <a
1289href="#dss_twine">Twine</a> class.</li>
Chris Lattner3b4f4172011-07-22 21:34:12 +0000[diff] [blame]1290</ol>
1291
1292<p>Because of its strengths and limitations, it is very common for a function to
1293take a StringRef and for a method on an object to return a StringRef that
1294points into some string that it owns.</p>
1295
1296</div>
1297
1298<!-- _______________________________________________________________________ -->
1299<h4>
1300 <a name="dss_twine">llvm/ADT/Twine.h</a>
1301</h4>
1302
1303<div>
1304 <p>
1305 The Twine class is used as an intermediary datatype for APIs that want to take
1306 a string that can be constructed inline with a series of concatenations.
1307 Twine works by forming recursive instances of the Twine datatype (a simple
1308 value object) on the stack as temporary objects, linking them together into a
1309 tree which is then linearized when the Twine is consumed. Twine is only safe
1310 to use as the argument to a function, and should always be a const reference,
1311 e.g.:
1312 </p>
1313
1314 <pre>
1315 void foo(const Twine &amp;T);
1316 ...
1317 StringRef X = ...
1318 unsigned i = ...
1319 foo(X + "." + Twine(i));
1320 </pre>
1321
1322 <p>This example forms a string like "blarg.42" by concatenating the values
1323 together, and does not form intermediate strings containing "blarg" or
1324 "blarg.".
1325 </p>
1326
1327 <p>Because Twine is constructed with temporary objects on the stack, and
1328 because these instances are destroyed at the end of the current statement,
1329 it is an inherently dangerous API. For example, this simple variant contains
1330 undefined behavior and will probably crash:</p>
1331
1332 <pre>
1333 void foo(const Twine &amp;T);
1334 ...
1335 StringRef X = ...
1336 unsigned i = ...
1337 const Twine &amp;Tmp = X + "." + Twine(i);
1338 foo(Tmp);
1339 </pre>
1340
1341 <p>... because the temporaries are destroyed before the call. That said,
1342 Twine's are much more efficient than intermediate std::string temporaries, and
1343 they work really well with StringRef. Just be aware of their limitations.</p>
1344
1345</div>
1346
1347
1348<!-- _______________________________________________________________________ -->
1349<h4>
1350 <a name="dss_smallstring">llvm/ADT/SmallString.h</a>
1351</h4>
1352
1353<div>
1354
1355<p>SmallString is a subclass of <a href="#dss_smallvector">SmallVector</a> that
1356adds some convenience APIs like += that takes StringRef's. SmallString avoids
1357allocating memory in the case when the preallocated space is enough to hold its
1358data, and it calls back to general heap allocation when required. Since it owns
1359its data, it is very safe to use and supports full mutation of the string.</p>
1360
1361<p>Like SmallVector's, the big downside to SmallString is their sizeof. While
1362they are optimized for small strings, they themselves are not particularly
1363small. This means that they work great for temporary scratch buffers on the
1364stack, but should not generally be put into the heap: it is very rare to
1365see a SmallString as the member of a frequently-allocated heap data structure
1366or returned by-value.
Chris Lattner7314a202011-07-22 20:46:49 +0000[diff] [blame]1367</p>
1368
1369</div>
Chris Lattner3b4f4172011-07-22 21:34:12 +0000[diff] [blame]1370
1371<!-- _______________________________________________________________________ -->
1372<h4>
1373 <a name="dss_stdstring">std::string</a>
1374</h4>
1375
1376<div>
1377
1378 <p>The standard C++ std::string class is a very general class that (like
1379 SmallString) owns its underlying data. sizeof(std::string) is very reasonable
1380 so it can be embedded into heap data structures and returned by-value.
1381 On the other hand, std::string is highly inefficient for inline editing (e.g.
1382 concatenating a bunch of stuff together) and because it is provided by the
1383 standard library, its performance characteristics depend a lot of the host
1384 standard library (e.g. libc++ and MSVC provide a highly optimized string
1385 class, GCC contains a really slow implementation).
1386 </p>
1387
1388 <p>The major disadvantage of std::string is that almost every operation that
1389 makes them larger can allocate memory, which is slow. As such, it is better
1390 to use SmallVector or Twine as a scratch buffer, but then use std::string to
1391 persist the result.</p>
1392
1393
1394</div>
1395
1396<!-- end of strings -->
1397</div>
1398
Chris Lattner7314a202011-07-22 20:46:49 +0000[diff] [blame]1399
1400<!-- ======================================================================= -->
1401<h3>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1402 <a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1403</h3>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1404
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1405<div>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1406
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1407<p>Set-like containers are useful when you need to canonicalize multiple values
1408into a single representation. There are several different choices for how to do
1409this, providing various trade-offs.</p>
1410
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1411<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1412<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1413 <a name="dss_sortedvectorset">A sorted 'vector'</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1414</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1415
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1416<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1417
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]1418<p>If you intend to insert a lot of elements, then do a lot of queries, a
1419great approach is to use a vector (or other sequential container) with
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1420std::sort+std::unique to remove duplicates. This approach works really well if
Chris Lattner3b23a8c2007-02-03 08:10:45 +0000[diff] [blame]1421your usage pattern has these two distinct phases (insert then query), and can be
1422coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
1423</p>
1424
1425<p>
1426This combination provides the several nice properties: the result data is
1427contiguous in memory (good for cache locality), has few allocations, is easy to
1428address (iterators in the final vector are just indices or pointers), and can be
1429efficiently queried with a standard binary or radix search.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1430
1431</div>
1432
1433<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1434<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1435 <a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1436</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1437
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1438<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1439
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]1440<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]1441are 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]1442has space for N elements in place (thus, if the set is dynamically smaller than
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1443N, no malloc traffic is required) and accesses them with a simple linear search.
1444When the set grows beyond 'N' elements, it allocates a more expensive representation that
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1445guarantees efficient access (for most types, it falls back to std::set, but for
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1446pointers it uses something far better, <a
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1447href="#dss_smallptrset">SmallPtrSet</a>).</p>
1448
1449<p>The magic of this class is that it handles small sets extremely efficiently,
1450but gracefully handles extremely large sets without loss of efficiency. The
1451drawback is that the interface is quite small: it supports insertion, queries
1452and erasing, but does not support iteration.</p>
1453
1454</div>
1455
1456<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1457<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1458 <a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1459</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1460
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1461<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1462
Gabor Greif4de73682010-03-26 19:30:47 +0000[diff] [blame]1463<p>SmallPtrSet has all the advantages of <tt>SmallSet</tt> (and a <tt>SmallSet</tt> of pointers is
1464transparently implemented with a <tt>SmallPtrSet</tt>), but also supports iterators. If
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1465more than 'N' insertions are performed, a single quadratically
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1466probed hash table is allocated and grows as needed, providing extremely
1467efficient access (constant time insertion/deleting/queries with low constant
1468factors) and is very stingy with malloc traffic.</p>
1469
Gabor Greif4de73682010-03-26 19:30:47 +0000[diff] [blame]1470<p>Note that, unlike <tt>std::set</tt>, the iterators of <tt>SmallPtrSet</tt> are invalidated
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1471whenever an insertion occurs. Also, the values visited by the iterators are not
1472visited in sorted order.</p>
1473
1474</div>
1475
1476<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1477<h4>
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]1478 <a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1479</h4>
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]1480
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1481<div>
Chris Lattnerc28476f2007-09-30 00:58:59 +0000[diff] [blame]1482
1483<p>
1484DenseSet is a simple quadratically probed hash table. It excels at supporting
1485small values: it uses a single allocation to hold all of the pairs that
1486are currently inserted in the set. DenseSet is a great way to unique small
1487values that are not simple pointers (use <a
1488href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
1489the same requirements for the value type that <a
1490href="#dss_densemap">DenseMap</a> has.
1491</p>
1492
1493</div>
1494
1495<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1496<h4>
Jakob Stoklund Olesen62588622012-02-22 00:56:08 +0000[diff] [blame]1497 <a name="dss_sparseset">"llvm/ADT/SparseSet.h"</a>
1498</h4>
1499
1500<div>
1501
1502<p>SparseSet holds a small number of objects identified by unsigned keys of
1503moderate size. It uses a lot of memory, but provides operations that are
1504almost as fast as a vector. Typical keys are physical registers, virtual
1505registers, or numbered basic blocks.</p>
1506
1507<p>SparseSet is useful for algorithms that need very fast clear/find/insert/erase
1508and fast iteration over small sets. It is not intended for building composite
1509data structures.</p>
1510
1511</div>
1512
1513<!-- _______________________________________________________________________ -->
1514<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1515 <a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1516</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1517
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1518<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1519
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1520<p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1521FoldingSet is an aggregate class that is really good at uniquing
1522expensive-to-create or polymorphic objects. It is a combination of a chained
1523hash table with intrusive links (uniqued objects are required to inherit from
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1524FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
1525its ID process.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1526
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1527<p>Consider a case where you want to implement a "getOrCreateFoo" method for
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1528a complex object (for example, a node in the code generator). The client has a
1529description of *what* it wants to generate (it knows the opcode and all the
1530operands), 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]1531only to find out it already exists, at which point we would have to delete it
1532and return the node that already exists.
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1533</p>
1534
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1535<p>To support this style of client, FoldingSet perform a query with a
1536FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1537element that we want to query for. The query either returns the element
1538matching the ID or it returns an opaque ID that indicates where insertion should
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1539take place. Construction of the ID usually does not require heap traffic.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1540
1541<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
1542in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1543Because the elements are individually allocated, pointers to the elements are
1544stable: inserting or removing elements does not invalidate any pointers to other
1545elements.
1546</p>
1547
1548</div>
1549
1550<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1551<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1552 <a name="dss_set">&lt;set&gt;</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1553</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1554
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1555<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1556
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1557<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
1558many things but great at nothing. std::set allocates memory for each element
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1559inserted (thus it is very malloc intensive) and typically stores three pointers
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1560per element in the set (thus adding a large amount of per-element space
1561overhead). It offers guaranteed log(n) performance, which is not particularly
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1562fast from a complexity standpoint (particularly if the elements of the set are
1563expensive to compare, like strings), and has extremely high constant factors for
1564lookup, insertion and removal.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1565
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1566<p>The advantages of std::set are that its iterators are stable (deleting or
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1567inserting an element from the set does not affect iterators or pointers to other
1568elements) and that iteration over the set is guaranteed to be in sorted order.
1569If the elements in the set are large, then the relative overhead of the pointers
1570and malloc traffic is not a big deal, but if the elements of the set are small,
1571std::set is almost never a good choice.</p>
1572
1573</div>
1574
1575<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1576<h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1577 <a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1578</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1579
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1580<div>
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1581<p>LLVM's SetVector&lt;Type&gt; is an adapter class that combines your choice of
1582a set-like container along with a <a href="#ds_sequential">Sequential
1583Container</a>. The important property
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1584that this provides is efficient insertion with uniquing (duplicate elements are
1585ignored) with iteration support. It implements this by inserting elements into
1586both a set-like container and the sequential container, using the set-like
1587container for uniquing and the sequential container for iteration.
1588</p>
1589
1590<p>The difference between SetVector and other sets is that the order of
1591iteration is guaranteed to match the order of insertion into the SetVector.
1592This property is really important for things like sets of pointers. Because
1593pointer values are non-deterministic (e.g. vary across runs of the program on
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1594different machines), iterating over the pointers in the set will
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1595not be in a well-defined order.</p>
1596
1597<p>
1598The drawback of SetVector is that it requires twice as much space as a normal
1599set and has the sum of constant factors from the set-like container and the
1600sequential container that it uses. Use it *only* if you need to iterate over
1601the elements in a deterministic order. SetVector is also expensive to delete
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1602elements out of (linear time), unless you use it's "pop_back" method, which is
1603faster.
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1604</p>
1605
Bill Wendling34781732011-10-11 06:33:56 +0000[diff] [blame]1606<p><tt>SetVector</tt> is an adapter class that defaults to
1607 using <tt>std::vector</tt> and a size 16 <tt>SmallSet</tt> for the underlying
1608 containers, so it is quite expensive. However,
1609 <tt>"llvm/ADT/SetVector.h"</tt> also provides a <tt>SmallSetVector</tt>
1610 class, which defaults to using a <tt>SmallVector</tt> and <tt>SmallSet</tt>
1611 of a specified size. If you use this, and if your sets are dynamically
1612 smaller than <tt>N</tt>, you will save a lot of heap traffic.</p>
Chris Lattneredca3c52007-02-04 00:00:26 +0000[diff] [blame]1613
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1614</div>
1615
1616<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1617<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1618 <a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1619</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1620
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1621<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1622
1623<p>
1624UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
1625retains a unique ID for each element inserted into the set. It internally
1626contains a map and a vector, and it assigns a unique ID for each value inserted
1627into the set.</p>
1628
1629<p>UniqueVector is very expensive: its cost is the sum of the cost of
1630maintaining both the map and vector, it has high complexity, high constant
1631factors, and produces a lot of malloc traffic. It should be avoided.</p>
1632
1633</div>
1634
Chris Lattner2fdd0052011-11-15 22:40:14 +0000[diff] [blame]1635<!-- _______________________________________________________________________ -->
1636<h4>
1637 <a name="dss_immutableset">"llvm/ADT/ImmutableSet.h"</a>
1638</h4>
1639
1640<div>
1641
1642<p>
1643ImmutableSet is an immutable (functional) set implementation based on an AVL
1644tree.
