Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / ed58a95c33737ffe52745b3a4c17c8ec0b092402 / . / llvm / docs / ProgrammersManual.html
blob: 3bec325d6ce078240c539eeb4491b0a12e3644fc [file] [log] [blame]
Misha Brukman7f67e372004-01-15 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 Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3<html>
4<head>
Owen Andersoneb105f92009-06-16 17:40:28 +0000[diff] [blame]5 <meta http-equiv="Content-type" content="text/html;charset=UTF-8">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]6 <title>LLVM Programmer's Manual</title>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]7 <link rel="stylesheet" href="llvm.css" type="text/css">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]8</head>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]9<body>
10
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]11<h1>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]12 LLVM Programmer's Manual
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]13</h1>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]14
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]15<ol>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]16 <li><a href="#introduction">Introduction</a></li>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]17 <li><a href="#general">General Information</a>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]18 <ul>
Reid Spencerf0714dd2004-11-01 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 Lattner292e6602004-12-08 19:05:44 +0000[diff] [blame]24
Reid Spencerf0714dd2004-11-01 09:02:53 +0000[diff] [blame]25-->
Chris Lattner5a0c4c62003-08-01 22:20:59 +0000[diff] [blame]26 </ul>
Chris Lattner48b383b02003-11-25 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 Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]32 <li><a href="#string_apis">Passing strings (the <tt>StringRef</tt>
Benjamin Kramereaccdd32009-08-05 15:42:44 +0000[diff] [blame]33and <tt>Twine</tt> classes)</a>
Daniel Dunbar4975db62009-07-25 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 Kramereaccdd32009-08-05 15:42:44 +0000[diff] [blame]38 </li>
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]39 <li><a href="#DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt>
Chris Lattner48b383b02003-11-25 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 Lattner58e53452006-12-19 21:46:21 +0000[diff] [blame]46 <li><a href="#Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Reid Spencerf0714dd2004-11-01 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 Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]52 <li><a href="#ViewGraph">Viewing graphs while debugging code</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]53 </ul>
54 </li>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]55 <li><a href="#datastructure">Picking the Right Data Structure for a Task</a>
56 <ul>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]57 <li><a href="#ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
58 <ul>
Chris Lattnera76507f2011-04-05 23:18:20 +0000[diff] [blame]59 <li><a href="#dss_arrayref">llvm/ADT/ArrayRef.h</a></li>
Chris Lattnerac760292007-02-03 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>
62 <li><a href="#dss_smallvector">"llvm/ADT/SmallVector.h"</a></li>
63 <li><a href="#dss_vector">&lt;vector&gt;</a></li>
64 <li><a href="#dss_deque">&lt;deque&gt;</a></li>
65 <li><a href="#dss_list">&lt;list&gt;</a></li>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]66 <li><a href="#dss_ilist">llvm/ADT/ilist.h</a></li>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]67 <li><a href="#dss_other">Other Sequential Container Options</a></li>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]68 </ul></li>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]69 <li><a href="#ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
70 <ul>
71 <li><a href="#dss_sortedvectorset">A sorted 'vector'</a></li>
72 <li><a href="#dss_smallset">"llvm/ADT/SmallSet.h"</a></li>
73 <li><a href="#dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a></li>
Chris Lattnerbd0079c2007-09-30 00:58:59 +0000[diff] [blame]74 <li><a href="#dss_denseset">"llvm/ADT/DenseSet.h"</a></li>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]75 <li><a href="#dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a></li>
76 <li><a href="#dss_set">&lt;set&gt;</a></li>
77 <li><a href="#dss_setvector">"llvm/ADT/SetVector.h"</a></li>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]78 <li><a href="#dss_uniquevector">"llvm/ADT/UniqueVector.h"</a></li>
79 <li><a href="#dss_otherset">Other Set-Like ContainerOptions</a></li>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]80 </ul></li>
Chris Lattner4e406a92007-02-03 19:51:56 +0000[diff] [blame]81 <li><a href="#ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
82 <ul>
83 <li><a href="#dss_sortedvectormap">A sorted 'vector'</a></li>
Chris Lattnerd81d7be2007-02-08 19:14:21 +0000[diff] [blame]84 <li><a href="#dss_stringmap">"llvm/ADT/StringMap.h"</a></li>
Chris Lattner4e406a92007-02-03 19:51:56 +0000[diff] [blame]85 <li><a href="#dss_indexedmap">"llvm/ADT/IndexedMap.h"</a></li>
86 <li><a href="#dss_densemap">"llvm/ADT/DenseMap.h"</a></li>
Jeffrey Yasskin4546d312009-10-22 22:11:22 +0000[diff] [blame]87 <li><a href="#dss_valuemap">"llvm/ADT/ValueMap.h"</a></li>
Jakob Stoklund Olesen36eab1c2010-12-14 00:55:51 +0000[diff] [blame]88 <li><a href="#dss_intervalmap">"llvm/ADT/IntervalMap.h"</a></li>
Chris Lattner4e406a92007-02-03 19:51:56 +0000[diff] [blame]89 <li><a href="#dss_map">&lt;map&gt;</a></li>
Jakob Stoklund Olesen04123d42011-04-05 20:56:08 +0000[diff] [blame]90 <li><a href="#dss_inteqclasses">"llvm/ADT/IntEqClasses.h"</a></li>
Chris Lattner4e406a92007-02-03 19:51:56 +0000[diff] [blame]91 <li><a href="#dss_othermap">Other Map-Like Container Options</a></li>
92 </ul></li>
Chris Lattner47bb37c2009-07-25 07:22:20 +0000[diff] [blame]93 <li><a href="#ds_string">String-like containers</a>
Benjamin Kramereaccdd32009-08-05 15:42:44 +0000[diff] [blame]94 <!--<ul>
95 todo
96 </ul>--></li>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]97 <li><a href="#ds_bit">BitVector-like containers</a>
98 <ul>
99 <li><a href="#dss_bitvector">A dense bitvector</a></li>
Dan Gohman60b0a862010-01-05 18:24:00 +0000[diff] [blame]100 <li><a href="#dss_smallbitvector">A "small" dense bitvector</a></li>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]101 <li><a href="#dss_sparsebitvector">A sparse bitvector</a></li>
102 </ul></li>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]103 </ul>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]104 </li>
Chris Lattnercb29fd52002-09-06 18:31:18 +0000[diff] [blame]105 <li><a href="#common">Helpful Hints for Common Operations</a>
Chris Lattnercb29fd52002-09-06 18:31:18 +0000[diff] [blame]106 <ul>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]107 <li><a href="#inspection">Basic Inspection and Traversal Routines</a>
108 <ul>
109 <li><a href="#iterate_function">Iterating over the <tt>BasicBlock</tt>s
110in a <tt>Function</tt></a> </li>
111 <li><a href="#iterate_basicblock">Iterating over the <tt>Instruction</tt>s
112in a <tt>BasicBlock</tt></a> </li>
113 <li><a href="#iterate_institer">Iterating over the <tt>Instruction</tt>s
114in a <tt>Function</tt></a> </li>
115 <li><a href="#iterate_convert">Turning an iterator into a
116class pointer</a> </li>
117 <li><a href="#iterate_complex">Finding call sites: a more
118complex example</a> </li>
119 <li><a href="#calls_and_invokes">Treating calls and invokes
120the same way</a> </li>
121 <li><a href="#iterate_chains">Iterating over def-use &amp;
122use-def chains</a> </li>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]123 <li><a href="#iterate_preds">Iterating over predecessors &amp;
124successors of blocks</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]125 </ul>
126 </li>
127 <li><a href="#simplechanges">Making simple changes</a>
128 <ul>
129 <li><a href="#schanges_creating">Creating and inserting new
130 <tt>Instruction</tt>s</a> </li>
131 <li><a href="#schanges_deleting">Deleting <tt>Instruction</tt>s</a> </li>
132 <li><a href="#schanges_replacing">Replacing an <tt>Instruction</tt>
133with another <tt>Value</tt></a> </li>
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]134 <li><a href="#schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a> </li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]135 </ul>
Reid Spencerf0714dd2004-11-01 09:02:53 +0000[diff] [blame]136 </li>
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]137 <li><a href="#create_types">How to Create Types</a></li>
Chris Lattnercb29fd52002-09-06 18:31:18 +0000[diff] [blame]138<!--
139 <li>Working with the Control Flow Graph
140 <ul>
141 <li>Accessing predecessors and successors of a <tt>BasicBlock</tt>
142 <li>
143 <li>
144 </ul>
Reid Spencerf0714dd2004-11-01 09:02:53 +0000[diff] [blame]145-->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]146 </ul>
147 </li>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]148
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]149 <li><a href="#threading">Threads and LLVM</a>
150 <ul>
Owen Anderson4741b572009-06-16 18:04:19 +0000[diff] [blame]151 <li><a href="#startmultithreaded">Entering and Exiting Multithreaded Mode
152 </a></li>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]153 <li><a href="#shutdown">Ending execution with <tt>llvm_shutdown()</tt></a></li>
154 <li><a href="#managedstatic">Lazy initialization with <tt>ManagedStatic</tt></a></li>
Owen Andersonfe7bbc92009-08-19 17:58:52 +0000[diff] [blame]155 <li><a href="#llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a></li>
Jeffrey Yasskin9fb8ce82010-01-29 19:10:38 +0000[diff] [blame]156 <li><a href="#jitthreading">Threads and the JIT</a></li>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]157 </ul>
158 </li>
159
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]160 <li><a href="#advanced">Advanced Topics</a>
161 <ul>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]162 <li><a href="#TypeResolve">LLVM Type Resolution</a>
163 <ul>
164 <li><a href="#BuildRecType">Basic Recursive Type Construction</a></li>
165 <li><a href="#refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a></li>
166 <li><a href="#PATypeHolder">The PATypeHolder Class</a></li>
167 <li><a href="#AbstractTypeUser">The AbstractTypeUser Class</a></li>
168 </ul></li>
169
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]170 <li><a href="#SymbolTable">The <tt>ValueSymbolTable</tt> and <tt>TypeSymbolTable</tt> classes</a></li>
171 <li><a href="#UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a></li>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]172 </ul></li>
173
Joel Stanley64cfdbb2002-09-06 21:55:13 +0000[diff] [blame]174 <li><a href="#coreclasses">The Core LLVM Class Hierarchy Reference</a>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]175 <ul>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]176 <li><a href="#Type">The <tt>Type</tt> class</a> </li>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]177 <li><a href="#Module">The <tt>Module</tt> class</a></li>
Reid Spencerf0714dd2004-11-01 09:02:53 +0000[diff] [blame]178 <li><a href="#Value">The <tt>Value</tt> class</a>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]179 <ul>
180 <li><a href="#User">The <tt>User</tt> class</a>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]181 <ul>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]182 <li><a href="#Instruction">The <tt>Instruction</tt> class</a></li>
183 <li><a href="#Constant">The <tt>Constant</tt> class</a>
184 <ul>
185 <li><a href="#GlobalValue">The <tt>GlobalValue</tt> class</a>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]186 <ul>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]187 <li><a href="#Function">The <tt>Function</tt> class</a></li>
188 <li><a href="#GlobalVariable">The <tt>GlobalVariable</tt> class</a></li>
189 </ul>
190 </li>
191 </ul>
192 </li>
Reid Spencerf0714dd2004-11-01 09:02:53 +0000[diff] [blame]193 </ul>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]194 </li>
195 <li><a href="#BasicBlock">The <tt>BasicBlock</tt> class</a></li>
196 <li><a href="#Argument">The <tt>Argument</tt> class</a></li>
197 </ul>
Reid Spencerf0714dd2004-11-01 09:02:53 +0000[diff] [blame]198 </li>
199 </ul>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]200 </li>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]201</ol>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]202
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]203<div class="doc_author">
204 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
Chris Lattnercc834172004-05-26 16:52:55 +0000[diff] [blame]205 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a>,
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]206 <a href="mailto:ggreif@gmail.com">Gabor Greif</a>,
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]207 <a href="mailto:jstanley@cs.uiuc.edu">Joel Stanley</a>,
208 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a> and
209 <a href="mailto:owen@apple.com">Owen Anderson</a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]210</div>
211
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]212<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]213<h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]214 <a name="introduction">Introduction </a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]215</h2>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]216<!-- *********************************************************************** -->
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]217
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]218<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]219
220<p>This document is meant to highlight some of the important classes and
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]221interfaces available in the LLVM source-base. This manual is not
222intended to explain what LLVM is, how it works, and what LLVM code looks
223like. It assumes that you know the basics of LLVM and are interested
224in writing transformations or otherwise analyzing or manipulating the
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]225code.</p>
226
227<p>This document should get you oriented so that you can find your
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]228way in the continuously growing source code that makes up the LLVM
229infrastructure. Note that this manual is not intended to serve as a
230replacement for reading the source code, so if you think there should be
231a method in one of these classes to do something, but it's not listed,
232check the source. Links to the <a href="/doxygen/">doxygen</a> sources
233are provided to make this as easy as possible.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]234
235<p>The first section of this document describes general information that is
236useful to know when working in the LLVM infrastructure, and the second describes
237the Core LLVM classes. In the future this manual will be extended with
238information describing how to use extension libraries, such as dominator