1645Adding or removing elements is done through a Factory object and results in the
1646creation of a new ImmutableSet object.
1647If an ImmutableSet already exists with the given contents, then the existing one
1648is returned; equality is compared with a FoldingSetNodeID.
1649The time and space complexity of add or remove operations is logarithmic in the
1650size of the original set.
1651
1652<p>
1653There is no method for returning an element of the set, you can only check for
1654membership.
1655
1656</div>
1657
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1658
1659<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1660<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1661 <a name="dss_otherset">Other Set-Like Container Options</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1662</h4>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1663
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1664<div>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1665
1666<p>
1667The STL provides several other options, such as std::multiset and the various
Chris Lattnerf1a30822009-03-09 05:20:45 +0000[diff] [blame]1668"hash_set" like containers (whether from C++ TR1 or from the SGI library). We
1669never use hash_set and unordered_set because they are generally very expensive
1670(each insertion requires a malloc) and very non-portable.
1671</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1672
1673<p>std::multiset is useful if you're not interested in elimination of
Chris Lattner14868db2007-02-03 08:20:15 +0000[diff] [blame]1674duplicates, but has all the drawbacks of std::set. A sorted vector (where you
1675don't delete duplicate entries) or some other approach is almost always
1676better.</p>
Chris Lattner74c4ca12007-02-03 07:59:07 +0000[diff] [blame]1677
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1678</div>
1679
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1680</div>
1681
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1682<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1683<h3>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1684 <a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1685</h3>
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1686
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1687<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1688Map-like containers are useful when you want to associate data to a key. As
1689usual, there are a lot of different ways to do this. :)
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1690
1691<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1692<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1693 <a name="dss_sortedvectormap">A sorted 'vector'</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1694</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1695
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1696<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1697
1698<p>
1699If your usage pattern follows a strict insert-then-query approach, you can
1700trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
1701for set-like containers</a>. The only difference is that your query function
1702(which uses std::lower_bound to get efficient log(n) lookup) should only compare
1703the key, not both the key and value. This yields the same advantages as sorted
1704vectors for sets.
1705</p>
1706</div>
1707
1708<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1709<h4>
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1710 <a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1711</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1712
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1713<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1714
1715<p>
1716Strings are commonly used as keys in maps, and they are difficult to support
1717efficiently: they are variable length, inefficient to hash and compare when
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1718long, expensive to copy, etc. StringMap is a specialized container designed to
1719cope with these issues. It supports mapping an arbitrary range of bytes to an
1720arbitrary other object.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1721
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1722<p>The StringMap implementation uses a quadratically-probed hash table, where
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1723the buckets store a pointer to the heap allocated entries (and some other
1724stuff). The entries in the map must be heap allocated because the strings are
1725variable length. The string data (key) and the element object (value) are
1726stored in the same allocation with the string data immediately after the element
1727object. This container guarantees the "<tt>(char*)(&amp;Value+1)</tt>" points
1728to the key string for a value.</p>
1729
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1730<p>The StringMap is very fast for several reasons: quadratic probing is very
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1731cache efficient for lookups, the hash value of strings in buckets is not
Nick Lewycky2a80aca2010-08-01 23:18:45 +0000[diff] [blame]1732recomputed when looking up an element, StringMap rarely has to touch the
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1733memory for unrelated objects when looking up a value (even when hash collisions
1734happen), hash table growth does not recompute the hash values for strings
1735already in the table, and each pair in the map is store in a single allocation
1736(the string data is stored in the same allocation as the Value of a pair).</p>
1737
Chris Lattner796f9fa2007-02-08 19:14:21 +0000[diff] [blame]1738<p>StringMap also provides query methods that take byte ranges, so it only ever
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1739copies a string if a value is inserted into the table.</p>
Jim Grosbach838b7a32012-04-18 20:28:55 +0000[diff] [blame]1740
1741<p>StringMap iteratation order, however, is not guaranteed to be deterministic,
1742so any uses which require that should instead use a std::map.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1743</div>
1744
1745<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1746<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1747 <a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1748</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1749
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1750<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1751<p>
1752IndexedMap is a specialized container for mapping small dense integers (or
1753values that can be mapped to small dense integers) to some other type. It is
1754internally implemented as a vector with a mapping function that maps the keys to
1755the dense integer range.
1756</p>
1757
1758<p>
1759This is useful for cases like virtual registers in the LLVM code generator: they
1760have a dense mapping that is offset by a compile-time constant (the first
1761virtual register ID).</p>
1762
1763</div>
1764
1765<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1766<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1767 <a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1768</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1769
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1770<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1771
1772<p>
1773DenseMap is a simple quadratically probed hash table. It excels at supporting
1774small keys and values: it uses a single allocation to hold all of the pairs that
1775are currently inserted in the map. DenseMap is a great way to map pointers to
1776pointers, or map other small types to each other.
1777</p>
1778
1779<p>
1780There are several aspects of DenseMap that you should be aware of, however. The
Talinbabd5982012-01-30 06:55:43 +0000[diff] [blame]1781iterators in a DenseMap are invalidated whenever an insertion occurs, unlike
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1782map. Also, because DenseMap allocates space for a large number of key/value
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1783pairs (it starts with 64 by default), it will waste a lot of space if your keys
1784or values are large. Finally, you must implement a partial specialization of
Chris Lattner76c1b972007-09-17 18:34:04 +0000[diff] [blame]1785DenseMapInfo for the key that you want, if it isn't already supported. This
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1786is required to tell DenseMap about two special marker values (which can never be
Chris Lattnera4a264d2007-02-03 20:17:53 +0000[diff] [blame]1787inserted into the map) that it needs internally.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1788
Talinbabd5982012-01-30 06:55:43 +0000[diff] [blame]1789<p>
1790DenseMap's find_as() method supports lookup operations using an alternate key
1791type. This is useful in cases where the normal key type is expensive to
1792construct, but cheap to compare against. The DenseMapInfo is responsible for
1793defining the appropriate comparison and hashing methods for each alternate
1794key type used.
1795</p>
1796
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1797</div>
1798
1799<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1800<h4>
Jeffrey Yasskin71a5c222009-10-22 22:11:22 +0000[diff] [blame]1801 <a name="dss_valuemap">"llvm/ADT/ValueMap.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1802</h4>
Jeffrey Yasskin71a5c222009-10-22 22:11:22 +0000[diff] [blame]1803
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1804<div>
Jeffrey Yasskin71a5c222009-10-22 22:11:22 +0000[diff] [blame]1805
1806<p>
1807ValueMap is a wrapper around a <a href="#dss_densemap">DenseMap</a> mapping
1808Value*s (or subclasses) to another type. When a Value is deleted or RAUW'ed,
1809ValueMap will update itself so the new version of the key is mapped to the same
1810value, just as if the key were a WeakVH. You can configure exactly how this
1811happens, and what else happens on these two events, by passing
1812a <code>Config</code> parameter to the ValueMap template.</p>
1813
1814</div>
1815
1816<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1817<h4>
Benjamin Kramerb856d552012-04-25 18:01:58 +0000[diff] [blame^]1818 <a name="dss_multiimplmap">"llvm/ADT/MultiImplMap.h"</a>
1819</h4>
1820
1821<div>
1822
1823<p>
1824MultiImplMap is map that has two modes, one for small amount of elements and
1825one for big amount. User should set map implementation for both of them.
1826User also should set the maximum possible number of elements for small mode.
1827</p>
1828
1829<p>
1830If user want to use MultiImplMap instead of
1831<a href="#dss_densemap">DenseMap</a>, he should pass template parameter
1832DenseMapCompatible = true. Note, that in this case map implementations
1833should present additional DenseMap specific methods (see below):
1834<code>isPointerIntoBucketsArray</code>, <code>getPointerIntoBucketsArray</code>
1835and <code>FindAndConstruct</code>.
1836</p>
1837
1838<p>
1839Initially MultiImplMap uses small mode and small map implementation. It
1840triggered to the big mode when the number of contained elements exceeds
1841maximum possible elements for small mode.
1842</p>
1843
1844</div>
1845
1846<!-- _______________________________________________________________________ -->
1847<h4>
1848 <a name="dss_flatarraymap">"llvm/ADT/FlatArrayMap.h"</a>
1849</h4>
1850
1851<div>
1852
1853<p>
1854FlatArrayMap optimized for small amount of elements. It uses flat array
1855implementation inside:
1856</p>
1857<pre>[ key0, value0, key1, value1, ... keyN, valueN ]</pre>
1858
1859
1860<p>
1861User should pass key type, mapped type (type of value), and maximum
1862number of elements.
1863</p>
1864
1865<p>
1866After maximum number of elements is reached, map declines any further
1867attempts to insert new elements ("insert" method returns &#60;end(),
1868false&#62;).
1869</p>
1870
1871<p>
1872FlatArrayMap has interface that is compatible with
1873<a href="#dss_densemap">DenseMap</a>, so user can replace it with DenseMap
1874without any code changing and vice versa.
1875</p>
1876
1877</div>
1878
1879<!-- _______________________________________________________________________ -->
1880<h4>
1881 <a name="dss_smallmap">"llvm/ADT/SmallMap.h"</a>
1882</h4>
1883
1884<div>
1885
1886<p>
1887SmallMap is wrapper around <a href="#dss_multiimplmap">MultiImplMap</a>.
1888It uses <a href="#dss_flatarraymap">FlatArrayMap</a> for small mode, and
1889<a href="#dss_densemap">DenseMap</a> for big mode.
1890</p>
1891
1892</div>
1893
1894<!-- _______________________________________________________________________ -->
1895<h4>
Jakob Stoklund Olesenaca0da62010-12-14 00:55:51 +0000[diff] [blame]1896 <a name="dss_intervalmap">"llvm/ADT/IntervalMap.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1897</h4>
Jakob Stoklund Olesenaca0da62010-12-14 00:55:51 +0000[diff] [blame]1898
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1899<div>
Jakob Stoklund Olesenaca0da62010-12-14 00:55:51 +0000[diff] [blame]1900
1901<p> IntervalMap is a compact map for small keys and values. It maps key
1902intervals instead of single keys, and it will automatically coalesce adjacent
1903intervals. When then map only contains a few intervals, they are stored in the
1904map object itself to avoid allocations.</p>
1905
1906<p> The IntervalMap iterators are quite big, so they should not be passed around
1907as STL iterators. The heavyweight iterators allow a smaller data structure.</p>
1908
1909</div>
1910
1911<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1912<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1913 <a name="dss_map">&lt;map&gt;</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1914</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1915
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1916<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1917
1918<p>
1919std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
1920a single allocation per pair inserted into the map, it offers log(n) lookup with
1921an extremely large constant factor, imposes a space penalty of 3 pointers per
1922pair in the map, etc.</p>
1923
1924<p>std::map is most useful when your keys or values are very large, if you need
1925to iterate over the collection in sorted order, or if you need stable iterators
1926into the map (i.e. they don't get invalidated if an insertion or deletion of
1927another element takes place).</p>
1928
1929</div>
1930
1931<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1932<h4>
Jakob Stoklund Olesen4359e5e2011-04-05 20:56:08 +0000[diff] [blame]1933 <a name="dss_inteqclasses">"llvm/ADT/IntEqClasses.h"</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1934</h4>
Jakob Stoklund Olesen4359e5e2011-04-05 20:56:08 +0000[diff] [blame]1935
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1936<div>
Jakob Stoklund Olesen4359e5e2011-04-05 20:56:08 +0000[diff] [blame]1937
1938<p>IntEqClasses provides a compact representation of equivalence classes of
1939small integers. Initially, each integer in the range 0..n-1 has its own
1940equivalence class. Classes can be joined by passing two class representatives to
1941the join(a, b) method. Two integers are in the same class when findLeader()
1942returns the same representative.</p>
1943
1944<p>Once all equivalence classes are formed, the map can be compressed so each
1945integer 0..n-1 maps to an equivalence class number in the range 0..m-1, where m
1946is the total number of equivalence classes. The map must be uncompressed before
1947it can be edited again.</p>
1948
1949</div>
1950
1951<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1952<h4>
Chris Lattner2fdd0052011-11-15 22:40:14 +0000[diff] [blame]1953 <a name="dss_immutablemap">"llvm/ADT/ImmutableMap.h"</a>
1954</h4>
1955
1956<div>
1957
1958<p>
1959ImmutableMap is an immutable (functional) map implementation based on an AVL
1960tree.
1961Adding or removing elements is done through a Factory object and results in the
1962creation of a new ImmutableMap object.
1963If an ImmutableMap already exists with the given key set, then the existing one
1964is returned; equality is compared with a FoldingSetNodeID.
1965The time and space complexity of add or remove operations is logarithmic in the
1966size of the original map.