239information, CFG traversal routines, and useful utilities like the <tt><a
240href="/doxygen/InstVisitor_8h-source.html">InstVisitor</a></tt> template.</p>
241
242</div>
243
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]244<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]245<h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]246 <a name="general">General Information</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]247</h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]248<!-- *********************************************************************** -->
249
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]250<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]251
252<p>This section contains general information that is useful if you are working
253in the LLVM source-base, but that isn't specific to any particular API.</p>
254
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]255<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]256<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]257 <a name="stl">The C++ Standard Template Library</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]258</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]259
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]260<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]261
262<p>LLVM makes heavy use of the C++ Standard Template Library (STL),
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]263perhaps much more than you are used to, or have seen before. Because of
264this, you might want to do a little background reading in the
265techniques used and capabilities of the library. There are many good
266pages that discuss the STL, and several books on the subject that you
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]267can get, so it will not be discussed in this document.</p>
268
269<p>Here are some useful links:</p>
270
271<ol>
272
Nick Lewyckycf263b02010-10-09 21:12:29 +0000[diff] [blame]273<li><a href="http://www.dinkumware.com/manuals/#Standard C++ Library">Dinkumware
274C++ Library reference</a> - an excellent reference for the STL and other parts
275of the standard C++ library.</li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]276
277<li><a href="http://www.tempest-sw.com/cpp/">C++ In a Nutshell</a> - This is an
Gabor Greif4ef9bc02009-03-12 09:47:03 +0000[diff] [blame]278O'Reilly book in the making. It has a decent Standard Library
279Reference that rivals Dinkumware's, and is unfortunately no longer free since the
280book has been published.</li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]281
282<li><a href="http://www.parashift.com/c++-faq-lite/">C++ Frequently Asked
283Questions</a></li>
284
285<li><a href="http://www.sgi.com/tech/stl/">SGI's STL Programmer's Guide</a> -
286Contains a useful <a
287href="http://www.sgi.com/tech/stl/stl_introduction.html">Introduction to the
288STL</a>.</li>
289
290<li><a href="http://www.research.att.com/%7Ebs/C++.html">Bjarne Stroustrup's C++
291Page</a></li>
292
Tanya Lattner0e662212004-12-08 18:34:56 +0000[diff] [blame]293<li><a href="http://64.78.49.204/">
Reid Spencer9aed59b2004-05-26 08:41:35 +0000[diff] [blame]294Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 (even better, get
295the book).</a></li>
296
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]297</ol>
298
299<p>You are also encouraged to take a look at the <a
300href="CodingStandards.html">LLVM Coding Standards</a> guide which focuses on how
301to write maintainable code more than where to put your curly braces.</p>
302
303</div>
304
305<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]306<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]307 <a name="stl">Other useful references</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]308</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]309
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]310<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]311
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]312<ol>
Misha Brukman5f056c72004-06-18 18:39:00 +0000[diff] [blame]313<li><a href="http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html">Using
314static and shared libraries across platforms</a></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]315</ol>
316
317</div>
318
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]319</div>
320
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]321<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]322<h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]323 <a name="apis">Important and useful LLVM APIs</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]324</h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]325<!-- *********************************************************************** -->
326
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]327<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]328
329<p>Here we highlight some LLVM APIs that are generally useful and good to
330know about when writing transformations.</p>
331
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]332<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]333<h3>
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]334 <a name="isa">The <tt>isa&lt;&gt;</tt>, <tt>cast&lt;&gt;</tt> and
335 <tt>dyn_cast&lt;&gt;</tt> templates</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]336</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]337
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]338<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]339
340<p>The LLVM source-base makes extensive use of a custom form of RTTI.
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]341These templates have many similarities to the C++ <tt>dynamic_cast&lt;&gt;</tt>
342operator, but they don't have some drawbacks (primarily stemming from
343the fact that <tt>dynamic_cast&lt;&gt;</tt> only works on classes that
344have a v-table). Because they are used so often, you must know what they
345do and how they work. All of these templates are defined in the <a
Chris Lattner730a5da1a2005-04-26 22:56:16 +0000[diff] [blame]346 href="/doxygen/Casting_8h-source.html"><tt>llvm/Support/Casting.h</tt></a>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]347file (note that you very rarely have to include this file directly).</p>
348
349<dl>
350 <dt><tt>isa&lt;&gt;</tt>: </dt>
351
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]352 <dd><p>The <tt>isa&lt;&gt;</tt> operator works exactly like the Java
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]353 "<tt>instanceof</tt>" operator. It returns true or false depending on whether
354 a reference or pointer points to an instance of the specified class. This can
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]355 be very useful for constraint checking of various sorts (example below).</p>
356 </dd>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]357
358 <dt><tt>cast&lt;&gt;</tt>: </dt>
359
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]360 <dd><p>The <tt>cast&lt;&gt;</tt> operator is a "checked cast" operation. It
Chris Lattner643b00a2008-06-20 05:03:17 +0000[diff] [blame]361 converts a pointer or reference from a base class to a derived class, causing
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]362 an assertion failure if it is not really an instance of the right type. This
363 should be used in cases where you have some information that makes you believe
364 that something is of the right type. An example of the <tt>isa&lt;&gt;</tt>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]365 and <tt>cast&lt;&gt;</tt> template is:</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]366
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]367<div class="doc_code">
368<pre>
369static bool isLoopInvariant(const <a href="#Value">Value</a> *V, const Loop *L) {
370 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))
371 return true;
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]372
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]373 // <i>Otherwise, it must be an instruction...</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]374 return !L-&gt;contains(cast&lt;<a href="#Instruction">Instruction</a>&gt;(V)-&gt;getParent());
375}
376</pre>
377</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]378
379 <p>Note that you should <b>not</b> use an <tt>isa&lt;&gt;</tt> test followed
380 by a <tt>cast&lt;&gt;</tt>, for that use the <tt>dyn_cast&lt;&gt;</tt>
381 operator.</p>
382
383 </dd>
384
385 <dt><tt>dyn_cast&lt;&gt;</tt>:</dt>
386
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]387 <dd><p>The <tt>dyn_cast&lt;&gt;</tt> operator is a "checking cast" operation.
388 It checks to see if the operand is of the specified type, and if so, returns a
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]389 pointer to it (this operator does not work with references). If the operand is
390 not of the correct type, a null pointer is returned. Thus, this works very
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]391 much like the <tt>dynamic_cast&lt;&gt;</tt> operator in C++, and should be
392 used in the same circumstances. Typically, the <tt>dyn_cast&lt;&gt;</tt>
393 operator is used in an <tt>if</tt> statement or some other flow control
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]394 statement like this:</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]395
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]396<div class="doc_code">
397<pre>
398if (<a href="#AllocationInst">AllocationInst</a> *AI = dyn_cast&lt;<a href="#AllocationInst">AllocationInst</a>&gt;(Val)) {
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]399 // <i>...</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]400}
401</pre>
402</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]403
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]404 <p>This form of the <tt>if</tt> statement effectively combines together a call
405 to <tt>isa&lt;&gt;</tt> and a call to <tt>cast&lt;&gt;</tt> into one
406 statement, which is very convenient.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]407
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]408 <p>Note that the <tt>dyn_cast&lt;&gt;</tt> operator, like C++'s
409 <tt>dynamic_cast&lt;&gt;</tt> or Java's <tt>instanceof</tt> operator, can be
410 abused. In particular, you should not use big chained <tt>if/then/else</tt>
411 blocks to check for lots of different variants of classes. If you find
412 yourself wanting to do this, it is much cleaner and more efficient to use the
413 <tt>InstVisitor</tt> class to dispatch over the instruction type directly.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]414
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]415 </dd>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]416
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]417 <dt><tt>cast_or_null&lt;&gt;</tt>: </dt>
418
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]419 <dd><p>The <tt>cast_or_null&lt;&gt;</tt> operator works just like the
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]420 <tt>cast&lt;&gt;</tt> operator, except that it allows for a null pointer as an
421 argument (which it then propagates). This can sometimes be useful, allowing
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]422 you to combine several null checks into one.</p></dd>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]423
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]424 <dt><tt>dyn_cast_or_null&lt;&gt;</tt>: </dt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]425
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]426 <dd><p>The <tt>dyn_cast_or_null&lt;&gt;</tt> operator works just like the
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]427 <tt>dyn_cast&lt;&gt;</tt> operator, except that it allows for a null pointer
428 as an argument (which it then propagates). This can sometimes be useful,
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]429 allowing you to combine several null checks into one.</p></dd>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]430
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]431</dl>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]432
433<p>These five templates can be used with any classes, whether they have a
434v-table or not. To add support for these templates, you simply need to add
435<tt>classof</tt> static methods to the class you are interested casting
436to. Describing this is currently outside the scope of this document, but there
437are lots of examples in the LLVM source base.</p>
438
439</div>
440
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]441
442<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]443<h3>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]444 <a name="string_apis">Passing strings (the <tt>StringRef</tt>
445and <tt>Twine</tt> classes)</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]446</h3>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]447
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]448<div>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]449
450<p>Although LLVM generally does not do much string manipulation, we do have
Chris Lattner375a3f42009-07-25 07:16:59 +0000[diff] [blame]451several important APIs which take strings. Two important examples are the
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]452Value class -- which has names for instructions, functions, etc. -- and the
453StringMap class which is used extensively in LLVM and Clang.</p>
454
455<p>These are generic classes, and they need to be able to accept strings which
456may have embedded null characters. Therefore, they cannot simply take
Chris Lattner375a3f42009-07-25 07:16:59 +0000[diff] [blame]457a <tt>const char *</tt>, and taking a <tt>const std::string&amp;</tt> requires
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]458clients to perform a heap allocation which is usually unnecessary. Instead,
Benjamin Kramer92d89982010-07-14 22:38:02 +0000[diff] [blame]459many LLVM APIs use a <tt>StringRef</tt> or a <tt>const Twine&amp;</tt> for
460passing strings efficiently.</p>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]461
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]462<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]463<h4>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]464 <a name="StringRef">The <tt>StringRef</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]465</h4>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]466
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]467<div>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]468
469<p>The <tt>StringRef</tt> data type represents a reference to a constant string
470(a character array and a length) and supports the common operations available
471on <tt>std:string</tt>, but does not require heap allocation.</p>
472
Chris Lattner375a3f42009-07-25 07:16:59 +0000[diff] [blame]473<p>It can be implicitly constructed using a C style null-terminated string,
474an <tt>std::string</tt>, or explicitly with a character pointer and length.