1967
1968</div>
1969
1970<!-- _______________________________________________________________________ -->
1971<h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1972 <a name="dss_othermap">Other Map-Like Container Options</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1973</h4>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1974
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1975<div>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1976
1977<p>
1978The STL provides several other options, such as std::multimap and the various
Chris Lattnerf1a30822009-03-09 05:20:45 +0000[diff] [blame]1979"hash_map" like containers (whether from C++ TR1 or from the SGI library). We
1980never use hash_set and unordered_set because they are generally very expensive
1981(each insertion requires a malloc) and very non-portable.</p>
Chris Lattnerc5722432007-02-03 19:49:31 +0000[diff] [blame]1982
1983<p>std::multimap is useful if you want to map a key to multiple values, but has
1984all the drawbacks of std::map. A sorted vector or some other approach is almost
1985always better.</p>
1986
Chris Lattner098129a2007-02-03 03:04:03 +0000[diff] [blame]1987</div>
1988
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1989</div>
1990
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1991<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1992<h3>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1993 <a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1994</h3>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]1995
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1996<div>
Chris Lattner7086ce72007-09-25 22:37:50 +0000[diff] [blame]1997<p>Unlike the other containers, there are only two bit storage containers, and
1998choosing when to use each is relatively straightforward.</p>
1999
2000<p>One additional option is
2001<tt>std::vector&lt;bool&gt;</tt>: we discourage its use for two reasons 1) the
2002implementation in many common compilers (e.g. commonly available versions of
2003GCC) is extremely inefficient and 2) the C++ standards committee is likely to
2004deprecate this container and/or change it significantly somehow. In any case,
2005please don't use it.</p>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2006
2007<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2008<h4>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2009 <a name="dss_bitvector">BitVector</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2010</h4>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2011
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2012<div>
Dan Gohman5f7775c2010-01-05 18:24:00 +0000[diff] [blame]2013<p> The BitVector container provides a dynamic size set of bits for manipulation.
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2014It supports individual bit setting/testing, as well as set operations. The set
2015operations take time O(size of bitvector), but operations are performed one word
2016at a time, instead of one bit at a time. This makes the BitVector very fast for
2017set operations compared to other containers. Use the BitVector when you expect
2018the number of set bits to be high (IE a dense set).
2019</p>
2020</div>
2021
2022<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2023<h4>
Dan Gohman5f7775c2010-01-05 18:24:00 +0000[diff] [blame]2024 <a name="dss_smallbitvector">SmallBitVector</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2025</h4>
Dan Gohman5f7775c2010-01-05 18:24:00 +0000[diff] [blame]2026
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2027<div>
Dan Gohman5f7775c2010-01-05 18:24:00 +0000[diff] [blame]2028<p> The SmallBitVector container provides the same interface as BitVector, but
2029it is optimized for the case where only a small number of bits, less than
203025 or so, are needed. It also transparently supports larger bit counts, but
2031slightly less efficiently than a plain BitVector, so SmallBitVector should
2032only be used when larger counts are rare.
2033</p>
2034
2035<p>
2036At this time, SmallBitVector does not support set operations (and, or, xor),
2037and its operator[] does not provide an assignable lvalue.
2038</p>
2039</div>
2040
2041<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2042<h4>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2043 <a name="dss_sparsebitvector">SparseBitVector</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2044</h4>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2045
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2046<div>
Daniel Berlin1939ace2007-09-24 17:52:25 +0000[diff] [blame]2047<p> The SparseBitVector container is much like BitVector, with one major
2048difference: Only the bits that are set, are stored. This makes the
2049SparseBitVector much more space efficient than BitVector when the set is sparse,
2050as well as making set operations O(number of set bits) instead of O(size of
2051universe). 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
2052(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).
2053</p>
2054</div>
Chris Lattnerf623a082005-10-17 01:36:23 +0000[diff] [blame]2055
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2056</div>
2057
2058</div>
2059
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2060<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2061<h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2062 <a name="common">Helpful Hints for Common Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2063</h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2064<!-- *********************************************************************** -->
2065
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2066<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2067
2068<p>This section describes how to perform some very simple transformations of
2069LLVM code. This is meant to give examples of common idioms used, showing the
2070practical side of LLVM transformations. <p> Because this is a "how-to" section,
2071you should also read about the main classes that you will be working with. The
2072<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
2073and descriptions of the main classes that you should know about.</p>
2074
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2075<!-- NOTE: this section should be heavy on example code -->
2076<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2077<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2078 <a name="inspection">Basic Inspection and Traversal Routines</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2079</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2080
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2081<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2082
2083<p>The LLVM compiler infrastructure have many different data structures that may
2084be traversed. Following the example of the C++ standard template library, the
2085techniques used to traverse these various data structures are all basically the
2086same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
2087method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
2088function returns an iterator pointing to one past the last valid element of the
2089sequence, and there is some <tt>XXXiterator</tt> data type that is common
2090between the two operations.</p>
2091
2092<p>Because the pattern for iteration is common across many different aspects of
2093the program representation, the standard template library algorithms may be used
2094on them, and it is easier to remember how to iterate. First we show a few common
2095examples of the data structures that need to be traversed. Other data
2096structures are traversed in very similar ways.</p>
2097
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2098<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2099<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2100 <a name="iterate_function">Iterating over the </a><a
2101 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
2102 href="#Function"><tt>Function</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2103</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2104
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2105<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2106
2107<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
2108transform in some way; in particular, you'd like to manipulate its
2109<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
2110the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
2111an example that prints the name of a <tt>BasicBlock</tt> and the number of
2112<tt>Instruction</tt>s it contains:</p>
2113
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2114<div class="doc_code">
2115<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2116// <i>func is a pointer to a Function instance</i>
2117for (Function::iterator i = func-&gt;begin(), e = func-&gt;end(); i != e; ++i)
2118 // <i>Print out the name of the basic block if it has one, and then the</i>
2119 // <i>number of instructions that it contains</i>
Chris Lattner3fee6ed2009-09-08 05:15:50 +0000[diff] [blame]2120 errs() &lt;&lt; "Basic block (name=" &lt;&lt; i-&gt;getName() &lt;&lt; ") has "
Bill Wendling832171c2006-12-07 20:04:42 +0000[diff] [blame]2121 &lt;&lt; i-&gt;size() &lt;&lt; " instructions.\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2122</pre>
2123</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2124
2125<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]2126invoking member functions of the <tt>Instruction</tt> class. This is
2127because the indirection operator is overloaded for the iterator
Chris Lattner7496ec52003-08-05 22:54:23 +0000[diff] [blame]2128classes. In the above code, the expression <tt>i-&gt;size()</tt> is
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2129exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
2130
2131</div>
2132
2133<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2134<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2135 <a name="iterate_basicblock">Iterating over the </a><a
2136 href="#Instruction"><tt>Instruction</tt></a>s in a <a
2137 href="#BasicBlock"><tt>BasicBlock</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2138</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2139
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2140<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2141
2142<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
2143easy to iterate over the individual instructions that make up
2144<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
2145a <tt>BasicBlock</tt>:</p>
2146
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2147<div class="doc_code">
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]2148<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2149// <i>blk is a pointer to a BasicBlock instance</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2150for (BasicBlock::iterator i = blk-&gt;begin(), e = blk-&gt;end(); i != e; ++i)
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2151 // <i>The next statement works since operator&lt;&lt;(ostream&amp;,...)</i>
2152 // <i>is overloaded for Instruction&amp;</i>
Chris Lattner3fee6ed2009-09-08 05:15:50 +0000[diff] [blame]2153 errs() &lt;&lt; *i &lt;&lt; "\n";
Chris Lattner55c04612005-03-06 06:00:13 +0000[diff] [blame]2154</pre>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2155</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2156
2157<p>However, this isn't really the best way to print out the contents of a
2158<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
2159anything you'll care about, you could have just invoked the print routine on the
Chris Lattner3fee6ed2009-09-08 05:15:50 +0000[diff] [blame]2160basic block itself: <tt>errs() &lt;&lt; *blk &lt;&lt; "\n";</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2161
2162</div>
2163
2164<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2165<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2166 <a name="iterate_institer">Iterating over the </a><a
2167 href="#Instruction"><tt>Instruction</tt></a>s in a <a
2168 href="#Function"><tt>Function</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2169</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2170
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2171<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2172
2173<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
2174<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
2175<tt>InstIterator</tt> should be used instead. You'll need to include <a
2176href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
2177and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]2178small 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]2179
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2180<div class="doc_code">
2181<pre>
2182#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
2183
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]2184// <i>F is a pointer to a Function instance</i>
Chris Lattnerda021aa2008-06-04 18:20:42 +0000[diff] [blame]2185for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
Chris Lattner3fee6ed2009-09-08 05:15:50 +0000[diff] [blame]2186 errs() &lt;&lt; *I &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2187</pre>
2188</div>
2189
2190<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]2191work list with its initial contents. For example, if you wanted to
2192initialize a work list to contain all instructions in a <tt>Function</tt>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2193F, all you would need to do is something like:</p>
2194
2195<div class="doc_code">
2196<pre>
2197std::set&lt;Instruction*&gt; worklist;
Chris Lattnerda021aa2008-06-04 18:20:42 +0000[diff] [blame]2198// or better yet, SmallPtrSet&lt;Instruction*, 64&gt; worklist;
2199
2200for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
2201 worklist.insert(&amp;*I);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2202</pre>
2203</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2204
2205<p>The STL set <tt>worklist</tt> would now contain all instructions in the
2206<tt>Function</tt> pointed to by F.</p>
2207
2208</div>
2209
2210<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2211<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2212 <a name="iterate_convert">Turning an iterator into a class pointer (and
2213 vice-versa)</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2214</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2215
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2216<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2217
2218<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]2219instance when all you've got at hand is an iterator. Well, extracting
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]2220a reference or a pointer from an iterator is very straight-forward.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2221Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2222is a <tt>BasicBlock::const_iterator</tt>:</p>
2223
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2224<div class="doc_code">
2225<pre>
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2226Instruction&amp; inst = *i; // <i>Grab reference to instruction reference</i>
2227Instruction* pinst = &amp;*i; // <i>Grab pointer to instruction reference</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2228const Instruction&amp; inst = *j;
2229</pre>
2230</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2231
2232<p>However, the iterators you'll be working with in the LLVM framework are
2233special: they will automatically convert to a ptr-to-instance type whenever they
2234need to. Instead of dereferencing the iterator and then taking the address of
2235the result, you can simply assign the iterator to the proper pointer type and
2236you get the dereference and address-of operation as a result of the assignment
2237(behind the scenes, this is a result of overloading casting mechanisms). Thus
2238the last line of the last example,</p>
2239
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2240<div class="doc_code">
2241<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2242Instruction *pinst = &amp;*i;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2243</pre>
2244</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2245
2246<p>is semantically equivalent to</p>
2247
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2248<div class="doc_code">
2249<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2250Instruction *pinst = i;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2251</pre>
2252</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2253
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]2254<p>It's also possible to turn a class pointer into the corresponding iterator,
2255and this is a constant time operation (very efficient). The following code
2256snippet illustrates use of the conversion constructors provided by LLVM
2257iterators. By using these, you can explicitly grab the iterator of something
2258without actually obtaining it via iteration over some structure:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2259
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2260<div class="doc_code">
2261<pre>
2262void printNextInstruction(Instruction* inst) {
2263 BasicBlock::iterator it(inst);
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2264 ++it; // <i>After this line, it refers to the instruction after *inst</i>
Chris Lattner3fee6ed2009-09-08 05:15:50 +0000[diff] [blame]2265 if (it != inst-&gt;getParent()-&gt;end()) errs() &lt;&lt; *it &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2266}
2267</pre>
2268</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2269
Dan Gohman525bf8e2010-03-26 19:39:05 +0000[diff] [blame]2270<p>Unfortunately, these implicit conversions come at a cost; they prevent
2271these iterators from conforming to standard iterator conventions, and thus
Dan Gohman0d91c112010-03-26 19:51:14 +0000[diff] [blame]2272from being usable with standard algorithms and containers. For example, they
2273prevent the following code, where <tt>B</tt> is a <tt>BasicBlock</tt>,
Dan Gohman525bf8e2010-03-26 19:39:05 +0000[diff] [blame]2274from compiling:</p>
2275
2276<div class="doc_code">
2277<pre>
2278 llvm::SmallVector&lt;llvm::Instruction *, 16&gt;(B-&gt;begin(), B-&gt;end());
2279</pre>
2280</div>
2281
2282<p>Because of this, these implicit conversions may be removed some day,
Dan Gohman0d91c112010-03-26 19:51:14 +0000[diff] [blame]2283and <tt>operator*</tt> changed to return a pointer instead of a reference.</p>
Dan Gohman525bf8e2010-03-26 19:39:05 +0000[diff] [blame]2284
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2285</div>
2286
2287<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2288<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2289 <a name="iterate_complex">Finding call sites: a slightly more complex
2290 example</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2291</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2292
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2293<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2294
2295<p>Say that you're writing a FunctionPass and would like to count all the
2296locations in the entire module (that is, across every <tt>Function</tt>) where a
2297certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
2298learn 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]2299much more straight-forward manner, but this example will allow us to explore how
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]2300you'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]2301is what we want to do:</p>
2302
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2303<div class="doc_code">
2304<pre>
2305initialize callCounter to zero
2306for each Function f in the Module
2307 for each BasicBlock b in f
2308 for each Instruction i in b
2309 if (i is a CallInst and calls the given function)
2310 increment callCounter
2311</pre>
2312</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2313
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2314<p>And the actual code is (remember, because we're writing a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2315<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2316override the <tt>runOnFunction</tt> method):</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2317
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2318<div class="doc_code">
2319<pre>
2320Function* targetFunc = ...;
2321
2322class OurFunctionPass : public FunctionPass {
2323 public:
2324 OurFunctionPass(): callCounter(0) { }
2325
2326 virtual runOnFunction(Function&amp; F) {
2327 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
Eric Christopher203e71d2008-11-08 08:20:49 +0000[diff] [blame]2328 for (BasicBlock::iterator i = b-&gt;begin(), ie = b-&gt;end(); i != ie; ++i) {
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2329 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a>&lt;<a
2330 href="#CallInst">CallInst</a>&gt;(&amp;*i)) {
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2331 // <i>We know we've encountered a call instruction, so we</i>
2332 // <i>need to determine if it's a call to the</i>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2333 // <i>function pointed to by m_func or not.</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2334 if (callInst-&gt;getCalledFunction() == targetFunc)
2335 ++callCounter;
2336 }
2337 }
2338 }
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2339 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2340
2341 private:
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2342 unsigned callCounter;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2343};
2344</pre>
2345</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2346
2347</div>
2348
Brian Gaekef1972c62003-11-07 19:25:45 +0000[diff] [blame]2349<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2350<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2351 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2352</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2353
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2354<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2355
2356<p>You may have noticed that the previous example was a bit oversimplified in
2357that it did not deal with call sites generated by 'invoke' instructions. In
2358this, and in other situations, you may find that you want to treat
2359<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
2360most-specific common base class is <tt>Instruction</tt>, which includes lots of
2361less closely-related things. For these cases, LLVM provides a handy wrapper
2362class called <a
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]2363href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]2364It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
2365methods that provide functionality common to <tt>CallInst</tt>s and
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2366<tt>InvokeInst</tt>s.</p>
2367
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]2368<p>This class has "value semantics": it should be passed by value, not by
2369reference and it should not be dynamically allocated or deallocated using
2370<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
2371assignable and constructable, with costs equivalents to that of a bare pointer.