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]475For example, the <tt>StringRef</tt> find function is declared as:</p>
Chris Lattner375a3f42009-07-25 07:16:59 +0000[diff] [blame]476
Benjamin Kramer92d89982010-07-14 22:38:02 +0000[diff] [blame]477<pre class="doc_code">
478 iterator find(StringRef Key);
479</pre>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]480
481<p>and clients can call it using any one of:</p>
482
Benjamin Kramer92d89982010-07-14 22:38:02 +0000[diff] [blame]483<pre class="doc_code">
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]484 Map.find("foo"); <i>// Lookup "foo"</i>
485 Map.find(std::string("bar")); <i>// Lookup "bar"</i>
486 Map.find(StringRef("\0baz", 4)); <i>// Lookup "\0baz"</i>
487</pre>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]488
489<p>Similarly, APIs which need to return a string may return a <tt>StringRef</tt>
490instance, which can be used directly or converted to an <tt>std::string</tt>
491using the <tt>str</tt> member function. See
492"<tt><a href="/doxygen/classllvm_1_1StringRef_8h-source.html">llvm/ADT/StringRef.h</a></tt>"
493for more information.</p>
494
495<p>You should rarely use the <tt>StringRef</tt> class directly, because it contains
496pointers to external memory it is not generally safe to store an instance of the
Benjamin Kramer92d89982010-07-14 22:38:02 +0000[diff] [blame]497class (unless you know that the external storage will not be freed). StringRef is
498small and pervasive enough in LLVM that it should always be passed by value.</p>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]499
500</div>
501
502<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]503<h4>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]504 <a name="Twine">The <tt>Twine</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]505</h4>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]506
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]507<div>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]508
509<p>The <tt>Twine</tt> class is an efficient way for APIs to accept concatenated
510strings. For example, a common LLVM paradigm is to name one instruction based on
511the name of another instruction with a suffix, for example:</p>
512
513<div class="doc_code">
514<pre>
515 New = CmpInst::Create(<i>...</i>, SO->getName() + ".cmp");
516</pre>
517</div>
518
519<p>The <tt>Twine</tt> class is effectively a
520lightweight <a href="http://en.wikipedia.org/wiki/Rope_(computer_science)">rope</a>
521which points to temporary (stack allocated) objects. Twines can be implicitly
522constructed as the result of the plus operator applied to strings (i.e., a C
523strings, an <tt>std::string</tt>, or a <tt>StringRef</tt>). The twine delays the
Dan Gohmand0c7fd82010-02-25 23:51:27 +0000[diff] [blame]524actual concatenation of strings until it is actually required, at which point
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]525it can be efficiently rendered directly into a character array. This avoids
526unnecessary heap allocation involved in constructing the temporary results of
527string concatenation. See
528"<tt><a href="/doxygen/classllvm_1_1Twine_8h-source.html">llvm/ADT/Twine.h</a></tt>"
Benjamin Kramereaccdd32009-08-05 15:42:44 +0000[diff] [blame]529for more information.</p>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]530
531<p>As with a <tt>StringRef</tt>, <tt>Twine</tt> objects point to external memory
532and should almost never be stored or mentioned directly. They are intended
533solely for use when defining a function which should be able to efficiently
534accept concatenated strings.</p>
535
536</div>
537
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]538</div>
Daniel Dunbar4975db62009-07-25 04:41:11 +0000[diff] [blame]539
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]540<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]541<h3>
Misha Brukman93b8cb12005-11-01 21:12:49 +0000[diff] [blame]542 <a name="DEBUG">The <tt>DEBUG()</tt> macro and <tt>-debug</tt> option</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]543</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]544
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]545<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]546
547<p>Often when working on your pass you will put a bunch of debugging printouts
548and other code into your pass. After you get it working, you want to remove
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]549it, but you may need it again in the future (to work out new bugs that you run
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]550across).</p>
551
552<p> Naturally, because of this, you don't want to delete the debug printouts,
553but you don't want them to always be noisy. A standard compromise is to comment
554them out, allowing you to enable them if you need them in the future.</p>
555
Chris Lattner730a5da1a2005-04-26 22:56:16 +0000[diff] [blame]556<p>The "<tt><a href="/doxygen/Debug_8h-source.html">llvm/Support/Debug.h</a></tt>"
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]557file provides a macro named <tt>DEBUG()</tt> that is a much nicer solution to
558this problem. Basically, you can put arbitrary code into the argument of the
559<tt>DEBUG</tt> macro, and it is only executed if '<tt>opt</tt>' (or any other
560tool) is run with the '<tt>-debug</tt>' command line argument:</p>
561
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]562<div class="doc_code">
563<pre>
Daniel Dunbar04bbd9c2009-07-25 01:55:32 +0000[diff] [blame]564DEBUG(errs() &lt;&lt; "I am here!\n");
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]565</pre>
566</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]567
568<p>Then you can run your pass like this:</p>
569
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]570<div class="doc_code">
571<pre>
572$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]573<i>&lt;no output&gt;</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]574$ opt &lt; a.bc &gt; /dev/null -mypass -debug
575I am here!
576</pre>
577</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]578
579<p>Using the <tt>DEBUG()</tt> macro instead of a home-brewed solution allows you
580to not have to create "yet another" command line option for the debug output for
581your pass. Note that <tt>DEBUG()</tt> macros are disabled for optimized builds,
582so they do not cause a performance impact at all (for the same reason, they
583should also not contain side-effects!).</p>
584
585<p>One additional nice thing about the <tt>DEBUG()</tt> macro is that you can
586enable or disable it directly in gdb. Just use "<tt>set DebugFlag=0</tt>" or
587"<tt>set DebugFlag=1</tt>" from the gdb if the program is running. If the
588program hasn't been started yet, you can always just run it with
589<tt>-debug</tt>.</p>
590
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]591<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]592<h4>
Chris Lattner2bfe3622005-04-26 22:57:07 +0000[diff] [blame]593 <a name="DEBUG_TYPE">Fine grained debug info with <tt>DEBUG_TYPE</tt> and
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]594 the <tt>-debug-only</tt> option</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]595</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]596
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]597<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]598
599<p>Sometimes you may find yourself in a situation where enabling <tt>-debug</tt>
600just turns on <b>too much</b> information (such as when working on the code
601generator). If you want to enable debug information with more fine-grained
602control, you define the <tt>DEBUG_TYPE</tt> macro and the <tt>-debug</tt> only
603option as follows:</p>
604
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]605<div class="doc_code">
606<pre>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]607#undef DEBUG_TYPE
Daniel Dunbar04bbd9c2009-07-25 01:55:32 +0000[diff] [blame]608DEBUG(errs() &lt;&lt; "No debug type\n");
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]609#define DEBUG_TYPE "foo"
Daniel Dunbar04bbd9c2009-07-25 01:55:32 +0000[diff] [blame]610DEBUG(errs() &lt;&lt; "'foo' debug type\n");
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]611#undef DEBUG_TYPE
612#define DEBUG_TYPE "bar"
Daniel Dunbar04bbd9c2009-07-25 01:55:32 +0000[diff] [blame]613DEBUG(errs() &lt;&lt; "'bar' debug type\n"));
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]614#undef DEBUG_TYPE
615#define DEBUG_TYPE ""
Daniel Dunbar04bbd9c2009-07-25 01:55:32 +0000[diff] [blame]616DEBUG(errs() &lt;&lt; "No debug type (2)\n");
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]617</pre>
618</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]619
620<p>Then you can run your pass like this:</p>
621
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]622<div class="doc_code">
623<pre>
624$ opt &lt; a.bc &gt; /dev/null -mypass
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]625<i>&lt;no output&gt;</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]626$ opt &lt; a.bc &gt; /dev/null -mypass -debug
627No debug type
628'foo' debug type
629'bar' debug type
630No debug type (2)
631$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=foo
632'foo' debug type
633$ opt &lt; a.bc &gt; /dev/null -mypass -debug-only=bar
634'bar' debug type
635</pre>
636</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]637
638<p>Of course, in practice, you should only set <tt>DEBUG_TYPE</tt> at the top of
639a file, to specify the debug type for the entire module (if you do this before
Chris Lattner730a5da1a2005-04-26 22:56:16 +0000[diff] [blame]640you <tt>#include "llvm/Support/Debug.h"</tt>, you don't have to insert the ugly
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]641<tt>#undef</tt>'s). Also, you should use names more meaningful than "foo" and
642"bar", because there is no system in place to ensure that names do not
643conflict. If two different modules use the same string, they will all be turned
644on when the name is specified. This allows, for example, all debug information
645for instruction scheduling to be enabled with <tt>-debug-type=InstrSched</tt>,
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]646even if the source lives in multiple files.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]647
Daniel Dunbar11595b22009-08-07 23:48:59 +0000[diff] [blame]648<p>The <tt>DEBUG_WITH_TYPE</tt> macro is also available for situations where you
649would like to set <tt>DEBUG_TYPE</tt>, but only for one specific <tt>DEBUG</tt>
650statement. It takes an additional first parameter, which is the type to use. For
Benjamin Kramer0f420382009-10-12 14:46:08 +0000[diff] [blame]651example, the preceding example could be written as:</p>
Daniel Dunbar11595b22009-08-07 23:48:59 +0000[diff] [blame]652
653
654<div class="doc_code">
655<pre>
656DEBUG_WITH_TYPE("", errs() &lt;&lt; "No debug type\n");
657DEBUG_WITH_TYPE("foo", errs() &lt;&lt; "'foo' debug type\n");
658DEBUG_WITH_TYPE("bar", errs() &lt;&lt; "'bar' debug type\n"));
659DEBUG_WITH_TYPE("", errs() &lt;&lt; "No debug type (2)\n");
660</pre>
661</div>
662
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]663</div>
664
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]665</div>
666
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]667<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]668<h3>
Chris Lattner58e53452006-12-19 21:46:21 +0000[diff] [blame]669 <a name="Statistic">The <tt>Statistic</tt> class &amp; <tt>-stats</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]670 option</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]671</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]672
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]673<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]674
675<p>The "<tt><a
Chris Lattner730a5da1a2005-04-26 22:56:16 +0000[diff] [blame]676href="/doxygen/Statistic_8h-source.html">llvm/ADT/Statistic.h</a></tt>" file
Chris Lattner58e53452006-12-19 21:46:21 +0000[diff] [blame]677provides a class named <tt>Statistic</tt> that is used as a unified way to
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]678keep track of what the LLVM compiler is doing and how effective various
679optimizations are. It is useful to see what optimizations are contributing to
680making a particular program run faster.</p>
681
682<p>Often you may run your pass on some big program, and you're interested to see
683how many times it makes a certain transformation. Although you can do this with
684hand inspection, or some ad-hoc method, this is a real pain and not very useful
Chris Lattner58e53452006-12-19 21:46:21 +0000[diff] [blame]685for big programs. Using the <tt>Statistic</tt> class makes it very easy to
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]686keep track of this information, and the calculated information is presented in a
687uniform manner with the rest of the passes being executed.</p>
688
689<p>There are many examples of <tt>Statistic</tt> uses, but the basics of using
690it are as follows:</p>
691
692<ol>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]693 <li><p>Define your statistic like this:</p>
694
695<div class="doc_code">
696<pre>
Chris Lattner58e53452006-12-19 21:46:21 +0000[diff] [blame]697#define <a href="#DEBUG_TYPE">DEBUG_TYPE</a> "mypassname" <i>// This goes before any #includes.</i>
698STATISTIC(NumXForms, "The # of times I did stuff");
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]699</pre>
700</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]701
Chris Lattner58e53452006-12-19 21:46:21 +0000[diff] [blame]702 <p>The <tt>STATISTIC</tt> macro defines a static variable, whose name is
703 specified by the first argument. The pass name is taken from the DEBUG_TYPE
704 macro, and the description is taken from the second argument. The variable
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]705 defined ("NumXForms" in this case) acts like an unsigned integer.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]706
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]707 <li><p>Whenever you make a transformation, bump the counter:</p>
708
709<div class="doc_code">
710<pre>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]711++NumXForms; // <i>I did stuff!</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]712</pre>
713</div>
714
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]715 </li>
716 </ol>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]717
718 <p>That's all you have to do. To get '<tt>opt</tt>' to print out the
719 statistics gathered, use the '<tt>-stats</tt>' option:</p>
720
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]721<div class="doc_code">
722<pre>
723$ opt -stats -mypassname &lt; program.bc &gt; /dev/null
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]724<i>... statistics output ...</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]725</pre>
726</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]727
Reid Spencer2543a222007-02-09 16:00:28 +0000[diff] [blame]728 <p> When running <tt>opt</tt> on a C file from the SPEC benchmark
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]729suite, it gives a report that looks like this:</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]730
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]731<div class="doc_code">
732<pre>
Gabor Greifa54634a2007-07-06 22:07:22 +0000[diff] [blame]733 7646 bitcodewriter - Number of normal instructions
734 725 bitcodewriter - Number of oversized instructions
735 129996 bitcodewriter - Number of bitcode bytes written
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]736 2817 raise - Number of insts DCEd or constprop'd
737 3213 raise - Number of cast-of-self removed
738 5046 raise - Number of expression trees converted
739 75 raise - Number of other getelementptr's formed
740 138 raise - Number of load/store peepholes
741 42 deadtypeelim - Number of unused typenames removed from symtab
742 392 funcresolve - Number of varargs functions resolved
743 27 globaldce - Number of global variables removed
744 2 adce - Number of basic blocks removed
745 134 cee - Number of branches revectored
746 49 cee - Number of setcc instruction eliminated
747 532 gcse - Number of loads removed
748 2919 gcse - Number of instructions removed
749 86 indvars - Number of canonical indvars added
750 87 indvars - Number of aux indvars removed
751 25 instcombine - Number of dead inst eliminate
752 434 instcombine - Number of insts combined
753 248 licm - Number of load insts hoisted
754 1298 licm - Number of insts hoisted to a loop pre-header
755 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
756 75 mem2reg - Number of alloca's promoted
757 1444 cfgsimplify - Number of blocks simplified
758</pre>
759</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]760
761<p>Obviously, with so many optimizations, having a unified framework for this
762stuff is very nice. Making your pass fit well into the framework makes it more
763maintainable and useful.</p>
764
765</div>
766
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]767<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]768<h3>
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]769 <a name="ViewGraph">Viewing graphs while debugging code</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]770</h3>
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]771
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]772<div>
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]773
774<p>Several of the important data structures in LLVM are graphs: for example
775CFGs made out of LLVM <a href="#BasicBlock">BasicBlock</a>s, CFGs made out of
776LLVM <a href="CodeGenerator.html#machinebasicblock">MachineBasicBlock</a>s, and
777<a href="CodeGenerator.html#selectiondag_intro">Instruction Selection
778DAGs</a>. In many cases, while debugging various parts of the compiler, it is
779nice to instantly visualize these graphs.</p>
780
781<p>LLVM provides several callbacks that are available in a debug build to do
782exactly that. If you call the <tt>Function::viewCFG()</tt> method, for example,
783the current LLVM tool will pop up a window containing the CFG for the function
784where each basic block is a node in the graph, and each node contains the
785instructions in the block. Similarly, there also exists
786<tt>Function::viewCFGOnly()</tt> (does not include the instructions), the
787<tt>MachineFunction::viewCFG()</tt> and <tt>MachineFunction::viewCFGOnly()</tt>,
788and the <tt>SelectionDAG::viewGraph()</tt> methods. Within GDB, for example,
Jim Laskey37c2c6c2006-10-02 12:28:07 +0000[diff] [blame]789you can usually use something like <tt>call DAG.viewGraph()</tt> to pop
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]790up a window. Alternatively, you can sprinkle calls to these functions in your
791code in places you want to debug.</p>
792
793<p>Getting this to work requires a small amount of configuration. On Unix
794systems with X11, install the <a href="http://www.graphviz.org">graphviz</a>
795toolkit, and make sure 'dot' and 'gv' are in your path. If you are running on
796Mac OS/X, download and install the Mac OS/X <a
797href="http://www.pixelglow.com/graphviz/">Graphviz program</a>, and add
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]798<tt>/Applications/Graphviz.app/Contents/MacOS/</tt> (or wherever you install
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]799it) to your path. Once in your system and path are set up, rerun the LLVM
800configure script and rebuild LLVM to enable this functionality.</p>
801
Jim Laskey37c2c6c2006-10-02 12:28:07 +0000[diff] [blame]802<p><tt>SelectionDAG</tt> has been extended to make it easier to locate
803<i>interesting</i> nodes in large complex graphs. From gdb, if you
804<tt>call DAG.setGraphColor(<i>node</i>, "<i>color</i>")</tt>, then the
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]805next <tt>call DAG.viewGraph()</tt> would highlight the node in the
Jim Laskey37c2c6c2006-10-02 12:28:07 +0000[diff] [blame]806specified color (choices of colors can be found at <a
Chris Lattner099213b2007-02-03 03:05:57 +0000[diff] [blame]807href="http://www.graphviz.org/doc/info/colors.html">colors</a>.) More
Jim Laskey37c2c6c2006-10-02 12:28:07 +0000[diff] [blame]808complex node attributes can be provided with <tt>call
809DAG.setGraphAttrs(<i>node</i>, "<i>attributes</i>")</tt> (choices can be
810found at <a href="http://www.graphviz.org/doc/info/attrs.html">Graph
811Attributes</a>.) If you want to restart and clear all the current graph
812attributes, then you can <tt>call DAG.clearGraphAttrs()</tt>. </p>
813
Chris Lattnered58a952011-06-13 15:59:35 +0000[diff] [blame^]814<p>Note that graph visualization features are compiled out of Release builds
815to reduce file size. This means that you need a Debug+Asserts or
816Release+Asserts build to use these features.</p>
817
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]818</div>
819
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]820</div>
821
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]822<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]823<h2>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]824 <a name="datastructure">Picking the Right Data Structure for a Task</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]825</h2>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]826<!-- *********************************************************************** -->
827
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]828<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]829
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]830<p>LLVM has a plethora of data structures in the <tt>llvm/ADT/</tt> directory,
831 and we commonly use STL data structures. This section describes the trade-offs
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]832 you should consider when you pick one.</p>
833
834<p>
835The first step is a choose your own adventure: do you want a sequential
836container, a set-like container, or a map-like container? The most important
837thing when choosing a container is the algorithmic properties of how you plan to
838access the container. Based on that, you should use:</p>
839
840<ul>
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]841<li>a <a href="#ds_map">map-like</a> container if you need efficient look-up
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]842 of an value based on another value. Map-like containers also support
843 efficient queries for containment (whether a key is in the map). Map-like
844 containers generally do not support efficient reverse mapping (values to
845 keys). If you need that, use two maps. Some map-like containers also
846 support efficient iteration through the keys in sorted order. Map-like
847 containers are the most expensive sort, only use them if you need one of
848 these capabilities.</li>
849
850<li>a <a href="#ds_set">set-like</a> container if you need to put a bunch of
851 stuff into a container that automatically eliminates duplicates. Some
852 set-like containers support efficient iteration through the elements in
853 sorted order. Set-like containers are more expensive than sequential
854 containers.
855</li>
856
857<li>a <a href="#ds_sequential">sequential</a> container provides
858 the most efficient way to add elements and keeps track of the order they are
859 added to the collection. They permit duplicates and support efficient
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]860 iteration, but do not support efficient look-up based on a key.
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]861</li>
862
Chris Lattner47bb37c2009-07-25 07:22:20 +0000[diff] [blame]863<li>a <a href="#ds_string">string</a> container is a specialized sequential
864 container or reference structure that is used for character or byte
865 arrays.</li>
866
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]867<li>a <a href="#ds_bit">bit</a> container provides an efficient way to store and
868 perform set operations on sets of numeric id's, while automatically
869 eliminating duplicates. Bit containers require a maximum of 1 bit for each
870 identifier you want to store.
871</li>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]872</ul>
873
874<p>
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]875Once the proper category of container is determined, you can fine tune the
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]876memory use, constant factors, and cache behaviors of access by intelligently
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]877picking a member of the category. Note that constant factors and cache behavior
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]878can be a big deal. If you have a vector that usually only contains a few
879elements (but could contain many), for example, it's much better to use
880<a href="#dss_smallvector">SmallVector</a> than <a href="#dss_vector">vector</a>
881. Doing so avoids (relatively) expensive malloc/free calls, which dwarf the
882cost of adding the elements to the container. </p>
883
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]884<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]885<h3>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]886 <a name="ds_sequential">Sequential Containers (std::vector, std::list, etc)</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]887</h3>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]888
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]889<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]890There are a variety of sequential containers available for you, based on your
891needs. Pick the first in this section that will do what you want.
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]892
893<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]894<h4>
Chris Lattnera76507f2011-04-05 23:18:20 +0000[diff] [blame]895 <a name="dss_arrayref">llvm/ADT/ArrayRef.h</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]896</h4>
Chris Lattnera76507f2011-04-05 23:18:20 +0000[diff] [blame]897
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]898<div>
Chris Lattnera76507f2011-04-05 23:18:20 +0000[diff] [blame]899<p>The llvm::ArrayRef class is the preferred class to use in an interface that
900 accepts a sequential list of elements in memory and just reads from them. By
901 taking an ArrayRef, the API can be passed a fixed size array, an std::vector,
902 an llvm::SmallVector and anything else that is contiguous in memory.
903</p>
904</div>
905
906
907
908<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]909<h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]910 <a name="dss_fixedarrays">Fixed Size Arrays</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]911</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]912
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]913<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]914<p>Fixed size arrays are very simple and very fast. They are good if you know
915exactly how many elements you have, or you have a (low) upper bound on how many
916you have.</p>
917</div>
918
919<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]920<h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]921 <a name="dss_heaparrays">Heap Allocated Arrays</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]922</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]923
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]924<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]925<p>Heap allocated arrays (new[] + delete[]) are also simple. They are good if
926the number of elements is variable, if you know how many elements you will need
927before the array is allocated, and if the array is usually large (if not,
928consider a <a href="#dss_smallvector">SmallVector</a>). The cost of a heap
929allocated array is the cost of the new/delete (aka malloc/free). Also note that
930if you are allocating an array of a type with a constructor, the constructor and
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]931destructors will be run for every element in the array (re-sizable vectors only
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]932construct those elements actually used).</p>
933</div>
934
935<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]936<h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]937 <a name="dss_smallvector">"llvm/ADT/SmallVector.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]938</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]939
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]940<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]941<p><tt>SmallVector&lt;Type, N&gt;</tt> is a simple class that looks and smells
942just like <tt>vector&lt;Type&gt;</tt>:
943it supports efficient iteration, lays out elements in memory order (so you can
944do pointer arithmetic between elements), supports efficient push_back/pop_back
945operations, supports efficient random access to its elements, etc.</p>
946
947<p>The advantage of SmallVector is that it allocates space for
948some number of elements (N) <b>in the object itself</b>. Because of this, if
949the SmallVector is dynamically smaller than N, no malloc is performed. This can
950be a big win in cases where the malloc/free call is far more expensive than the
951code that fiddles around with the elements.</p>
952
953<p>This is good for vectors that are "usually small" (e.g. the number of
954predecessors/successors of a block is usually less than 8). On the other hand,
955this makes the size of the SmallVector itself large, so you don't want to
956allocate lots of them (doing so will waste a lot of space). As such,
957SmallVectors are most useful when on the stack.</p>
958
959<p>SmallVector also provides a nice portable and efficient replacement for
960<tt>alloca</tt>.</p>
961
962</div>
963
964<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]965<h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]966 <a name="dss_vector">&lt;vector&gt;</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]967</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]968
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]969<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]970<p>
971std::vector is well loved and respected. It is useful when SmallVector isn't:
972when the size of the vector is often large (thus the small optimization will
973rarely be a benefit) or if you will be allocating many instances of the vector
974itself (which would waste space for elements that aren't in the container).
975vector is also useful when interfacing with code that expects vectors :).
976</p>
Chris Lattner9132f162007-02-05 06:30:51 +0000[diff] [blame]977
978<p>One worthwhile note about std::vector: avoid code like this:</p>
979
980<div class="doc_code">
981<pre>
982for ( ... ) {
Chris Lattner3d9c1bd2007-03-28 18:27:57 +0000[diff] [blame]983 std::vector&lt;foo&gt; V;
Chris Lattner9132f162007-02-05 06:30:51 +0000[diff] [blame]984 use V;
985}
986</pre>
987</div>
988
989<p>Instead, write this as:</p>
990
991<div class="doc_code">
992<pre>
Chris Lattner3d9c1bd2007-03-28 18:27:57 +0000[diff] [blame]993std::vector&lt;foo&gt; V;
Chris Lattner9132f162007-02-05 06:30:51 +0000[diff] [blame]994for ( ... ) {
995 use V;
996 V.clear();
997}
998</pre>
999</div>
1000
1001<p>Doing so will save (at least) one heap allocation and free per iteration of
1002the loop.</p>
1003
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1004</div>
1005
1006<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1007<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1008 <a name="dss_deque">&lt;deque&gt;</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1009</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1010
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1011<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1012<p>std::deque is, in some senses, a generalized version of std::vector. Like
1013std::vector, it provides constant time random access and other similar
1014properties, but it also provides efficient access to the front of the list. It
1015does not guarantee continuity of elements within memory.</p>
1016
1017<p>In exchange for this extra flexibility, std::deque has significantly higher
1018constant factor costs than std::vector. If possible, use std::vector or
1019something cheaper.</p>
1020</div>
1021
1022<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1023<h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1024 <a name="dss_list">&lt;list&gt;</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1025</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1026
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1027<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1028<p>std::list is an extremely inefficient class that is rarely useful.