2372If you look at its definition, it has only a single pointer member.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2373
2374</div>
2375
Chris Lattner1a3105b2002-09-09 05:49:39 +0000[diff] [blame]2376<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2377<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2378 <a name="iterate_chains">Iterating over def-use &amp; use-def chains</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2379</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2380
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2381<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2382
2383<p>Frequently, we might have an instance of the <a
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]2384href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]2385determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
2386<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
2387For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
2388particular function <tt>foo</tt>. Finding all of the instructions that
2389<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
2390of <tt>F</tt>:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2391
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2392<div class="doc_code">
2393<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2394Function *F = ...;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2395
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2396for (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]2397 if (Instruction *Inst = dyn_cast&lt;Instruction&gt;(*i)) {
Chris Lattner3fee6ed2009-09-08 05:15:50 +0000[diff] [blame]2398 errs() &lt;&lt; "F is used in instruction:\n";
2399 errs() &lt;&lt; *Inst &lt;&lt; "\n";
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2400 }
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2401</pre>
Gabor Greif394fdfb2010-03-26 19:35:48 +0000[diff] [blame]2402</div>
2403
Gabor Greifce94319532010-03-26 19:40:38 +0000[diff] [blame]2404<p>Note that dereferencing a <tt>Value::use_iterator</tt> is not a very cheap
Gabor Greif4de73682010-03-26 19:30:47 +0000[diff] [blame]2405operation. Instead of performing <tt>*i</tt> above several times, consider
Gabor Greifce94319532010-03-26 19:40:38 +0000[diff] [blame]2406doing it only once in the loop body and reusing its result.</p>
Gabor Greif4de73682010-03-26 19:30:47 +0000[diff] [blame]2407
Gabor Greif6091ff32010-03-26 19:04:42 +0000[diff] [blame]2408<p>Alternatively, it's common to have an instance of the <a
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]2409href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2410<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
2411<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
2412<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
2413all of the values that a particular instruction uses (that is, the operands of
2414the particular <tt>Instruction</tt>):</p>
2415
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2416<div class="doc_code">
2417<pre>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2418Instruction *pi = ...;
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2419
2420for (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]2421 Value *v = *i;
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2422 // <i>...</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2423}
2424</pre>
2425</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2426
Gabor Greif4de73682010-03-26 19:30:47 +0000[diff] [blame]2427<p>Declaring objects as <tt>const</tt> is an important tool of enforcing
Gabor Greifce94319532010-03-26 19:40:38 +0000[diff] [blame]2428mutation free algorithms (such as analyses, etc.). For this purpose above
Gabor Greif4de73682010-03-26 19:30:47 +0000[diff] [blame]2429iterators come in constant flavors as <tt>Value::const_use_iterator</tt>
2430and <tt>Value::const_op_iterator</tt>. They automatically arise when
2431calling <tt>use/op_begin()</tt> on <tt>const Value*</tt>s or
2432<tt>const User*</tt>s respectively. Upon dereferencing, they return
Gabor Greifce94319532010-03-26 19:40:38 +0000[diff] [blame]2433<tt>const Use*</tt>s. Otherwise the above patterns remain unchanged.</p>
2434
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2435</div>
2436
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2437<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2438<h4>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2439 <a name="iterate_preds">Iterating over predecessors &amp;
2440successors of blocks</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2441</h4>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2442
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2443<div>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2444
2445<p>Iterating over the predecessors and successors of a block is quite easy
2446with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
2447this to iterate over all predecessors of BB:</p>
2448
2449<div class="doc_code">
2450<pre>
2451#include "llvm/Support/CFG.h"
2452BasicBlock *BB = ...;
2453
2454for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
2455 BasicBlock *Pred = *PI;
2456 // <i>...</i>
2457}
2458</pre>
2459</div>
2460
2461<p>Similarly, to iterate over successors use
2462succ_iterator/succ_begin/succ_end.</p>
2463
2464</div>
2465
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2466</div>
Chris Lattner2e438ca2008-01-03 16:56:04 +0000[diff] [blame]2467
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2468<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2469<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2470 <a name="simplechanges">Making simple changes</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2471</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2472
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2473<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2474
2475<p>There are some primitive transformation operations present in the LLVM
Joel Stanley753eb712002-09-11 22:32:24 +0000[diff] [blame]2476infrastructure that are worth knowing about. When performing
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2477transformations, it's fairly common to manipulate the contents of basic
2478blocks. This section describes some of the common methods for doing so
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2479and gives example code.</p>
2480
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2481<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2482<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2483 <a name="schanges_creating">Creating and inserting new
2484 <tt>Instruction</tt>s</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2485</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2486
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2487<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2488
2489<p><i>Instantiating Instructions</i></p>
2490
Chris Lattner69bf8a92004-05-23 21:06:58 +0000[diff] [blame]2491<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2492constructor for the kind of instruction to instantiate and provide the necessary
2493parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
2494(const-ptr-to) <tt>Type</tt>. Thus:</p>
2495
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2496<div class="doc_code">
2497<pre>
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]2498AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2499</pre>
2500</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2501
2502<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]2503one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2504subclass is likely to have varying default parameters which change the semantics
2505of the instruction, so refer to the <a
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]2506href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2507Instruction</a> that you're interested in instantiating.</p>
2508
2509<p><i>Naming values</i></p>
2510
2511<p>It is very useful to name the values of instructions when you're able to, as
2512this facilitates the debugging of your transformations. If you end up looking
2513at generated LLVM machine code, you definitely want to have logical names
2514associated with the results of instructions! By supplying a value for the
2515<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
2516associate a logical name with the result of the instruction's execution at
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]2517run time. For example, say that I'm writing a transformation that dynamically
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2518allocates space for an integer on the stack, and that integer is going to be
2519used as some kind of index by some other code. To accomplish this, I place an
2520<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
2521<tt>Function</tt>, and I'm intending to use it within the same
2522<tt>Function</tt>. I might do:</p>
2523
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2524<div class="doc_code">
2525<pre>
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]2526AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2527</pre>
2528</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2529
2530<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
Reid Spencer128a7a72007-02-03 21:06:43 +0000[diff] [blame]2531execution value, which is a pointer to an integer on the run time stack.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2532
2533<p><i>Inserting instructions</i></p>
2534
2535<p>There are essentially two ways to insert an <tt>Instruction</tt>
2536into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
2537
Joel Stanley9dd1ad62002-09-18 03:17:23 +0000[diff] [blame]2538<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2539 <li>Insertion into an explicit instruction list
2540
2541 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
2542 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
2543 before <tt>*pi</tt>, we do the following: </p>
2544
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2545<div class="doc_code">
2546<pre>
2547BasicBlock *pb = ...;
2548Instruction *pi = ...;
2549Instruction *newInst = new Instruction(...);
2550
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2551pb-&gt;getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2552</pre>
2553</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]2554
2555 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
2556 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
2557 classes provide constructors which take a pointer to a
2558 <tt>BasicBlock</tt> to be appended to. For example code that
2559 looked like: </p>
2560
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2561<div class="doc_code">
2562<pre>
2563BasicBlock *pb = ...;
2564Instruction *newInst = new Instruction(...);
2565
Bill Wendling82e2eea2006-10-11 18:00:22 +0000[diff] [blame]2566pb-&gt;getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2567</pre>
2568</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]2569
2570 <p>becomes: </p>
2571
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2572<div class="doc_code">
2573<pre>
2574BasicBlock *pb = ...;
2575Instruction *newInst = new Instruction(..., pb);
2576</pre>
2577</div>
Alkis Evlogimenos9a5dc4f2004-05-27 00:57:51 +0000[diff] [blame]2578
2579 <p>which is much cleaner, especially if you are creating
2580 long instruction streams.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2581
2582 <li>Insertion into an implicit instruction list
2583
2584 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
2585 are implicitly associated with an existing instruction list: the instruction
2586 list of the enclosing basic block. Thus, we could have accomplished the same
2587 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
2588 </p>
2589
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2590<div class="doc_code">
2591<pre>
2592Instruction *pi = ...;
2593Instruction *newInst = new Instruction(...);
2594
2595pi-&gt;getParent()-&gt;getInstList().insert(pi, newInst);
2596</pre>
2597</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2598
2599 <p>In fact, this sequence of steps occurs so frequently that the
2600 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
2601 constructors which take (as a default parameter) a pointer to an
2602 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
2603 precede. That is, <tt>Instruction</tt> constructors are capable of
2604 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
2605 provided instruction, immediately before that instruction. Using an
2606 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
2607 parameter, the above code becomes:</p>
2608
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2609<div class="doc_code">
2610<pre>
2611Instruction* pi = ...;
2612Instruction* newInst = new Instruction(..., pi);
2613</pre>
2614</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2615
2616 <p>which is much cleaner, especially if you're creating a lot of
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2617 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2618</ul>
2619
2620</div>
2621
2622<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2623<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2624 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2625</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2626
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2627<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2628
2629<p>Deleting an instruction from an existing sequence of instructions that form a
Chris Lattner49e0ccf2011-03-24 16:13:31 +0000[diff] [blame]2630<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward: just
2631call the instruction's eraseFromParent() method. For example:</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2632
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2633<div class="doc_code">
2634<pre>
2635<a href="#Instruction">Instruction</a> *I = .. ;
Chris Lattner9f8ec252008-02-15 22:57:17 +0000[diff] [blame]2636I-&gt;eraseFromParent();
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2637</pre>
2638</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2639
Chris Lattner49e0ccf2011-03-24 16:13:31 +0000[diff] [blame]2640<p>This unlinks the instruction from its containing basic block and deletes
2641it. If you'd just like to unlink the instruction from its containing basic
2642block but not delete it, you can use the <tt>removeFromParent()</tt> method.</p>
2643
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2644</div>
2645
2646<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2647<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2648 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
2649 <tt>Value</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2650</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2651
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2652<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2653
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]2654<h5><i>Replacing individual instructions</i></h5>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2655
2656<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]2657permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2658and <tt>ReplaceInstWithInst</tt>.</p>
2659
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000[diff] [blame]2660<h5><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h5>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2661
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]2662<div>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2663<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2664 <li><tt>ReplaceInstWithValue</tt>
2665
Nick Lewyckyb6d1f392008-09-15 06:31:52 +0000[diff] [blame]2666 <p>This function replaces all uses of a given instruction with a value,
2667 and then removes the original instruction. The following example
2668 illustrates the replacement of the result of a particular
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]2669 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2670 pointer to an integer.</p>
2671
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2672<div class="doc_code">
2673<pre>
2674AllocaInst* instToReplace = ...;
2675BasicBlock::iterator ii(instToReplace);
2676
2677ReplaceInstWithValue(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Daniel Dunbar58c2ac02008-10-03 22:17:25 +0000[diff] [blame]2678 Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2679</pre></div></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2680
2681 <li><tt>ReplaceInstWithInst</tt>
2682
2683 <p>This function replaces a particular instruction with another
Nick Lewyckyb6d1f392008-09-15 06:31:52 +0000[diff] [blame]2684 instruction, inserting the new instruction into the basic block at the
2685 location where the old instruction was, and replacing any uses of the old
2686 instruction with the new instruction. The following example illustrates
2687 the replacement of one <tt>AllocaInst</tt> with another.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2688
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2689<div class="doc_code">
2690<pre>
2691AllocaInst* instToReplace = ...;
2692BasicBlock::iterator ii(instToReplace);
2693
2694ReplaceInstWithInst(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Nick Lewycky10d64b92007-12-03 01:52:52 +0000[diff] [blame]2695 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]2696</pre></div></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2697</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2698
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]2699</div>
2700
2701<h5><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></h5>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2702
2703<p>You can use <tt>Value::replaceAllUsesWith</tt> and
2704<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]2705doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]2706and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]2707information.</p>
2708
2709<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
2710include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
2711ReplaceInstWithValue, ReplaceInstWithInst -->
2712
2713</div>
2714
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2715<!--_______________________________________________________________________-->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2716<h4>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2717 <a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2718</h4>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2719
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2720<div>
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2721
Tanya Lattnerc5dfcdb2007-06-20 20:46:37 +0000[diff] [blame]2722<p>Deleting a global variable from a module is just as easy as deleting an
2723Instruction. First, you must have a pointer to the global variable that you wish
2724 to delete. You use this pointer to erase it from its parent, the module.
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2725 For example:</p>
2726
2727<div class="doc_code">
2728<pre>
2729<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2730
Tanya Lattnerc5dfcdb2007-06-20 20:46:37 +0000[diff] [blame]2731GV-&gt;eraseFromParent();
Tanya Lattnerb011c662007-06-20 18:33:15 +0000[diff] [blame]2732</pre>
2733</div>
2734
2735</div>
2736
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2737</div>
2738
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2739<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2740<h3>
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2741 <a name="create_types">How to Create Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2742</h3>
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2743
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2744<div>
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2745
2746<p>In generating IR, you may need some complex types. If you know these types
Misha Brukman1af789f2009-05-01 20:40:51 +0000[diff] [blame]2747statically, you can use <tt>TypeBuilder&lt;...&gt;::get()</tt>, defined
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2748in <tt>llvm/Support/TypeBuilder.h</tt>, to retrieve them. <tt>TypeBuilder</tt>
2749has two forms depending on whether you're building types for cross-compilation
Misha Brukman1af789f2009-05-01 20:40:51 +0000[diff] [blame]2750or native library use. <tt>TypeBuilder&lt;T, true&gt;</tt> requires
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2751that <tt>T</tt> be independent of the host environment, meaning that it's built
2752out of types from
2753the <a href="/doxygen/namespacellvm_1_1types.html"><tt>llvm::types</tt></a>
2754namespace and pointers, functions, arrays, etc. built of
Misha Brukman1af789f2009-05-01 20:40:51 +0000[diff] [blame]2755those. <tt>TypeBuilder&lt;T, false&gt;</tt> additionally allows native C types
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2756whose size may depend on the host compiler. For example,</p>
2757
2758<div class="doc_code">
2759<pre>
Misha Brukman1af789f2009-05-01 20:40:51 +0000[diff] [blame]2760FunctionType *ft = TypeBuilder&lt;types::i&lt;8&gt;(types::i&lt;32&gt;*), true&gt;::get();
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2761</pre>
2762</div>
2763
2764<p>is easier to read and write than the equivalent</p>
2765
2766<div class="doc_code">
2767<pre>
Owen Anderson5e8c50e2009-06-16 17:40:28 +0000[diff] [blame]2768std::vector&lt;const Type*&gt; params;
Jeffrey Yasskin714257f2009-04-30 22:33:41 +0000[diff] [blame]2769params.push_back(PointerType::getUnqual(Type::Int32Ty));
2770FunctionType *ft = FunctionType::get(Type::Int8Ty, params, false);
2771</pre>
2772</div>
2773
2774<p>See the <a href="/doxygen/TypeBuilder_8h-source.html#l00001">class
2775comment</a> for more details.</p>
2776
2777</div>
2778
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2779</div>
2780
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]2781<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2782<h2>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2783 <a name="threading">Threads and LLVM</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2784</h2>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2785<!-- *********************************************************************** -->
2786
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2787<div>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2788<p>
2789This section describes the interaction of the LLVM APIs with multithreading,
2790both on the part of client applications, and in the JIT, in the hosted
2791application.
2792</p>
2793
2794<p>
2795Note that LLVM's support for multithreading is still relatively young. Up
2796through version 2.5, the execution of threaded hosted applications was
2797supported, but not threaded client access to the APIs. While this use case is
2798now supported, clients <em>must</em> adhere to the guidelines specified below to
2799ensure proper operation in multithreaded mode.
2800</p>
2801
2802<p>
2803Note that, on Unix-like platforms, LLVM requires the presence of GCC's atomic
2804intrinsics in order to support threaded operation. If you need a
2805multhreading-capable LLVM on a platform without a suitably modern system
2806compiler, consider compiling LLVM and LLVM-GCC in single-threaded mode, and
2807using the resultant compiler to build a copy of LLVM with multithreading
2808support.
2809</p>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2810
2811<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2812<h3>
Owen Anderson1ad70e32009-06-16 18:04:19 +0000[diff] [blame]2813 <a name="startmultithreaded">Entering and Exiting Multithreaded Mode</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2814</h3>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2815
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2816<div>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2817
2818<p>
2819In order to properly protect its internal data structures while avoiding
Owen Anderson1ad70e32009-06-16 18:04:19 +0000[diff] [blame]2820excessive locking overhead in the single-threaded case, the LLVM must intialize
2821certain data structures necessary to provide guards around its internals. To do
2822so, the client program must invoke <tt>llvm_start_multithreaded()</tt> before
2823making any concurrent LLVM API calls. To subsequently tear down these
2824structures, use the <tt>llvm_stop_multithreaded()</tt> call. You can also use
2825the <tt>llvm_is_multithreaded()</tt> call to check the status of multithreaded
2826mode.
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2827</p>
2828
2829<p>
Owen Anderson1ad70e32009-06-16 18:04:19 +0000[diff] [blame]2830Note that both of these calls must be made <em>in isolation</em>. That is to
2831say that no other LLVM API calls may be executing at any time during the
2832execution of <tt>llvm_start_multithreaded()</tt> or <tt>llvm_stop_multithreaded
2833</tt>. It's is the client's responsibility to enforce this isolation.
2834</p>
2835
2836<p>
2837The return value of <tt>llvm_start_multithreaded()</tt> indicates the success or
2838failure of the initialization. Failure typically indicates that your copy of
2839LLVM was built without multithreading support, typically because GCC atomic
2840intrinsics were not found in your system compiler. In this case, the LLVM API
2841will not be safe for concurrent calls. However, it <em>will</em> be safe for
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]2842hosting threaded applications in the JIT, though <a href="#jitthreading">care
2843must be taken</a> to ensure that side exits and the like do not accidentally
2844result in concurrent LLVM API calls.
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2845</p>
2846</div>
2847
2848<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2849<h3>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2850 <a name="shutdown">Ending Execution with <tt>llvm_shutdown()</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2851</h3>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2852
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2853<div>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2854<p>
2855When you are done using the LLVM APIs, you should call <tt>llvm_shutdown()</tt>
Owen Anderson1ad70e32009-06-16 18:04:19 +0000[diff] [blame]2856to deallocate memory used for internal structures. This will also invoke
2857<tt>llvm_stop_multithreaded()</tt> if LLVM is operating in multithreaded mode.
2858As such, <tt>llvm_shutdown()</tt> requires the same isolation guarantees as
2859<tt>llvm_stop_multithreaded()</tt>.
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2860</p>
2861
2862<p>
2863Note that, if you use scope-based shutdown, you can use the
2864<tt>llvm_shutdown_obj</tt> class, which calls <tt>llvm_shutdown()</tt> in its
2865destructor.
2866</div>
2867
2868<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2869<h3>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2870 <a name="managedstatic">Lazy Initialization with <tt>ManagedStatic</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2871</h3>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2872
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2873<div>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2874<p>
2875<tt>ManagedStatic</tt> is a utility class in LLVM used to implement static
2876initialization of static resources, such as the global type tables. Before the
2877invocation of <tt>llvm_shutdown()</tt>, it implements a simple lazy
2878initialization scheme. Once <tt>llvm_start_multithreaded()</tt> returns,
2879however, it uses double-checked locking to implement thread-safe lazy
2880initialization.
2881</p>
2882
2883<p>
2884Note that, because no other threads are allowed to issue LLVM API calls before
2885<tt>llvm_start_multithreaded()</tt> returns, it is possible to have
2886<tt>ManagedStatic</tt>s of <tt>llvm::sys::Mutex</tt>s.
2887</p>
Owen Anderson1ad70e32009-06-16 18:04:19 +0000[diff] [blame]2888
2889<p>
2890The <tt>llvm_acquire_global_lock()</tt> and <tt>llvm_release_global_lock</tt>
2891APIs provide access to the global lock used to implement the double-checked
2892locking for lazy initialization. These should only be used internally to LLVM,
2893and only if you know what you're doing!
2894</p>
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2895</div>
2896
Owen Andersone0c951a2009-08-19 17:58:52 +0000[diff] [blame]2897<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2898<h3>
Owen Andersone0c951a2009-08-19 17:58:52 +0000[diff] [blame]2899 <a name="llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2900</h3>
Owen Andersone0c951a2009-08-19 17:58:52 +0000[diff] [blame]2901
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2902<div>
Owen Andersone0c951a2009-08-19 17:58:52 +0000[diff] [blame]2903<p>
2904<tt>LLVMContext</tt> is an opaque class in the LLVM API which clients can use
2905to operate multiple, isolated instances of LLVM concurrently within the same
2906address space. For instance, in a hypothetical compile-server, the compilation
2907of an individual translation unit is conceptually independent from all the
2908others, and it would be desirable to be able to compile incoming translation
2909units concurrently on independent server threads. Fortunately,
2910<tt>LLVMContext</tt> exists to enable just this kind of scenario!
2911</p>
2912
2913<p>
2914Conceptually, <tt>LLVMContext</tt> provides isolation. Every LLVM entity
2915(<tt>Module</tt>s, <tt>Value</tt>s, <tt>Type</tt>s, <tt>Constant</tt>s, etc.)
Chris Lattner38eee3c2009-08-20 03:10:14 +0000[diff] [blame]2916in LLVM's in-memory IR belongs to an <tt>LLVMContext</tt>. Entities in
Owen Andersone0c951a2009-08-19 17:58:52 +0000[diff] [blame]2917different contexts <em>cannot</em> interact with each other: <tt>Module</tt>s in
2918different contexts cannot be linked together, <tt>Function</tt>s cannot be added
2919to <tt>Module</tt>s in different contexts, etc. What this means is that is is
2920safe to compile on multiple threads simultaneously, as long as no two threads
2921operate on entities within the same context.
2922</p>
2923
2924<p>
2925In practice, very few places in the API require the explicit specification of a
2926<tt>LLVMContext</tt>, other than the <tt>Type</tt> creation/lookup APIs.
2927Because every <tt>Type</tt> carries a reference to its owning context, most
2928other entities can determine what context they belong to by looking at their
2929own <tt>Type</tt>. If you are adding new entities to LLVM IR, please try to
2930maintain this interface design.
2931</p>
2932
2933<p>
2934For clients that do <em>not</em> require the benefits of isolation, LLVM
2935provides a convenience API <tt>getGlobalContext()</tt>. This returns a global,
2936lazily initialized <tt>LLVMContext</tt> that may be used in situations where
2937isolation is not a concern.
2938</p>
2939</div>
2940
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]2941<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2942<h3>
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]2943 <a name="jitthreading">Threads and the JIT</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2944</h3>
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]2945
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2946<div>
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]2947<p>
2948LLVM's "eager" JIT compiler is safe to use in threaded programs. Multiple
2949threads can call <tt>ExecutionEngine::getPointerToFunction()</tt> or
2950<tt>ExecutionEngine::runFunction()</tt> concurrently, and multiple threads can
2951run code output by the JIT concurrently. The user must still ensure that only
2952one thread accesses IR in a given <tt>LLVMContext</tt> while another thread
2953might be modifying it. One way to do that is to always hold the JIT lock while
2954accessing IR outside the JIT (the JIT <em>modifies</em> the IR by adding
2955<tt>CallbackVH</tt>s). Another way is to only
2956call <tt>getPointerToFunction()</tt> from the <tt>LLVMContext</tt>'s thread.
2957</p>
2958
2959<p>When the JIT is configured to compile lazily (using
2960<tt>ExecutionEngine::DisableLazyCompilation(false)</tt>), there is currently a
2961<a href="http://llvm.org/bugs/show_bug.cgi?id=5184">race condition</a> in
2962updating call sites after a function is lazily-jitted. It's still possible to
2963use the lazy JIT in a threaded program if you ensure that only one thread at a
2964time can call any particular lazy stub and that the JIT lock guards any IR
2965access, but we suggest using only the eager JIT in threaded programs.
2966</p>
2967</div>
2968
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2969</div>
2970
Owen Anderson8bc1b3b2009-06-16 01:17:16 +0000[diff] [blame]2971<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2972<h2>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2973 <a name="advanced">Advanced Topics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2974</h2>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2975<!-- *********************************************************************** -->
2976
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2977<div>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2978<p>
2979This section describes some of the advanced or obscure API's that most clients
2980do not need to be aware of. These API's tend manage the inner workings of the
2981LLVM system, and only need to be accessed in unusual circumstances.