1029It performs a heap allocation for every element inserted into it, thus having an
1030extremely high constant factor, particularly for small data types. std::list
1031also only supports bidirectional iteration, not random access iteration.</p>
1032
1033<p>In exchange for this high cost, std::list supports efficient access to both
1034ends of the list (like std::deque, but unlike std::vector or SmallVector). In
1035addition, the iterator invalidation characteristics of std::list are stronger
1036than that of a vector class: inserting or removing an element into the list does
1037not invalidate iterator or pointers to other elements in the list.</p>
1038</div>
1039
1040<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1041<h4>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1042 <a name="dss_ilist">llvm/ADT/ilist.h</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1043</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1044
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1045<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1046<p><tt>ilist&lt;T&gt;</tt> implements an 'intrusive' doubly-linked list. It is
1047intrusive, because it requires the element to store and provide access to the
1048prev/next pointers for the list.</p>
1049
Gabor Greif8243b8c2009-02-27 12:02:19 +0000[diff] [blame]1050<p><tt>ilist</tt> has the same drawbacks as <tt>std::list</tt>, and additionally
1051requires an <tt>ilist_traits</tt> implementation for the element type, but it
1052provides some novel characteristics. In particular, it can efficiently store
1053polymorphic objects, the traits class is informed when an element is inserted or
Gabor Greif4ef9bc02009-03-12 09:47:03 +0000[diff] [blame]1054removed from the list, and <tt>ilist</tt>s are guaranteed to support a
1055constant-time splice operation.</p>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1056
Gabor Greif4ef9bc02009-03-12 09:47:03 +0000[diff] [blame]1057<p>These properties are exactly what we want for things like
1058<tt>Instruction</tt>s and basic blocks, which is why these are implemented with
1059<tt>ilist</tt>s.</p>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1060
1061Related classes of interest are explained in the following subsections:
1062 <ul>
Gabor Greifb4b608c2009-02-27 13:28:07 +0000[diff] [blame]1063 <li><a href="#dss_ilist_traits">ilist_traits</a></li>
Gabor Greif8243b8c2009-02-27 12:02:19 +0000[diff] [blame]1064 <li><a href="#dss_iplist">iplist</a></li>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1065 <li><a href="#dss_ilist_node">llvm/ADT/ilist_node.h</a></li>
Gabor Greif02ab9ab2009-03-12 10:30:31 +0000[diff] [blame]1066 <li><a href="#dss_ilist_sentinel">Sentinels</a></li>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1067 </ul>
1068</div>
1069
1070<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1071<h4>
Gabor Greifb4b608c2009-02-27 13:28:07 +0000[diff] [blame]1072 <a name="dss_ilist_traits">ilist_traits</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1073</h4>
Gabor Greifb4b608c2009-02-27 13:28:07 +0000[diff] [blame]1074
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1075<div>
Gabor Greifb4b608c2009-02-27 13:28:07 +0000[diff] [blame]1076<p><tt>ilist_traits&lt;T&gt;</tt> is <tt>ilist&lt;T&gt;</tt>'s customization
1077mechanism. <tt>iplist&lt;T&gt;</tt> (and consequently <tt>ilist&lt;T&gt;</tt>)
1078publicly derive from this traits class.</p>
1079</div>
1080
1081<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1082<h4>
Gabor Greif8243b8c2009-02-27 12:02:19 +0000[diff] [blame]1083 <a name="dss_iplist">iplist</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1084</h4>
Gabor Greif8243b8c2009-02-27 12:02:19 +0000[diff] [blame]1085
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1086<div>
Gabor Greif8243b8c2009-02-27 12:02:19 +0000[diff] [blame]1087<p><tt>iplist&lt;T&gt;</tt> is <tt>ilist&lt;T&gt;</tt>'s base and as such
Gabor Greif4ef9bc02009-03-12 09:47:03 +0000[diff] [blame]1088supports a slightly narrower interface. Notably, inserters from
1089<tt>T&amp;</tt> are absent.</p>
Gabor Greifb4b608c2009-02-27 13:28:07 +0000[diff] [blame]1090
1091<p><tt>ilist_traits&lt;T&gt;</tt> is a public base of this class and can be
1092used for a wide variety of customizations.</p>
Gabor Greif8243b8c2009-02-27 12:02:19 +0000[diff] [blame]1093</div>
1094
1095<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1096<h4>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1097 <a name="dss_ilist_node">llvm/ADT/ilist_node.h</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1098</h4>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1099
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1100<div>
Gabor Greifd5575ad2009-02-27 11:37:41 +0000[diff] [blame]1101<p><tt>ilist_node&lt;T&gt;</tt> implements a the forward and backward links
1102that are expected by the <tt>ilist&lt;T&gt;</tt> (and analogous containers)
1103in the default manner.</p>
1104
1105<p><tt>ilist_node&lt;T&gt;</tt>s are meant to be embedded in the node type
Gabor Greif4ef9bc02009-03-12 09:47:03 +0000[diff] [blame]1106<tt>T</tt>, usually <tt>T</tt> publicly derives from
1107<tt>ilist_node&lt;T&gt;</tt>.</p>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1108</div>
1109
1110<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1111<h4>
Gabor Greif02ab9ab2009-03-12 10:30:31 +0000[diff] [blame]1112 <a name="dss_ilist_sentinel">Sentinels</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1113</h4>
Gabor Greif02ab9ab2009-03-12 10:30:31 +0000[diff] [blame]1114
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1115<div>
Dan Gohmand0c7fd82010-02-25 23:51:27 +0000[diff] [blame]1116<p><tt>ilist</tt>s have another specialty that must be considered. To be a good
Gabor Greif02ab9ab2009-03-12 10:30:31 +0000[diff] [blame]1117citizen in the C++ ecosystem, it needs to support the standard container
1118operations, such as <tt>begin</tt> and <tt>end</tt> iterators, etc. Also, the
1119<tt>operator--</tt> must work correctly on the <tt>end</tt> iterator in the
1120case of non-empty <tt>ilist</tt>s.</p>
1121
1122<p>The only sensible solution to this problem is to allocate a so-called
1123<i>sentinel</i> along with the intrusive list, which serves as the <tt>end</tt>
1124iterator, providing the back-link to the last element. However conforming to the
1125C++ convention it is illegal to <tt>operator++</tt> beyond the sentinel and it
1126also must not be dereferenced.</p>
1127
1128<p>These constraints allow for some implementation freedom to the <tt>ilist</tt>
1129how to allocate and store the sentinel. The corresponding policy is dictated
1130by <tt>ilist_traits&lt;T&gt;</tt>. By default a <tt>T</tt> gets heap-allocated
1131whenever the need for a sentinel arises.</p>
1132
1133<p>While the default policy is sufficient in most cases, it may break down when
1134<tt>T</tt> does not provide a default constructor. Also, in the case of many
1135instances of <tt>ilist</tt>s, the memory overhead of the associated sentinels
1136is wasted. To alleviate the situation with numerous and voluminous
1137<tt>T</tt>-sentinels, sometimes a trick is employed, leading to <i>ghostly
1138sentinels</i>.</p>
1139
1140<p>Ghostly sentinels are obtained by specially-crafted <tt>ilist_traits&lt;T&gt;</tt>
1141which superpose the sentinel with the <tt>ilist</tt> instance in memory. Pointer
1142arithmetic is used to obtain the sentinel, which is relative to the
1143<tt>ilist</tt>'s <tt>this</tt> pointer. The <tt>ilist</tt> is augmented by an
1144extra pointer, which serves as the back-link of the sentinel. This is the only
1145field in the ghostly sentinel which can be legally accessed.</p>
1146</div>
1147
1148<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1149<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1150 <a name="dss_other">Other Sequential Container options</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1151</h4>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1152
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1153<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1154<p>Other STL containers are available, such as std::string.</p>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1155
1156<p>There are also various STL adapter classes such as std::queue,
1157std::priority_queue, std::stack, etc. These provide simplified access to an
1158underlying container but don't affect the cost of the container itself.</p>
1159
1160</div>
1161
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1162</div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1163
1164<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1165<h3>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1166 <a name="ds_set">Set-Like Containers (std::set, SmallSet, SetVector, etc)</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1167</h3>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1168
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1169<div>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1170
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1171<p>Set-like containers are useful when you need to canonicalize multiple values
1172into a single representation. There are several different choices for how to do
1173this, providing various trade-offs.</p>
1174
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1175<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1176<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1177 <a name="dss_sortedvectorset">A sorted 'vector'</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1178</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1179
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1180<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1181
Chris Lattnerf5ddd042007-02-03 08:10:45 +0000[diff] [blame]1182<p>If you intend to insert a lot of elements, then do a lot of queries, a
1183great approach is to use a vector (or other sequential container) with
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1184std::sort+std::unique to remove duplicates. This approach works really well if
Chris Lattnerf5ddd042007-02-03 08:10:45 +0000[diff] [blame]1185your usage pattern has these two distinct phases (insert then query), and can be
1186coupled with a good choice of <a href="#ds_sequential">sequential container</a>.
1187</p>
1188
1189<p>
1190This combination provides the several nice properties: the result data is
1191contiguous in memory (good for cache locality), has few allocations, is easy to
1192address (iterators in the final vector are just indices or pointers), and can be
1193efficiently queried with a standard binary or radix search.</p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1194
1195</div>
1196
1197<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1198<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1199 <a name="dss_smallset">"llvm/ADT/SmallSet.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1200</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1201
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1202<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1203
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]1204<p>If you have a set-like data structure that is usually small and whose elements
Chris Lattnercb292fc2007-02-03 07:59:51 +0000[diff] [blame]1205are reasonably small, a <tt>SmallSet&lt;Type, N&gt;</tt> is a good choice. This set
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1206has space for N elements in place (thus, if the set is dynamically smaller than
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1207N, no malloc traffic is required) and accesses them with a simple linear search.
1208When the set grows beyond 'N' elements, it allocates a more expensive representation that
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1209guarantees efficient access (for most types, it falls back to std::set, but for
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1210pointers it uses something far better, <a
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1211href="#dss_smallptrset">SmallPtrSet</a>).</p>
1212
1213<p>The magic of this class is that it handles small sets extremely efficiently,
1214but gracefully handles extremely large sets without loss of efficiency. The
1215drawback is that the interface is quite small: it supports insertion, queries
1216and erasing, but does not support iteration.</p>
1217
1218</div>
1219
1220<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1221<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1222 <a name="dss_smallptrset">"llvm/ADT/SmallPtrSet.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1223</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1224
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1225<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1226
Gabor Greif6d6cf8f2010-03-26 19:30:47 +0000[diff] [blame]1227<p>SmallPtrSet has all the advantages of <tt>SmallSet</tt> (and a <tt>SmallSet</tt> of pointers is
1228transparently implemented with a <tt>SmallPtrSet</tt>), but also supports iterators. If
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1229more than 'N' insertions are performed, a single quadratically
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1230probed hash table is allocated and grows as needed, providing extremely
1231efficient access (constant time insertion/deleting/queries with low constant
1232factors) and is very stingy with malloc traffic.</p>
1233
Gabor Greif6d6cf8f2010-03-26 19:30:47 +0000[diff] [blame]1234<p>Note that, unlike <tt>std::set</tt>, the iterators of <tt>SmallPtrSet</tt> are invalidated
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1235whenever an insertion occurs. Also, the values visited by the iterators are not
1236visited in sorted order.</p>
1237
1238</div>
1239
1240<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1241<h4>
Chris Lattnerbd0079c2007-09-30 00:58:59 +0000[diff] [blame]1242 <a name="dss_denseset">"llvm/ADT/DenseSet.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1243</h4>
Chris Lattnerbd0079c2007-09-30 00:58:59 +0000[diff] [blame]1244
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1245<div>
Chris Lattnerbd0079c2007-09-30 00:58:59 +0000[diff] [blame]1246
1247<p>
1248DenseSet is a simple quadratically probed hash table. It excels at supporting
1249small values: it uses a single allocation to hold all of the pairs that
1250are currently inserted in the set. DenseSet is a great way to unique small
1251values that are not simple pointers (use <a
1252href="#dss_smallptrset">SmallPtrSet</a> for pointers). Note that DenseSet has
1253the same requirements for the value type that <a
1254href="#dss_densemap">DenseMap</a> has.
1255</p>
1256
1257</div>
1258
1259<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1260<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1261 <a name="dss_FoldingSet">"llvm/ADT/FoldingSet.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1262</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1263
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1264<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1265
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1266<p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1267FoldingSet is an aggregate class that is really good at uniquing
1268expensive-to-create or polymorphic objects. It is a combination of a chained
1269hash table with intrusive links (uniqued objects are required to inherit from
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1270FoldingSetNode) that uses <a href="#dss_smallvector">SmallVector</a> as part of
1271its ID process.</p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1272
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1273<p>Consider a case where you want to implement a "getOrCreateFoo" method for
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1274a complex object (for example, a node in the code generator). The client has a
1275description of *what* it wants to generate (it knows the opcode and all the
1276operands), but we don't want to 'new' a node, then try inserting it into a set
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1277only to find out it already exists, at which point we would have to delete it
1278and return the node that already exists.
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1279</p>
1280
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1281<p>To support this style of client, FoldingSet perform a query with a
1282FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1283element that we want to query for. The query either returns the element
1284matching the ID or it returns an opaque ID that indicates where insertion should
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1285take place. Construction of the ID usually does not require heap traffic.</p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1286
1287<p>Because FoldingSet uses intrusive links, it can support polymorphic objects
1288in the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1289Because the elements are individually allocated, pointers to the elements are
1290stable: inserting or removing elements does not invalidate any pointers to other
1291elements.
1292</p>
1293
1294</div>
1295
1296<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1297<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1298 <a name="dss_set">&lt;set&gt;</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1299</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1300
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1301<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1302
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1303<p><tt>std::set</tt> is a reasonable all-around set class, which is decent at
1304many things but great at nothing. std::set allocates memory for each element
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1305inserted (thus it is very malloc intensive) and typically stores three pointers
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1306per element in the set (thus adding a large amount of per-element space
1307overhead). It offers guaranteed log(n) performance, which is not particularly
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1308fast from a complexity standpoint (particularly if the elements of the set are
1309expensive to compare, like strings), and has extremely high constant factors for
1310lookup, insertion and removal.</p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1311
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1312<p>The advantages of std::set are that its iterators are stable (deleting or
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1313inserting an element from the set does not affect iterators or pointers to other
1314elements) and that iteration over the set is guaranteed to be in sorted order.
1315If the elements in the set are large, then the relative overhead of the pointers
1316and malloc traffic is not a big deal, but if the elements of the set are small,
1317std::set is almost never a good choice.</p>
1318
1319</div>
1320
1321<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1322<h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1323 <a name="dss_setvector">"llvm/ADT/SetVector.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1324</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1325
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1326<div>
Chris Lattner0a165ee2007-02-04 00:00:26 +0000[diff] [blame]1327<p>LLVM's SetVector&lt;Type&gt; is an adapter class that combines your choice of
1328a set-like container along with a <a href="#ds_sequential">Sequential
1329Container</a>. The important property
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1330that this provides is efficient insertion with uniquing (duplicate elements are
1331ignored) with iteration support. It implements this by inserting elements into
1332both a set-like container and the sequential container, using the set-like
1333container for uniquing and the sequential container for iteration.
1334</p>
1335
1336<p>The difference between SetVector and other sets is that the order of
1337iteration is guaranteed to match the order of insertion into the SetVector.
1338This property is really important for things like sets of pointers. Because
1339pointer values are non-deterministic (e.g. vary across runs of the program on
Chris Lattner0a165ee2007-02-04 00:00:26 +0000[diff] [blame]1340different machines), iterating over the pointers in the set will
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1341not be in a well-defined order.</p>
1342
1343<p>
1344The drawback of SetVector is that it requires twice as much space as a normal
1345set and has the sum of constant factors from the set-like container and the
1346sequential container that it uses. Use it *only* if you need to iterate over
1347the elements in a deterministic order. SetVector is also expensive to delete
Chris Lattner0a165ee2007-02-04 00:00:26 +0000[diff] [blame]1348elements out of (linear time), unless you use it's "pop_back" method, which is
1349faster.
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1350</p>
1351
Chris Lattner0a165ee2007-02-04 00:00:26 +0000[diff] [blame]1352<p>SetVector is an adapter class that defaults to using std::vector and std::set
1353for the underlying containers, so it is quite expensive. However,
1354<tt>"llvm/ADT/SetVector.h"</tt> also provides a SmallSetVector class, which
1355defaults to using a SmallVector and SmallSet of a specified size. If you use
1356this, and if your sets are dynamically smaller than N, you will save a lot of
1357heap traffic.</p>
1358
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1359</div>
1360
1361<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1362<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1363 <a name="dss_uniquevector">"llvm/ADT/UniqueVector.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1364</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1365
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1366<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1367
1368<p>
1369UniqueVector is similar to <a href="#dss_setvector">SetVector</a>, but it
1370retains a unique ID for each element inserted into the set. It internally
1371contains a map and a vector, and it assigns a unique ID for each value inserted
1372into the set.</p>
1373
1374<p>UniqueVector is very expensive: its cost is the sum of the cost of
1375maintaining both the map and vector, it has high complexity, high constant
1376factors, and produces a lot of malloc traffic. It should be avoided.</p>
1377
1378</div>
1379
1380
1381<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1382<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1383 <a name="dss_otherset">Other Set-Like Container Options</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1384</h4>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1385
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1386<div>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1387
1388<p>
1389The STL provides several other options, such as std::multiset and the various
Chris Lattner4ce03e62009-03-09 05:20:45 +0000[diff] [blame]1390"hash_set" like containers (whether from C++ TR1 or from the SGI library). We
1391never use hash_set and unordered_set because they are generally very expensive
1392(each insertion requires a malloc) and very non-portable.
1393</p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1394
1395<p>std::multiset is useful if you're not interested in elimination of
Chris Lattner68b0ec52007-02-03 08:20:15 +0000[diff] [blame]1396duplicates, but has all the drawbacks of std::set. A sorted vector (where you
1397don't delete duplicate entries) or some other approach is almost always
1398better.</p>
Chris Lattnerac760292007-02-03 07:59:07 +0000[diff] [blame]1399
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1400</div>
1401
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1402</div>
1403
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1404<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1405<h3>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1406 <a name="ds_map">Map-Like Containers (std::map, DenseMap, etc)</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1407</h3>
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1408
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1409<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1410Map-like containers are useful when you want to associate data to a key. As
1411usual, there are a lot of different ways to do this. :)
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1412
1413<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1414<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1415 <a name="dss_sortedvectormap">A sorted 'vector'</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1416</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1417
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1418<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1419
1420<p>
1421If your usage pattern follows a strict insert-then-query approach, you can
1422trivially use the same approach as <a href="#dss_sortedvectorset">sorted vectors
1423for set-like containers</a>. The only difference is that your query function
1424(which uses std::lower_bound to get efficient log(n) lookup) should only compare
1425the key, not both the key and value. This yields the same advantages as sorted
1426vectors for sets.
1427</p>
1428</div>
1429
1430<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1431<h4>
Chris Lattnerd81d7be2007-02-08 19:14:21 +0000[diff] [blame]1432 <a name="dss_stringmap">"llvm/ADT/StringMap.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1433</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1434
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1435<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1436
1437<p>
1438Strings are commonly used as keys in maps, and they are difficult to support
1439efficiently: they are variable length, inefficient to hash and compare when
Chris Lattnerd81d7be2007-02-08 19:14:21 +0000[diff] [blame]1440long, expensive to copy, etc. StringMap is a specialized container designed to
1441cope with these issues. It supports mapping an arbitrary range of bytes to an
1442arbitrary other object.</p>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1443
Chris Lattnerd81d7be2007-02-08 19:14:21 +0000[diff] [blame]1444<p>The StringMap implementation uses a quadratically-probed hash table, where
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1445the buckets store a pointer to the heap allocated entries (and some other
1446stuff). The entries in the map must be heap allocated because the strings are
1447variable length. The string data (key) and the element object (value) are
1448stored in the same allocation with the string data immediately after the element
1449object. This container guarantees the "<tt>(char*)(&amp;Value+1)</tt>" points
1450to the key string for a value.</p>
1451
Chris Lattnerd81d7be2007-02-08 19:14:21 +0000[diff] [blame]1452<p>The StringMap is very fast for several reasons: quadratic probing is very
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1453cache efficient for lookups, the hash value of strings in buckets is not
Nick Lewyckydaf67722010-08-01 23:18:45 +0000[diff] [blame]1454recomputed when looking up an element, StringMap rarely has to touch the
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1455memory for unrelated objects when looking up a value (even when hash collisions
1456happen), hash table growth does not recompute the hash values for strings
1457already in the table, and each pair in the map is store in a single allocation
1458(the string data is stored in the same allocation as the Value of a pair).</p>
1459
Chris Lattnerd81d7be2007-02-08 19:14:21 +0000[diff] [blame]1460<p>StringMap also provides query methods that take byte ranges, so it only ever
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1461copies a string if a value is inserted into the table.</p>
1462</div>
1463
1464<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1465<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1466 <a name="dss_indexedmap">"llvm/ADT/IndexedMap.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1467</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1468
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1469<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1470<p>
1471IndexedMap is a specialized container for mapping small dense integers (or
1472values that can be mapped to small dense integers) to some other type. It is
1473internally implemented as a vector with a mapping function that maps the keys to
1474the dense integer range.
1475</p>
1476
1477<p>
1478This is useful for cases like virtual registers in the LLVM code generator: they
1479have a dense mapping that is offset by a compile-time constant (the first
1480virtual register ID).</p>
1481
1482</div>
1483
1484<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1485<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1486 <a name="dss_densemap">"llvm/ADT/DenseMap.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1487</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1488
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1489<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1490
1491<p>
1492DenseMap is a simple quadratically probed hash table. It excels at supporting
1493small keys and values: it uses a single allocation to hold all of the pairs that
1494are currently inserted in the map. DenseMap is a great way to map pointers to
1495pointers, or map other small types to each other.