2982</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2983
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2984
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2985<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2986<h3>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2987 <a name="SymbolTable">The <tt>ValueSymbolTable</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2988</h3>
Chris Lattnerf1b200b2005-04-23 17:27:36 +0000[diff] [blame]2989
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2990<div>
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2991<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
2992ValueSymbolTable</a></tt> class provides a symbol table that the <a
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2993href="#Function"><tt>Function</tt></a> and <a href="#Module">
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]2994<tt>Module</tt></a> classes use for naming value definitions. The symbol table
2995can provide a name for any <a href="#Value"><tt>Value</tt></a>.
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2996</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]2997
Reid Spencera6362242007-01-07 00:41:39 +0000[diff] [blame]2998<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
2999by most clients. It should only be used when iteration over the symbol table
3000names themselves are required, which is very special purpose. Note that not
3001all LLVM
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3002<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]3003an empty name) do not exist in the symbol table.
3004</p>
3005
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]3006<p>Symbol tables support iteration over the values in the symbol
Chris Lattner263a98e2007-02-16 04:37:31 +0000[diff] [blame]3007table with <tt>begin/end/iterator</tt> and supports querying to see if a
3008specific name is in the symbol table (with <tt>lookup</tt>). The
3009<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
3010simply call <tt>setName</tt> on a value, which will autoinsert it into the
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]3011appropriate symbol table.</p>
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]3012
Chris Lattnerd9d6e102005-04-23 16:10:52 +0000[diff] [blame]3013</div>
3014
3015
3016
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3017<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3018<h3>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3019 <a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3020</h3>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3021
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3022<div>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3023<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1User.html">
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3024User</a></tt> class provides a basis for expressing the ownership of <tt>User</tt>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3025towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
3026Value</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]3027Use</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]3028addition and removal.</p>
3029
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3030<!-- ______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3031<h4>
3032 <a name="Use2User">
3033 Interaction and relationship between <tt>User</tt> and <tt>Use</tt> objects
3034 </a>
3035</h4>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3036
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3037<div>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3038<p>
3039A subclass of <tt>User</tt> can choose between incorporating its <tt>Use</tt> objects
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3040or refer to them out-of-line by means of a pointer. A mixed variant
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3041(some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
3042that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
3043</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3044
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3045<p>
3046We have 2 different layouts in the <tt>User</tt> (sub)classes:
3047<ul>
3048<li><p>Layout a)
3049The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
3050object and there are a fixed number of them.</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3051
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3052<li><p>Layout b)
3053The <tt>Use</tt> object(s) are referenced by a pointer to an
3054array from the <tt>User</tt> object and there may be a variable
3055number of them.</p>
3056</ul>
3057<p>
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3058As of v2.4 each layout still possesses a direct pointer to the
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3059start of the array of <tt>Use</tt>s. Though not mandatory for layout a),
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3060we stick to this redundancy for the sake of simplicity.
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3061The <tt>User</tt> object also stores the number of <tt>Use</tt> objects it
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3062has. (Theoretically this information can also be calculated
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3063given the scheme presented below.)</p>
3064<p>
3065Special forms of allocation operators (<tt>operator new</tt>)
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3066enforce the following memory layouts:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3067
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3068<ul>
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3069<li><p>Layout a) is 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]3070
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3071<pre>
3072...---.---.---.---.-------...
3073 | P | P | P | P | User
3074'''---'---'---'---'-------'''
3075</pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3076
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3077<li><p>Layout b) is modelled by pointing at the <tt>Use[]</tt> array.</p>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3078<pre>
3079.-------...
3080| User
3081'-------'''
3082 |
3083 v
3084 .---.---.---.---...
3085 | P | P | P | P |
3086 '---'---'---'---'''
3087</pre>
3088</ul>
3089<i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
3090 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]3091
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3092</div>
3093
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3094<!-- ______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3095<h4>
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3096 <a name="Waymarking">The waymarking algorithm</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3097</h4>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3098
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3099<div>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3100<p>
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3101Since the <tt>Use</tt> objects are deprived of the direct (back)pointer to
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3102their <tt>User</tt> objects, there must be a fast and exact method to
3103recover it. This is accomplished by the following scheme:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3104
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3105A bit-encoding in the 2 LSBits (least significant bits) of the <tt>Use::Prev</tt> allows to find the
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3106start of the <tt>User</tt> object:
3107<ul>
3108<li><tt>00</tt> &mdash;&gt; binary digit 0</li>
3109<li><tt>01</tt> &mdash;&gt; binary digit 1</li>
3110<li><tt>10</tt> &mdash;&gt; stop and calculate (<tt>s</tt>)</li>
3111<li><tt>11</tt> &mdash;&gt; full stop (<tt>S</tt>)</li>
3112</ul>
3113<p>
3114Given a <tt>Use*</tt>, all we have to do is to walk till we get
3115a stop and we either have a <tt>User</tt> immediately behind or
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3116we have to walk to the next stop picking up digits
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3117and calculating the offset:</p>
3118<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3119.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
3120| 1 | s | 1 | 0 | 1 | 0 | s | 1 | 1 | 0 | s | 1 | 1 | s | 1 | S | User (or User*)
3121'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
3122 |+15 |+10 |+6 |+3 |+1
3123 | | | | |__>
3124 | | | |__________>
3125 | | |______________________>
3126 | |______________________________________>
3127 |__________________________________________________________>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3128</pre>
3129<p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3130Only the significant number of bits need to be stored between the
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3131stops, so that the <i>worst case is 20 memory accesses</i> when there are
31321000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3133
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3134</div>
3135
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3136<!-- ______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3137<h4>
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3138 <a name="ReferenceImpl">Reference implementation</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3139</h4>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3140
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3141<div>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3142<p>
3143The following literate Haskell fragment demonstrates the concept:</p>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3144
3145<div class="doc_code">
3146<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3147> import Test.QuickCheck
3148>
3149> digits :: Int -> [Char] -> [Char]
3150> digits 0 acc = '0' : acc
3151> digits 1 acc = '1' : acc
3152> digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
3153>
3154> dist :: Int -> [Char] -> [Char]
3155> dist 0 [] = ['S']
3156> dist 0 acc = acc
3157> dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
3158> dist n acc = dist (n - 1) $ dist 1 acc
3159>
3160> takeLast n ss = reverse $ take n $ reverse ss
3161>
3162> test = takeLast 40 $ dist 20 []
3163>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3164</pre>
3165</div>
3166<p>
3167Printing &lt;test&gt; gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
3168<p>
3169The reverse algorithm computes the length of the string just by examining
3170a certain prefix:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3171
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3172<div class="doc_code">
3173<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3174> pref :: [Char] -> Int
3175> pref "S" = 1
3176> pref ('s':'1':rest) = decode 2 1 rest
3177> pref (_:rest) = 1 + pref rest
3178>
3179> decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
3180> decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
3181> decode walk acc _ = walk + acc
3182>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3183</pre>
3184</div>
3185<p>
3186Now, as expected, printing &lt;pref test&gt; gives <tt>40</tt>.</p>
3187<p>
3188We can <i>quickCheck</i> this with following property:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3189
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3190<div class="doc_code">
3191<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3192> testcase = dist 2000 []
3193> testcaseLength = length testcase
3194>
3195> identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
3196> where arr = takeLast n testcase
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3197>
3198</pre>
3199</div>
3200<p>
3201As expected &lt;quickCheck identityProp&gt; gives:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3202
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3203<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3204*Main> quickCheck identityProp
3205OK, passed 100 tests.
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3206</pre>
3207<p>
3208Let's be a bit more exhaustive:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3209
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3210<div class="doc_code">
3211<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3212>
3213> deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
3214>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3215</pre>
3216</div>
3217<p>
3218And here is the result of &lt;deepCheck identityProp&gt;:</p>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3219
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3220<pre>
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3221*Main> deepCheck identityProp
3222OK, passed 500 tests.
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3223</pre>
3224
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3225</div>
3226
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3227<!-- ______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3228<h4>
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3229 <a name="Tagging">Tagging considerations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3230</h4>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3231
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3232<div>
3233
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3234<p>
3235To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
3236never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
3237new <tt>Use**</tt> on every modification. Accordingly getters must strip the
3238tag bits.</p>
3239<p>
Gabor Greifd41720a2008-06-25 00:10:22 +0000[diff] [blame]3240For layout b) instead of the <tt>User</tt> we find a pointer (<tt>User*</tt> with LSBit set).
3241Following this pointer brings us to the <tt>User</tt>. A portable trick ensures
3242that the first bytes of <tt>User</tt> (if interpreted as a pointer) never has
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3243the LSBit set. (Portability is relying on the fact that all known compilers place the
3244<tt>vptr</tt> in the first word of the instances.)</p>
Gabor Greifdfed1182008-06-18 13:44:57 +0000[diff] [blame]3245
Gabor Greife98fc272008-06-16 21:06:12 +0000[diff] [blame]3246</div>
3247
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3248</div>
3249
3250</div>
3251
3252<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3253<h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3254 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3255</h2>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3256<!-- *********************************************************************** -->
3257
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3258<div>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3259<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
3260<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3261
3262<p>The Core LLVM classes are the primary means of representing the program
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3263being inspected or transformed. The core LLVM classes are defined in
3264header files in the <tt>include/llvm/</tt> directory, and implemented in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3265the <tt>lib/VMCore</tt> directory.</p>
3266
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3267<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3268<h3>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3269 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3270</h3>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3271
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3272<div>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3273
3274 <p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
3275 a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
3276 through its subclasses. Certain primitive types (<tt>VoidType</tt>,
3277 <tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
3278 subclasses. They are hidden because they offer no useful functionality beyond
3279 what the <tt>Type</tt> class offers except to distinguish themselves from
3280 other subclasses of <tt>Type</tt>.</p>
3281 <p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
3282 named, but this is not a requirement. There exists exactly
3283 one instance of a given shape at any one time. This allows type equality to
3284 be performed with address equality of the Type Instance. That is, given two
3285 <tt>Type*</tt> values, the types are identical if the pointers are identical.
3286 </p>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3287
3288<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3289<h4>
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3290 <a name="m_Type">Important Public Methods</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3291</h4>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3292
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3293<div>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3294
3295<ul>
Duncan Sandsb0bc6c32010-02-15 16:12:20 +0000[diff] [blame]3296 <li><tt>bool isIntegerTy() const</tt>: Returns true for any integer type.</li>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3297
Duncan Sandsb0bc6c32010-02-15 16:12:20 +0000[diff] [blame]3298 <li><tt>bool isFloatingPointTy()</tt>: Return true if this is one of the five
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3299 floating point types.</li>
3300
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3301 <li><tt>bool isSized()</tt>: Return true if the type has known size. Things
3302 that don't have a size are abstract types, labels and void.</li>
3303
3304</ul>
3305</div>
3306
3307<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3308<h4>
Gabor Greiffd095b62009-01-05 16:05:32 +0000[diff] [blame]3309 <a name="derivedtypes">Important Derived Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3310</h4>
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3311<div>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3312<dl>
3313 <dt><tt>IntegerType</tt></dt>
3314 <dd>Subclass of DerivedType that represents integer types of any bit width.
3315 Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
3316 <tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
3317 <ul>
3318 <li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
3319 type of a specific bit width.</li>
3320 <li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
3321 type.</li>
3322 </ul>
3323 </dd>
3324 <dt><tt>SequentialType</tt></dt>
Tobias Grosserd475c102011-07-12 11:37:02 +0000[diff] [blame]3325 <dd>This is subclassed by ArrayType, PointerType and VectorType.
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3326 <ul>
3327 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
3328 of the elements in the sequential type. </li>
3329 </ul>
3330 </dd>
3331 <dt><tt>ArrayType</tt></dt>
3332 <dd>This is a subclass of SequentialType and defines the interface for array
3333 types.
3334 <ul>
3335 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
3336 elements in the array. </li>
3337 </ul>
3338 </dd>
3339 <dt><tt>PointerType</tt></dt>
Chris Lattner302da1e2007-02-03 03:05:57 +0000[diff] [blame]3340 <dd>Subclass of SequentialType for pointer types.</dd>
Reid Spencer9d6565a2007-02-15 02:26:10 +0000[diff] [blame]3341 <dt><tt>VectorType</tt></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]3342 <dd>Subclass of SequentialType for vector types. A
3343 vector type is similar to an ArrayType but is distinguished because it is
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]3344 a first class type whereas ArrayType is not. Vector types are used for
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3345 vector operations and are usually small vectors of of an integer or floating
3346 point type.</dd>
3347 <dt><tt>StructType</tt></dt>
3348 <dd>Subclass of DerivedTypes for struct types.</dd>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3349 <dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3350 <dd>Subclass of DerivedTypes for function types.
3351 <ul>
Dan Gohman4bb31bf2010-03-30 20:04:57 +0000[diff] [blame]3352 <li><tt>bool isVarArg() const</tt>: Returns true if it's a vararg
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3353 function</li>
3354 <li><tt> const Type * getReturnType() const</tt>: Returns the
3355 return type of the function.</li>
3356 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
3357 the type of the ith parameter.</li>
3358 <li><tt> const unsigned getNumParams() const</tt>: Returns the
3359 number of formal parameters.</li>
3360 </ul>
3361 </dd>
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3362</dl>
3363</div>
3364
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3365</div>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3366
3367<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3368<h3>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3369 <a name="Module">The <tt>Module</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3370</h3>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3371
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3372<div>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3373
3374<p><tt>#include "<a
3375href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
3376<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
3377
3378<p>The <tt>Module</tt> class represents the top level structure present in LLVM
3379programs. An LLVM module is effectively either a translation unit of the
3380original program or a combination of several translation units merged by the
3381linker. The <tt>Module</tt> class keeps track of a list of <a
3382href="#Function"><tt>Function</tt></a>s, a list of <a
3383href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
3384href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
3385helpful member functions that try to make common operations easy.</p>
3386
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3387<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3388<h4>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3389 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3390</h4>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3391
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3392<div>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3393
3394<ul>
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]3395 <li><tt>Module::Module(std::string name = "")</tt>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3396
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]3397 <p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3398provide a name for it (probably based on the name of the translation unit).</p>
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]3399 </li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3400
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3401 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
3402 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
3403
3404 <tt>begin()</tt>, <tt>end()</tt>
3405 <tt>size()</tt>, <tt>empty()</tt>
3406
3407 <p>These are forwarding methods that make it easy to access the contents of
3408 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
3409 list.</p></li>
3410
3411 <li><tt>Module::FunctionListType &amp;getFunctionList()</tt>
3412
3413 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
3414 necessary to use when you need to update the list or perform a complex
3415 action that doesn't have a forwarding method.</p>
3416
3417 <p><!-- Global Variable --></p></li>
3418</ul>
3419
3420<hr>
3421
3422<ul>
3423 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
3424
3425 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
3426
3427 <tt>global_begin()</tt>, <tt>global_end()</tt>
3428 <tt>global_size()</tt>, <tt>global_empty()</tt>
3429
3430 <p> These are forwarding methods that make it easy to access the contents of
3431 a <tt>Module</tt> object's <a
3432 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
3433
3434 <li><tt>Module::GlobalListType &amp;getGlobalList()</tt>
3435
3436 <p>Returns the list of <a
3437 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
3438 use when you need to update the list or perform a complex action that
3439 doesn't have a forwarding method.</p>
3440
3441 <p><!-- Symbol table stuff --> </p></li>
3442</ul>
3443
3444<hr>
3445
3446<ul>
3447 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
3448
3449 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3450 for this <tt>Module</tt>.</p>
3451
3452 <p><!-- Convenience methods --></p></li>
3453</ul>
3454
3455<hr>
3456
3457<ul>
3458 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
3459 &amp;Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
3460
3461 <p>Look up the specified function in the <tt>Module</tt> <a
3462 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
3463 <tt>null</tt>.</p></li>
3464
3465 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
3466 std::string &amp;Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
3467
3468 <p>Look up the specified function in the <tt>Module</tt> <a
3469 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
3470 external declaration for the function and return it.</p></li>
3471
3472 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
3473
3474 <p>If there is at least one entry in the <a
3475 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
3476 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
3477 string.</p></li>
3478
3479 <li><tt>bool addTypeName(const std::string &amp;Name, const <a
3480 href="#Type">Type</a> *Ty)</tt>
3481
3482 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3483 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
3484 name, true is returned and the <a
3485 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
3486</ul>
3487
3488</div>
3489
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3490</div>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3491
Reid Spencer303c4b42007-01-12 17:26:25 +0000[diff] [blame]3492<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3493<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3494 <a name="Value">The <tt>Value</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3495</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3496
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3497<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3498
3499<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
3500<br>
Chris Lattner00815172007-01-04 22:01:45 +0000[diff] [blame]3501doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3502
3503<p>The <tt>Value</tt> class is the most important class in the LLVM Source
3504base. It represents a typed value that may be used (among other things) as an
3505operand to an instruction. There are many different types of <tt>Value</tt>s,
3506such as <a href="#Constant"><tt>Constant</tt></a>s,<a
3507href="#Argument"><tt>Argument</tt></a>s. Even <a
3508href="#Instruction"><tt>Instruction</tt></a>s and <a
3509href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
3510
3511<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
3512for a program. For example, an incoming argument to a function (represented
3513with an instance of the <a href="#Argument">Argument</a> class) is "used" by
3514every instruction in the function that references the argument. To keep track
3515of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
3516href="#User"><tt>User</tt></a>s that is using it (the <a
3517href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
3518graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
3519def-use information in the program, and is accessible through the <tt>use_</tt>*
3520methods, shown below.</p>
3521
3522<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
3523and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
3524method. In addition, all LLVM values can be named. The "name" of the
3525<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
3526
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]3527<div class="doc_code">
3528<pre>
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]3529%<b>foo</b> = add i32 1, 2
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]3530</pre>
3531</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3532
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3533<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]3534that the name of any value may be missing (an empty string), so names should
3535<b>ONLY</b> be used for debugging (making the source code easier to read,
3536debugging printouts), they should not be used to keep track of values or map
3537between them. For this purpose, use a <tt>std::map</tt> of pointers to the
3538<tt>Value</tt> itself instead.</p>
3539
3540<p>One important aspect of LLVM is that there is no distinction between an SSA
3541variable and the operation that produces it. Because of this, any reference to
3542the value produced by an instruction (or the value available as an incoming
Chris Lattnerd5fc4fc2004-03-18 14:58:55 +0000[diff] [blame]3543argument, for example) is represented as a direct pointer to the instance of
3544the class that
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3545represents this value. Although this may take some getting used to, it
3546simplifies the representation and makes it easier to manipulate.</p>
3547
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3548<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3549<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3550 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3551</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3552
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3553<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3554
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3555<ul>
3556 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
3557use-list<br>
Gabor Greifbbbf9a22010-03-26 19:59:25 +0000[diff] [blame]3558 <tt>Value::const_use_iterator</tt> - Typedef for const_iterator over
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3559the use-list<br>
3560 <tt>unsigned use_size()</tt> - Returns the number of users of the
3561value.<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3562 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3563 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
3564the use-list.<br>
3565 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
3566use-list.<br>
3567 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
3568element in the list.
3569 <p> These methods are the interface to access the def-use
3570information in LLVM. As with all other iterators in LLVM, the naming
3571conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3572 </li>
3573 <li><tt><a href="#Type">Type</a> *getType() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3574 <p>This method returns the Type of the Value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3575 </li>
3576 <li><tt>bool hasName() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3577 <tt>std::string getName() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3578 <tt>void setName(const std::string &amp;Name)</tt>
3579 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
3580be aware of the <a href="#nameWarning">precaution above</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3581 </li>
3582 <li><tt>void replaceAllUsesWith(Value *V)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3583
3584 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
3585 href="#User"><tt>User</tt>s</a> of the current value to refer to
3586 "<tt>V</tt>" instead. For example, if you detect that an instruction always
3587 produces a constant value (for example through constant folding), you can
3588 replace all uses of the instruction with the constant like this:</p>
3589
Bill Wendling3cd5ca62006-10-11 06:30:10 +0000[diff] [blame]3590<div class="doc_code">
3591<pre>
3592Inst-&gt;replaceAllUsesWith(ConstVal);
3593</pre>
3594</div>
3595
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3596</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3597
3598</div>
3599
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3600</div>
3601
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3602<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3603<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3604 <a name="User">The <tt>User</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3605</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3606
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3607<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3608
3609<p>
3610<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]3611doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3612Superclass: <a href="#Value"><tt>Value</tt></a></p>
3613
3614<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
3615refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
3616that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
3617referring to. The <tt>User</tt> class itself is a subclass of
3618<tt>Value</tt>.</p>
3619
3620<p>The operands of a <tt>User</tt> point directly to the LLVM <a
3621href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
3622Single Assignment (SSA) form, there can only be one definition referred to,
3623allowing this direct connection. This connection provides the use-def
3624information in LLVM.</p>
3625
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3626<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3627<h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3628 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3629</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3630
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3631<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3632
3633<p>The <tt>User</tt> class exposes the operand list in two ways: through
3634an index access interface and through an iterator based interface.</p>
3635
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3636<ul>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3637 <li><tt>Value *getOperand(unsigned i)</tt><br>
3638 <tt>unsigned getNumOperands()</tt>
3639 <p> These two methods expose the operands of the <tt>User</tt> in a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3640convenient form for direct access.</p></li>
3641
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3642 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
3643list<br>
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]3644 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
3645the operand list.<br>
3646 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3647operand list.
3648 <p> Together, these methods make up the iterator based interface to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3649the operands of a <tt>User</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3650</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3651
3652</div>
3653
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3654</div>
3655
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3656<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3657<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3658 <a name="Instruction">The <tt>Instruction</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3659</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3660
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3661<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3662
3663<p><tt>#include "</tt><tt><a
3664href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]3665doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3666Superclasses: <a href="#User"><tt>User</tt></a>, <a
3667href="#Value"><tt>Value</tt></a></p>
3668
3669<p>The <tt>Instruction</tt> class is the common base class for all LLVM
3670instructions. It provides only a few methods, but is a very commonly used
3671class. The primary data tracked by the <tt>Instruction</tt> class itself is the
3672opcode (instruction type) and the parent <a
3673href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
3674into. To represent a specific type of instruction, one of many subclasses of
3675<tt>Instruction</tt> are used.</p>
3676
3677<p> Because the <tt>Instruction</tt> class subclasses the <a
3678href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
3679way as for other <a href="#User"><tt>User</tt></a>s (with the
3680<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
3681<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
3682the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
3683file contains some meta-data about the various different types of instructions
3684in LLVM. It describes the enum values that are used as opcodes (for example
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]3685<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3686concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
3687example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]3688href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3689this file confuses doxygen, so these enum values don't show up correctly in the
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]3690<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3691
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3692<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3693<h4>
3694 <a name="s_Instruction">
3695 Important Subclasses of the <tt>Instruction</tt> class
3696 </a>
3697</h4>
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3698<div>
Reid Spencerc92d25d2006-12-19 19:47:19 +0000[diff] [blame]3699 <ul>
3700 <li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
3701 <p>This subclasses represents all two operand instructions whose operands
3702 must be the same type, except for the comparison instructions.</p></li>
3703 <li><tt><a name="CastInst">CastInst</a></tt>
3704 <p>This subclass is the parent of the 12 casting instructions. It provides
3705 common operations on cast instructions.</p>
3706 <li><tt><a name="CmpInst">CmpInst</a></tt>
3707 <p>This subclass respresents the two comparison instructions,
3708 <a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
3709 <a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
3710 <li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
3711 <p>This subclass is the parent of all terminator instructions (those which
3712 can terminate a block).</p>
3713 </ul>
3714 </div>
3715
3716<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3717<h4>
3718 <a name="m_Instruction">
3719 Important Public Members of the <tt>Instruction</tt> class
3720 </a>
3721</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3722
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3723<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3724
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3725<ul>
3726 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3727 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
3728this <tt>Instruction</tt> is embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3729 <li><tt>bool mayWriteToMemory()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3730 <p>Returns true if the instruction writes to memory, i.e. it is a
3731 <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]3732 <li><tt>unsigned getOpcode()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3733 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3734 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3735 <p>Returns another instance of the specified instruction, identical
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3736in all ways to the original except that the instruction has no parent
3737(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3738and it has no name</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3739</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3740
3741</div>
3742
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3743</div>
3744
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3745<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3746<h3>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3747 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3748</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3749
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3750<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3751
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3752<p>Constant represents a base class for different types of constants. It
3753is subclassed by ConstantInt, ConstantArray, etc. for representing
3754the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
3755a subclass, which represents the address of a global variable or function.
3756</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3757
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3758<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3759<h4>Important Subclasses of Constant</h4>
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3760<div>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3761<ul>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3762 <li>ConstantInt : This subclass of Constant represents an integer constant of
3763 any width.
3764 <ul>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]3765 <li><tt>const APInt&amp; getValue() const</tt>: Returns the underlying
3766 value of this constant, an APInt value.</li>
3767 <li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
3768 value to an int64_t via sign extension. If the value (not the bit width)
3769 of the APInt is too large to fit in an int64_t, an assertion will result.
3770 For this reason, use of this method is discouraged.</li>
3771 <li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
3772 value to a uint64_t via zero extension. IF the value (not the bit width)
3773 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]3774 For this reason, use of this method is discouraged.</li>
Reid Spencer97b4ee32007-03-01 21:05:33 +0000[diff] [blame]3775 <li><tt>static ConstantInt* get(const APInt&amp; Val)</tt>: Returns the
3776 ConstantInt object that represents the value provided by <tt>Val</tt>.
3777 The type is implied as the IntegerType that corresponds to the bit width
3778 of <tt>Val</tt>.</li>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3779 <li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
3780 Returns the ConstantInt object that represents the value provided by
3781 <tt>Val</tt> for integer type <tt>Ty</tt>.</li>
3782 </ul>
3783 </li>
3784 <li>ConstantFP : This class represents a floating point constant.
3785 <ul>
3786 <li><tt>double getValue() const</tt>: Returns the underlying value of
3787 this constant. </li>
3788 </ul>
3789 </li>
3790 <li>ConstantArray : This represents a constant array.
3791 <ul>
3792 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
3793 a vector of component constants that makeup this array. </li>
3794 </ul>
3795 </li>
3796 <li>ConstantStruct : This represents a constant struct.
3797 <ul>
3798 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
3799 a vector of component constants that makeup this array. </li>
3800 </ul>
3801 </li>
3802 <li>GlobalValue : This represents either a global variable or a function. In
3803 either case, the value is a constant fixed address (after linking).