1496</p>
1497
1498<p>
1499There are several aspects of DenseMap that you should be aware of, however. The
1500iterators in a densemap are invalidated whenever an insertion occurs, unlike
1501map. Also, because DenseMap allocates space for a large number of key/value
Chris Lattnerf043f762007-02-03 20:17:53 +0000[diff] [blame]1502pairs (it starts with 64 by default), it will waste a lot of space if your keys
1503or values are large. Finally, you must implement a partial specialization of
Chris Lattner0625bd62007-09-17 18:34:04 +0000[diff] [blame]1504DenseMapInfo for the key that you want, if it isn't already supported. This
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1505is required to tell DenseMap about two special marker values (which can never be
Chris Lattnerf043f762007-02-03 20:17:53 +0000[diff] [blame]1506inserted into the map) that it needs internally.</p>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1507
1508</div>
1509
1510<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1511<h4>
Jeffrey Yasskin4546d312009-10-22 22:11:22 +0000[diff] [blame]1512 <a name="dss_valuemap">"llvm/ADT/ValueMap.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1513</h4>
Jeffrey Yasskin4546d312009-10-22 22:11:22 +0000[diff] [blame]1514
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1515<div>
Jeffrey Yasskin4546d312009-10-22 22:11:22 +0000[diff] [blame]1516
1517<p>
1518ValueMap is a wrapper around a <a href="#dss_densemap">DenseMap</a> mapping
1519Value*s (or subclasses) to another type. When a Value is deleted or RAUW'ed,
1520ValueMap will update itself so the new version of the key is mapped to the same
1521value, just as if the key were a WeakVH. You can configure exactly how this
1522happens, and what else happens on these two events, by passing
1523a <code>Config</code> parameter to the ValueMap template.</p>
1524
1525</div>
1526
1527<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1528<h4>
Jakob Stoklund Olesen36eab1c2010-12-14 00:55:51 +0000[diff] [blame]1529 <a name="dss_intervalmap">"llvm/ADT/IntervalMap.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1530</h4>
Jakob Stoklund Olesen36eab1c2010-12-14 00:55:51 +0000[diff] [blame]1531
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1532<div>
Jakob Stoklund Olesen36eab1c2010-12-14 00:55:51 +0000[diff] [blame]1533
1534<p> IntervalMap is a compact map for small keys and values. It maps key
1535intervals instead of single keys, and it will automatically coalesce adjacent
1536intervals. When then map only contains a few intervals, they are stored in the
1537map object itself to avoid allocations.</p>
1538
1539<p> The IntervalMap iterators are quite big, so they should not be passed around
1540as STL iterators. The heavyweight iterators allow a smaller data structure.</p>
1541
1542</div>
1543
1544<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1545<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1546 <a name="dss_map">&lt;map&gt;</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1547</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1548
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1549<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1550
1551<p>
1552std::map has similar characteristics to <a href="#dss_set">std::set</a>: it uses
1553a single allocation per pair inserted into the map, it offers log(n) lookup with
1554an extremely large constant factor, imposes a space penalty of 3 pointers per
1555pair in the map, etc.</p>
1556
1557<p>std::map is most useful when your keys or values are very large, if you need
1558to iterate over the collection in sorted order, or if you need stable iterators
1559into the map (i.e. they don't get invalidated if an insertion or deletion of
1560another element takes place).</p>
1561
1562</div>
1563
1564<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1565<h4>
Jakob Stoklund Olesen04123d42011-04-05 20:56:08 +0000[diff] [blame]1566 <a name="dss_inteqclasses">"llvm/ADT/IntEqClasses.h"</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1567</h4>
Jakob Stoklund Olesen04123d42011-04-05 20:56:08 +0000[diff] [blame]1568
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1569<div>
Jakob Stoklund Olesen04123d42011-04-05 20:56:08 +0000[diff] [blame]1570
1571<p>IntEqClasses provides a compact representation of equivalence classes of
1572small integers. Initially, each integer in the range 0..n-1 has its own
1573equivalence class. Classes can be joined by passing two class representatives to
1574the join(a, b) method. Two integers are in the same class when findLeader()
1575returns the same representative.</p>
1576
1577<p>Once all equivalence classes are formed, the map can be compressed so each
1578integer 0..n-1 maps to an equivalence class number in the range 0..m-1, where m
1579is the total number of equivalence classes. The map must be uncompressed before
1580it can be edited again.</p>
1581
1582</div>
1583
1584<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1585<h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1586 <a name="dss_othermap">Other Map-Like Container Options</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1587</h4>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1588
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1589<div>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1590
1591<p>
1592The STL provides several other options, such as std::multimap and the various
Chris Lattner4ce03e62009-03-09 05:20:45 +0000[diff] [blame]1593"hash_map" like containers (whether from C++ TR1 or from the SGI library). We
1594never use hash_set and unordered_set because they are generally very expensive
1595(each insertion requires a malloc) and very non-portable.</p>
Chris Lattnerf233c422007-02-03 19:49:31 +0000[diff] [blame]1596
1597<p>std::multimap is useful if you want to map a key to multiple values, but has
1598all the drawbacks of std::map. A sorted vector or some other approach is almost
1599always better.</p>
1600
Chris Lattner9d9985c2007-02-03 03:04:03 +0000[diff] [blame]1601</div>
1602
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1603</div>
1604
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1605<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1606<h3>
Chris Lattner47bb37c2009-07-25 07:22:20 +0000[diff] [blame]1607 <a name="ds_string">String-like containers</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1608</h3>
Chris Lattner47bb37c2009-07-25 07:22:20 +0000[diff] [blame]1609
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1610<div>
Chris Lattner47bb37c2009-07-25 07:22:20 +0000[diff] [blame]1611
1612<p>
1613TODO: const char* vs stringref vs smallstring vs std::string. Describe twine,
1614xref to #string_apis.
1615</p>
1616
1617</div>
1618
1619<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1620<h3>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1621 <a name="ds_bit">Bit storage containers (BitVector, SparseBitVector)</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1622</h3>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1623
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1624<div>
Chris Lattnerccc1b692007-09-25 22:37:50 +0000[diff] [blame]1625<p>Unlike the other containers, there are only two bit storage containers, and
1626choosing when to use each is relatively straightforward.</p>
1627
1628<p>One additional option is
1629<tt>std::vector&lt;bool&gt;</tt>: we discourage its use for two reasons 1) the
1630implementation in many common compilers (e.g. commonly available versions of
1631GCC) is extremely inefficient and 2) the C++ standards committee is likely to
1632deprecate this container and/or change it significantly somehow. In any case,
1633please don't use it.</p>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1634
1635<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1636<h4>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1637 <a name="dss_bitvector">BitVector</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1638</h4>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1639
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1640<div>
Dan Gohman60b0a862010-01-05 18:24:00 +0000[diff] [blame]1641<p> The BitVector container provides a dynamic size set of bits for manipulation.
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1642It supports individual bit setting/testing, as well as set operations. The set
1643operations take time O(size of bitvector), but operations are performed one word
1644at a time, instead of one bit at a time. This makes the BitVector very fast for
1645set operations compared to other containers. Use the BitVector when you expect
1646the number of set bits to be high (IE a dense set).
1647</p>
1648</div>
1649
1650<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1651<h4>
Dan Gohman60b0a862010-01-05 18:24:00 +0000[diff] [blame]1652 <a name="dss_smallbitvector">SmallBitVector</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1653</h4>
Dan Gohman60b0a862010-01-05 18:24:00 +0000[diff] [blame]1654
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1655<div>
Dan Gohman60b0a862010-01-05 18:24:00 +0000[diff] [blame]1656<p> The SmallBitVector container provides the same interface as BitVector, but
1657it is optimized for the case where only a small number of bits, less than
165825 or so, are needed. It also transparently supports larger bit counts, but
1659slightly less efficiently than a plain BitVector, so SmallBitVector should
1660only be used when larger counts are rare.
1661</p>
1662
1663<p>
1664At this time, SmallBitVector does not support set operations (and, or, xor),
1665and its operator[] does not provide an assignable lvalue.
1666</p>
1667</div>
1668
1669<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1670<h4>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1671 <a name="dss_sparsebitvector">SparseBitVector</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1672</h4>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1673
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1674<div>
Daniel Berlind746bbd2007-09-24 17:52:25 +0000[diff] [blame]1675<p> The SparseBitVector container is much like BitVector, with one major
1676difference: Only the bits that are set, are stored. This makes the
1677SparseBitVector much more space efficient than BitVector when the set is sparse,
1678as well as making set operations O(number of set bits) instead of O(size of
1679universe). 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
1680(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).
1681</p>
1682</div>
Chris Lattnered54f2f2005-10-17 01:36:23 +0000[diff] [blame]1683
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1684</div>
1685
1686</div>
1687
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1688<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1689<h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1690 <a name="common">Helpful Hints for Common Operations</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1691</h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1692<!-- *********************************************************************** -->
1693
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1694<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1695
1696<p>This section describes how to perform some very simple transformations of
1697LLVM code. This is meant to give examples of common idioms used, showing the
1698practical side of LLVM transformations. <p> Because this is a "how-to" section,
1699you should also read about the main classes that you will be working with. The
1700<a href="#coreclasses">Core LLVM Class Hierarchy Reference</a> contains details
1701and descriptions of the main classes that you should know about.</p>
1702
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1703<!-- NOTE: this section should be heavy on example code -->
1704<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1705<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1706 <a name="inspection">Basic Inspection and Traversal Routines</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1707</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1708
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1709<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1710
1711<p>The LLVM compiler infrastructure have many different data structures that may
1712be traversed. Following the example of the C++ standard template library, the
1713techniques used to traverse these various data structures are all basically the
1714same. For a enumerable sequence of values, the <tt>XXXbegin()</tt> function (or
1715method) returns an iterator to the start of the sequence, the <tt>XXXend()</tt>
1716function returns an iterator pointing to one past the last valid element of the
1717sequence, and there is some <tt>XXXiterator</tt> data type that is common
1718between the two operations.</p>
1719
1720<p>Because the pattern for iteration is common across many different aspects of
1721the program representation, the standard template library algorithms may be used
1722on them, and it is easier to remember how to iterate. First we show a few common
1723examples of the data structures that need to be traversed. Other data
1724structures are traversed in very similar ways.</p>
1725
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1726<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1727<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1728 <a name="iterate_function">Iterating over the </a><a
1729 href="#BasicBlock"><tt>BasicBlock</tt></a>s in a <a
1730 href="#Function"><tt>Function</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1731</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1732
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1733<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1734
1735<p>It's quite common to have a <tt>Function</tt> instance that you'd like to
1736transform in some way; in particular, you'd like to manipulate its
1737<tt>BasicBlock</tt>s. To facilitate this, you'll need to iterate over all of
1738the <tt>BasicBlock</tt>s that constitute the <tt>Function</tt>. The following is
1739an example that prints the name of a <tt>BasicBlock</tt> and the number of
1740<tt>Instruction</tt>s it contains:</p>
1741
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1742<div class="doc_code">
1743<pre>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1744// <i>func is a pointer to a Function instance</i>
1745for (Function::iterator i = func-&gt;begin(), e = func-&gt;end(); i != e; ++i)
1746 // <i>Print out the name of the basic block if it has one, and then the</i>
1747 // <i>number of instructions that it contains</i>
Chris Lattnerba7cdde2009-09-08 05:15:50 +0000[diff] [blame]1748 errs() &lt;&lt; "Basic block (name=" &lt;&lt; i-&gt;getName() &lt;&lt; ") has "
Bill Wendling22e978a2006-12-07 20:04:42 +0000[diff] [blame]1749 &lt;&lt; i-&gt;size() &lt;&lt; " instructions.\n";
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1750</pre>
1751</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1752
1753<p>Note that i can be used as if it were a pointer for the purposes of
Joel Stanley64cfdbb2002-09-06 21:55:13 +0000[diff] [blame]1754invoking member functions of the <tt>Instruction</tt> class. This is
1755because the indirection operator is overloaded for the iterator
Chris Lattner99948702003-08-05 22:54:23 +0000[diff] [blame]1756classes. In the above code, the expression <tt>i-&gt;size()</tt> is
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1757exactly equivalent to <tt>(*i).size()</tt> just like you'd expect.</p>
1758
1759</div>
1760
1761<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1762<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1763 <a name="iterate_basicblock">Iterating over the </a><a
1764 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1765 href="#BasicBlock"><tt>BasicBlock</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1766</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1767
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1768<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1769
1770<p>Just like when dealing with <tt>BasicBlock</tt>s in <tt>Function</tt>s, it's
1771easy to iterate over the individual instructions that make up
1772<tt>BasicBlock</tt>s. Here's a code snippet that prints out each instruction in
1773a <tt>BasicBlock</tt>:</p>
1774
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1775<div class="doc_code">
Chris Lattner897cc8e2005-03-06 06:00:13 +0000[diff] [blame]1776<pre>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1777// <i>blk is a pointer to a BasicBlock instance</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1778for (BasicBlock::iterator i = blk-&gt;begin(), e = blk-&gt;end(); i != e; ++i)
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1779 // <i>The next statement works since operator&lt;&lt;(ostream&amp;,...)</i>
1780 // <i>is overloaded for Instruction&amp;</i>
Chris Lattnerba7cdde2009-09-08 05:15:50 +0000[diff] [blame]1781 errs() &lt;&lt; *i &lt;&lt; "\n";
Chris Lattner897cc8e2005-03-06 06:00:13 +0000[diff] [blame]1782</pre>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1783</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1784
1785<p>However, this isn't really the best way to print out the contents of a
1786<tt>BasicBlock</tt>! Since the ostream operators are overloaded for virtually
1787anything you'll care about, you could have just invoked the print routine on the
Chris Lattnerba7cdde2009-09-08 05:15:50 +0000[diff] [blame]1788basic block itself: <tt>errs() &lt;&lt; *blk &lt;&lt; "\n";</tt>.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1789
1790</div>
1791
1792<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1793<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1794 <a name="iterate_institer">Iterating over the </a><a
1795 href="#Instruction"><tt>Instruction</tt></a>s in a <a
1796 href="#Function"><tt>Function</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1797</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1798
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1799<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1800
1801<p>If you're finding that you commonly iterate over a <tt>Function</tt>'s
1802<tt>BasicBlock</tt>s and then that <tt>BasicBlock</tt>'s <tt>Instruction</tt>s,
1803<tt>InstIterator</tt> should be used instead. You'll need to include <a
1804href="/doxygen/InstIterator_8h-source.html"><tt>llvm/Support/InstIterator.h</tt></a>,
1805and then instantiate <tt>InstIterator</tt>s explicitly in your code. Here's a
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]1806small example that shows how to dump all instructions in a function to the standard error stream:<p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1807
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1808<div class="doc_code">
1809<pre>
1810#include "<a href="/doxygen/InstIterator_8h-source.html">llvm/Support/InstIterator.h</a>"
1811
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]1812// <i>F is a pointer to a Function instance</i>
Chris Lattner09ebde22008-06-04 18:20:42 +0000[diff] [blame]1813for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
Chris Lattnerba7cdde2009-09-08 05:15:50 +0000[diff] [blame]1814 errs() &lt;&lt; *I &lt;&lt; "\n";
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1815</pre>
1816</div>
1817
1818<p>Easy, isn't it? You can also use <tt>InstIterator</tt>s to fill a
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]1819work list with its initial contents. For example, if you wanted to
1820initialize a work list to contain all instructions in a <tt>Function</tt>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1821F, all you would need to do is something like:</p>
1822
1823<div class="doc_code">
1824<pre>
1825std::set&lt;Instruction*&gt; worklist;
Chris Lattner09ebde22008-06-04 18:20:42 +0000[diff] [blame]1826// or better yet, SmallPtrSet&lt;Instruction*, 64&gt; worklist;
1827
1828for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1829 worklist.insert(&amp;*I);
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1830</pre>
1831</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1832
1833<p>The STL set <tt>worklist</tt> would now contain all instructions in the
1834<tt>Function</tt> pointed to by F.</p>
1835
1836</div>
1837
1838<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1839<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1840 <a name="iterate_convert">Turning an iterator into a class pointer (and
1841 vice-versa)</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1842</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1843
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1844<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1845
1846<p>Sometimes, it'll be useful to grab a reference (or pointer) to a class
Joel Stanley64cfdbb2002-09-06 21:55:13 +0000[diff] [blame]1847instance when all you've got at hand is an iterator. Well, extracting
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]1848a reference or a pointer from an iterator is very straight-forward.