3804 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3805</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3806</div>
3807
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3808</div>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]3809
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3810<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3811<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3812 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3813</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3814
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3815<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3816
3817<p><tt>#include "<a
3818href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
Misha Brukman384047f2004-06-03 23:29:12 +0000[diff] [blame]3819doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
3820Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3821Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
3822<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]3823
3824<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
3825href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
3826visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
3827Because they are visible at global scope, they are also subject to linking with
3828other globals defined in different translation units. To control the linking
3829process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
3830<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]3831defined by the <tt>LinkageTypes</tt> enumeration.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3832
3833<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
3834<tt>static</tt> in C), it is not visible to code outside the current translation
3835unit, and does not participate in linking. If it has external linkage, it is
3836visible to external code, and does participate in linking. In addition to
3837linkage information, <tt>GlobalValue</tt>s keep track of which <a
3838href="#Module"><tt>Module</tt></a> they are currently part of.</p>
3839
3840<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
3841by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
3842global is always a pointer to its contents. It is important to remember this
3843when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
3844be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
3845subclass 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]3846i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3847the address of the first element of this array and the value of the
3848<tt>GlobalVariable</tt> are the same, they have different types. The
Reid Spencer06565dc2007-01-12 17:11:23 +0000[diff] [blame]3849<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
3850is <tt>i32.</tt> Because of this, accessing a global value requires you to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3851dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
3852can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
3853Language Reference Manual</a>.</p>
3854
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3855<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3856<h4>
3857 <a name="m_GlobalValue">
3858 Important Public Members of the <tt>GlobalValue</tt> class
3859 </a>
3860</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3861
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3862<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3863
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3864<ul>
3865 <li><tt>bool hasInternalLinkage() const</tt><br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3866 <tt>bool hasExternalLinkage() const</tt><br>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3867 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
3868 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
3869 <p> </p>
3870 </li>
3871 <li><tt><a href="#Module">Module</a> *getParent()</tt>
3872 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3873GlobalValue is currently embedded into.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3874</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3875
3876</div>
3877
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3878</div>
3879
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3880<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3881<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3882 <a name="Function">The <tt>Function</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3883</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3884
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3885<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3886
3887<p><tt>#include "<a
3888href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
Misha Brukman31ca1de2004-06-03 23:35:54 +0000[diff] [blame]3889info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3890Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3891<a href="#Constant"><tt>Constant</tt></a>,
3892<a href="#User"><tt>User</tt></a>,
3893<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3894
3895<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
Torok Edwin87469292009-10-12 13:37:29 +0000[diff] [blame]3896actually one of the more complex classes in the LLVM hierarchy because it must
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3897keep track of a large amount of data. The <tt>Function</tt> class keeps track
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]3898of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
3899<a href="#Argument"><tt>Argument</tt></a>s, and a
3900<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3901
3902<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
3903commonly used part of <tt>Function</tt> objects. The list imposes an implicit
3904ordering of the blocks in the function, which indicate how the code will be
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]3905laid out by the backend. Additionally, the first <a
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3906href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
3907<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
3908block. There are no implicit exit nodes, and in fact there may be multiple exit
3909nodes from a single <tt>Function</tt>. If the <a
3910href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
3911the <tt>Function</tt> is actually a function declaration: the actual body of the
3912function hasn't been linked in yet.</p>
3913
3914<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
3915<tt>Function</tt> class also keeps track of the list of formal <a
3916href="#Argument"><tt>Argument</tt></a>s that the function receives. This
3917container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
3918nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
3919the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
3920
3921<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
3922LLVM feature that is only used when you have to look up a value by name. Aside
3923from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
3924internally to make sure that there are not conflicts between the names of <a
3925href="#Instruction"><tt>Instruction</tt></a>s, <a
3926href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
3927href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
3928
Reid Spencer8b2da7a2004-07-18 13:10:31 +0000[diff] [blame]3929<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
3930and therefore also a <a href="#Constant">Constant</a>. The value of the function
3931is its address (after linking) which is guaranteed to be constant.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3932
3933<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3934<h4>
3935 <a name="m_Function">
3936 Important Public Members of the <tt>Function</tt> class
3937 </a>
3938</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3939
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3940<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3941
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3942<ul>
3943 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]3944 *Ty, LinkageTypes Linkage, const std::string &amp;N = "", Module* Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3945
3946 <p>Constructor used when you need to create new <tt>Function</tt>s to add
3947 the the program. The constructor must specify the type of the function to
Chris Lattnerac479e52004-08-04 05:10:48 +0000[diff] [blame]3948 create and what type of linkage the function should have. The <a
3949 href="#FunctionType"><tt>FunctionType</tt></a> argument
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3950 specifies the formal arguments and return value for the function. The same
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3951 <a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3952 create multiple functions. The <tt>Parent</tt> argument specifies the Module
3953 in which the function is defined. If this argument is provided, the function
3954 will automatically be inserted into that module's list of
3955 functions.</p></li>
3956
Chris Lattner62810e32008-11-25 18:34:50 +0000[diff] [blame]3957 <li><tt>bool isDeclaration()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3958
3959 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
3960 function is "external", it does not have a body, and thus must be resolved
3961 by linking with a function defined in a different translation unit.</p></li>
3962
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3963 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]3964 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3965
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]3966 <tt>begin()</tt>, <tt>end()</tt>
3967 <tt>size()</tt>, <tt>empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3968
3969 <p>These are forwarding methods that make it easy to access the contents of
3970 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
3971 list.</p></li>
3972
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3973 <li><tt>Function::BasicBlockListType &amp;getBasicBlockList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3974
3975 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
3976 is necessary to use when you need to update the list or perform a complex
3977 action that doesn't have a forwarding method.</p></li>
3978
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]3979 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3980iterator<br>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]3981 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3982
Chris Lattner77d69242005-03-15 05:19:20 +0000[diff] [blame]3983 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
Chris Lattner89cc2652005-03-15 04:48:32 +0000[diff] [blame]3984 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3985
3986 <p>These are forwarding methods that make it easy to access the contents of
3987 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
3988 list.</p></li>
3989
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3990 <li><tt>Function::ArgumentListType &amp;getArgumentList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3991
3992 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
3993 necessary to use when you need to update the list or perform a complex
3994 action that doesn't have a forwarding method.</p></li>
3995
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3996 <li><tt><a href="#BasicBlock">BasicBlock</a> &amp;getEntryBlock()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]3997
3998 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
3999 function. Because the entry block for the function is always the first
4000 block, this returns the first block of the <tt>Function</tt>.</p></li>
4001
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4002 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
4003 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4004
4005 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
4006 <tt>Function</tt> and returns the return type of the function, or the <a
4007 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
4008 function.</p></li>
4009
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4010 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4011
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4012 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4013 for this <tt>Function</tt>.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4014</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4015
4016</div>
4017
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4018</div>
4019
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4020<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4021<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4022 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4023</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4024
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4025<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4026
4027<p><tt>#include "<a
4028href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
4029<br>
Tanya Lattnera3da7772004-06-22 08:02:25 +0000[diff] [blame]4030doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]4031 Class</a><br>
4032Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
4033<a href="#Constant"><tt>Constant</tt></a>,
4034<a href="#User"><tt>User</tt></a>,
4035<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4036
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]4037<p>Global variables are represented with the (surprise surprise)
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4038<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
4039subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
4040always referenced by their address (global values must live in memory, so their
Reid Spencerbe5e85e2006-04-14 14:11:48 +0000[diff] [blame]4041"name" refers to their constant address). See
4042<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
4043variables may have an initial value (which must be a
4044<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
4045they may be marked as "constant" themselves (indicating that their contents
4046never change at runtime).</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4047
4048<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4049<h4>
4050 <a name="m_GlobalVariable">
4051 Important Public Members of the <tt>GlobalVariable</tt> class
4052 </a>
4053</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4054
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4055<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4056
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4057<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4058 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
4059 isConstant, LinkageTypes&amp; Linkage, <a href="#Constant">Constant</a>
4060 *Initializer = 0, const std::string &amp;Name = "", Module* Parent = 0)</tt>
4061
4062 <p>Create a new global variable of the specified type. If
4063 <tt>isConstant</tt> is true then the global variable will be marked as
4064 unchanging for the program. The Linkage parameter specifies the type of
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]4065 linkage (internal, external, weak, linkonce, appending) for the variable.
4066 If the linkage is InternalLinkage, WeakAnyLinkage, WeakODRLinkage,
4067 LinkOnceAnyLinkage or LinkOnceODRLinkage,&nbsp; then the resultant
4068 global variable will have internal linkage. AppendingLinkage concatenates
4069 together all instances (in different translation units) of the variable
4070 into a single variable but is only applicable to arrays. &nbsp;See
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4071 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
4072 further details on linkage types. Optionally an initializer, a name, and the
4073 module to put the variable into may be specified for the global variable as
4074 well.</p></li>
4075
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4076 <li><tt>bool isConstant() const</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4077
4078 <p>Returns true if this is a global variable that is known not to
4079 be modified at runtime.</p></li>
4080
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4081 <li><tt>bool hasInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4082
4083 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
4084
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4085 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4086
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]4087 <p>Returns the initial value for a <tt>GlobalVariable</tt>. It is not legal
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4088 to call this method if there is no initializer.</p></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4089</ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4090
4091</div>
4092
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4093</div>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4094
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4095<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4096<h3>
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4097 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4098</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4099
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4100<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4101
4102<p><tt>#include "<a
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4103href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
Stefanus Du Toit24e04112009-06-17 21:12:26 +0000[diff] [blame]4104doxygen info: <a href="/doxygen/classllvm_1_1BasicBlock.html">BasicBlock
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4105Class</a><br>
4106Superclass: <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4107
Nick Lewyckyccd279d2011-02-17 02:19:22 +0000[diff] [blame]4108<p>This class represents a single entry single exit section of the code,
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4109commonly known as a basic block by the compiler community. The
4110<tt>BasicBlock</tt> class maintains a list of <a
4111href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
4112Matching the language definition, the last element of this list of instructions
4113is always a terminator instruction (a subclass of the <a
4114href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
4115
4116<p>In addition to tracking the list of instructions that make up the block, the
4117<tt>BasicBlock</tt> class also keeps track of the <a
4118href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
4119
4120<p>Note that <tt>BasicBlock</tt>s themselves are <a
4121href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
4122like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
4123<tt>label</tt>.</p>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4124
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4125<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4126<h4>
4127 <a name="m_BasicBlock">
4128 Important Public Members of the <tt>BasicBlock</tt> class
4129 </a>
4130</h4>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4131
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4132<div>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4133<ul>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4134
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4135<li><tt>BasicBlock(const std::string &amp;Name = "", </tt><tt><a
4136 href="#Function">Function</a> *Parent = 0)</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4137
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4138<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
4139insertion into a function. The constructor optionally takes a name for the new
4140block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
4141the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
4142automatically inserted at the end of the specified <a
4143href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
4144manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4145
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4146<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
4147<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
4148<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
4149<tt>size()</tt>, <tt>empty()</tt>
4150STL-style functions for accessing the instruction list.
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4151
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4152<p>These methods and typedefs are forwarding functions that have the same
4153semantics as the standard library methods of the same names. These methods
4154expose the underlying instruction list of a basic block in a way that is easy to
4155manipulate. To get the full complement of container operations (including
4156operations to update the list), you must use the <tt>getInstList()</tt>
4157method.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4158
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4159<li><tt>BasicBlock::InstListType &amp;getInstList()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4160
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4161<p>This method is used to get access to the underlying container that actually
4162holds the Instructions. This method must be used when there isn't a forwarding
4163function in the <tt>BasicBlock</tt> class for the operation that you would like
4164to perform. Because there are no forwarding functions for "updating"
4165operations, you need to use this if you want to update the contents of a
4166<tt>BasicBlock</tt>.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4167
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4168<li><tt><a href="#Function">Function</a> *getParent()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4169
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4170<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
4171embedded into, or a null pointer if it is homeless.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4172
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4173<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4174
Chris Lattner2b78d962007-02-03 20:02:25 +0000[diff] [blame]4175<p> Returns a pointer to the terminator instruction that appears at the end of
4176the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
4177instruction in the block is not a terminator, then a null pointer is
4178returned.</p></li>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4179
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4180</ul>
4181
4182</div>
4183
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4184</div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4185
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4186<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4187<h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4188 <a name="Argument">The <tt>Argument</tt> class</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4189</h3>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4190
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4191<div>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4192
4193<p>This subclass of Value defines the interface for incoming formal
Chris Lattner58360822005-01-17 00:12:04 +0000[diff] [blame]4194arguments to a function. A Function maintains a list of its formal
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4195arguments. An argument has a pointer to the parent Function.</p>
4196
4197</div>
4198
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4199</div>
4200
Chris Lattner9355b472002-09-06 02:50:58 +0000[diff] [blame]4201<!-- *********************************************************************** -->
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4202<hr>
4203<address>
4204 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]4205 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4206 <a href="http://validator.w3.org/check/referer"><img
Gabor Greifa9c0f2b2008-06-18 14:05:31 +0000[diff] [blame]4207 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]4208
4209 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
4210 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
NAKAMURA Takumib9a33632011-04-09 02:13:37 +0000[diff] [blame]4211 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Misha Brukman13fd15c2004-01-15 00:14:41 +0000[diff] [blame]4212 Last modified: $Date$
4213</address>
4214
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4215</body>
4216</html>