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1849Assuming that <tt>i</tt> is a <tt>BasicBlock::iterator</tt> and <tt>j</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1850is a <tt>BasicBlock::const_iterator</tt>:</p>
1851
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1852<div class="doc_code">
1853<pre>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1854Instruction&amp; inst = *i; // <i>Grab reference to instruction reference</i>
1855Instruction* pinst = &amp;*i; // <i>Grab pointer to instruction reference</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1856const Instruction&amp; inst = *j;
1857</pre>
1858</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1859
1860<p>However, the iterators you'll be working with in the LLVM framework are
1861special: they will automatically convert to a ptr-to-instance type whenever they
1862need to. Instead of dereferencing the iterator and then taking the address of
1863the result, you can simply assign the iterator to the proper pointer type and
1864you get the dereference and address-of operation as a result of the assignment
1865(behind the scenes, this is a result of overloading casting mechanisms). Thus
1866the last line of the last example,</p>
1867
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1868<div class="doc_code">
1869<pre>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]1870Instruction *pinst = &amp;*i;
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1871</pre>
1872</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1873
1874<p>is semantically equivalent to</p>
1875
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1876<div class="doc_code">
1877<pre>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]1878Instruction *pinst = i;
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1879</pre>
1880</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1881
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]1882<p>It's also possible to turn a class pointer into the corresponding iterator,
1883and this is a constant time operation (very efficient). The following code
1884snippet illustrates use of the conversion constructors provided by LLVM
1885iterators. By using these, you can explicitly grab the iterator of something
1886without actually obtaining it via iteration over some structure:</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1887
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1888<div class="doc_code">
1889<pre>
1890void printNextInstruction(Instruction* inst) {
1891 BasicBlock::iterator it(inst);
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1892 ++it; // <i>After this line, it refers to the instruction after *inst</i>
Chris Lattnerba7cdde2009-09-08 05:15:50 +0000[diff] [blame]1893 if (it != inst-&gt;getParent()-&gt;end()) errs() &lt;&lt; *it &lt;&lt; "\n";
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1894}
1895</pre>
1896</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1897
Dan Gohman37027c32010-03-26 19:39:05 +0000[diff] [blame]1898<p>Unfortunately, these implicit conversions come at a cost; they prevent
1899these iterators from conforming to standard iterator conventions, and thus
Dan Gohmanaedde1e2010-03-26 19:51:14 +0000[diff] [blame]1900from being usable with standard algorithms and containers. For example, they
1901prevent the following code, where <tt>B</tt> is a <tt>BasicBlock</tt>,
Dan Gohman37027c32010-03-26 19:39:05 +0000[diff] [blame]1902from compiling:</p>
1903
1904<div class="doc_code">
1905<pre>
1906 llvm::SmallVector&lt;llvm::Instruction *, 16&gt;(B-&gt;begin(), B-&gt;end());
1907</pre>
1908</div>
1909
1910<p>Because of this, these implicit conversions may be removed some day,
Dan Gohmanaedde1e2010-03-26 19:51:14 +0000[diff] [blame]1911and <tt>operator*</tt> changed to return a pointer instead of a reference.</p>
Dan Gohman37027c32010-03-26 19:39:05 +0000[diff] [blame]1912
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1913</div>
1914
1915<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1916<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1917 <a name="iterate_complex">Finding call sites: a slightly more complex
1918 example</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1919</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1920
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1921<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1922
1923<p>Say that you're writing a FunctionPass and would like to count all the
1924locations in the entire module (that is, across every <tt>Function</tt>) where a
1925certain function (i.e., some <tt>Function</tt>*) is already in scope. As you'll
1926learn later, you may want to use an <tt>InstVisitor</tt> to accomplish this in a
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]1927much more straight-forward manner, but this example will allow us to explore how
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]1928you'd do it if you didn't have <tt>InstVisitor</tt> around. In pseudo-code, this
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1929is what we want to do:</p>
1930
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1931<div class="doc_code">
1932<pre>
1933initialize callCounter to zero
1934for each Function f in the Module
1935 for each BasicBlock b in f
1936 for each Instruction i in b
1937 if (i is a CallInst and calls the given function)
1938 increment callCounter
1939</pre>
1940</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1941
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1942<p>And the actual code is (remember, because we're writing a
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1943<tt>FunctionPass</tt>, our <tt>FunctionPass</tt>-derived class simply has to
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1944override the <tt>runOnFunction</tt> method):</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1945
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1946<div class="doc_code">
1947<pre>
1948Function* targetFunc = ...;
1949
1950class OurFunctionPass : public FunctionPass {
1951 public:
1952 OurFunctionPass(): callCounter(0) { }
1953
1954 virtual runOnFunction(Function&amp; F) {
1955 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
Eric Christopher6e670382008-11-08 08:20:49 +0000[diff] [blame]1956 for (BasicBlock::iterator i = b-&gt;begin(), ie = b-&gt;end(); i != ie; ++i) {
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1957 if (<a href="#CallInst">CallInst</a>* callInst = <a href="#isa">dyn_cast</a>&lt;<a
1958 href="#CallInst">CallInst</a>&gt;(&amp;*i)) {
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1959 // <i>We know we've encountered a call instruction, so we</i>
1960 // <i>need to determine if it's a call to the</i>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]1961 // <i>function pointed to by m_func or not.</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1962 if (callInst-&gt;getCalledFunction() == targetFunc)
1963 ++callCounter;
1964 }
1965 }
1966 }
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]1967 }
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1968
1969 private:
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]1970 unsigned callCounter;
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]1971};
1972</pre>
1973</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1974
1975</div>
1976
Brian Gaekef578c052003-11-07 19:25:45 +0000[diff] [blame]1977<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1978<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1979 <a name="calls_and_invokes">Treating calls and invokes the same way</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]1980</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1981
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]1982<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1983
1984<p>You may have noticed that the previous example was a bit oversimplified in
1985that it did not deal with call sites generated by 'invoke' instructions. In
1986this, and in other situations, you may find that you want to treat
1987<tt>CallInst</tt>s and <tt>InvokeInst</tt>s the same way, even though their
1988most-specific common base class is <tt>Instruction</tt>, which includes lots of
1989less closely-related things. For these cases, LLVM provides a handy wrapper
1990class called <a
Reid Spencerca058542006-03-14 05:39:39 +0000[diff] [blame]1991href="http://llvm.org/doxygen/classllvm_1_1CallSite.html"><tt>CallSite</tt></a>.
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]1992It is essentially a wrapper around an <tt>Instruction</tt> pointer, with some
1993methods that provide functionality common to <tt>CallInst</tt>s and
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]1994<tt>InvokeInst</tt>s.</p>
1995
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]1996<p>This class has "value semantics": it should be passed by value, not by
1997reference and it should not be dynamically allocated or deallocated using
1998<tt>operator new</tt> or <tt>operator delete</tt>. It is efficiently copyable,
1999assignable and constructable, with costs equivalents to that of a bare pointer.
2000If you look at its definition, it has only a single pointer member.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2001
2002</div>
2003
Chris Lattner6755aa12002-09-09 05:49:39 +0000[diff] [blame]2004<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2005<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2006 <a name="iterate_chains">Iterating over def-use &amp; use-def chains</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2007</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2008
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2009<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2010
2011<p>Frequently, we might have an instance of the <a
Chris Lattnerbc5cc2e2007-01-04 22:01:45 +0000[diff] [blame]2012href="/doxygen/classllvm_1_1Value.html">Value Class</a> and we want to
Misha Brukmanfdcb3c22004-06-03 23:29:12 +0000[diff] [blame]2013determine which <tt>User</tt>s use the <tt>Value</tt>. The list of all
2014<tt>User</tt>s of a particular <tt>Value</tt> is called a <i>def-use</i> chain.
2015For example, let's say we have a <tt>Function*</tt> named <tt>F</tt> to a
2016particular function <tt>foo</tt>. Finding all of the instructions that
2017<i>use</i> <tt>foo</tt> is as simple as iterating over the <i>def-use</i> chain
2018of <tt>F</tt>:</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2019
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2020<div class="doc_code">
2021<pre>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2022Function *F = ...;
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2023
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2024for (Value::use_iterator i = F-&gt;use_begin(), e = F-&gt;use_end(); i != e; ++i)
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2025 if (Instruction *Inst = dyn_cast&lt;Instruction&gt;(*i)) {
Chris Lattnerba7cdde2009-09-08 05:15:50 +0000[diff] [blame]2026 errs() &lt;&lt; "F is used in instruction:\n";
2027 errs() &lt;&lt; *Inst &lt;&lt; "\n";
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2028 }
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2029</pre>
Gabor Greif450dfeb2010-03-26 19:35:48 +0000[diff] [blame]2030</div>
2031
Gabor Greif853daf42010-03-26 19:40:38 +0000[diff] [blame]2032<p>Note that dereferencing a <tt>Value::use_iterator</tt> is not a very cheap
Gabor Greif6d6cf8f2010-03-26 19:30:47 +0000[diff] [blame]2033operation. Instead of performing <tt>*i</tt> above several times, consider
Gabor Greif853daf42010-03-26 19:40:38 +0000[diff] [blame]2034doing it only once in the loop body and reusing its result.</p>
Gabor Greif6d6cf8f2010-03-26 19:30:47 +0000[diff] [blame]2035
Gabor Greif66de0492010-03-26 19:04:42 +0000[diff] [blame]2036<p>Alternatively, it's common to have an instance of the <a
Misha Brukmanfdcb3c22004-06-03 23:29:12 +0000[diff] [blame]2037href="/doxygen/classllvm_1_1User.html">User Class</a> and need to know what
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2038<tt>Value</tt>s are used by it. The list of all <tt>Value</tt>s used by a
2039<tt>User</tt> is known as a <i>use-def</i> chain. Instances of class
2040<tt>Instruction</tt> are common <tt>User</tt>s, so we might want to iterate over
2041all of the values that a particular instruction uses (that is, the operands of
2042the particular <tt>Instruction</tt>):</p>
2043
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2044<div class="doc_code">
2045<pre>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2046Instruction *pi = ...;
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2047
2048for (User::op_iterator i = pi-&gt;op_begin(), e = pi-&gt;op_end(); i != e; ++i) {
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2049 Value *v = *i;
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2050 // <i>...</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2051}
2052</pre>
2053</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2054
Gabor Greif6d6cf8f2010-03-26 19:30:47 +0000[diff] [blame]2055<p>Declaring objects as <tt>const</tt> is an important tool of enforcing
Gabor Greif853daf42010-03-26 19:40:38 +0000[diff] [blame]2056mutation free algorithms (such as analyses, etc.). For this purpose above
Gabor Greif6d6cf8f2010-03-26 19:30:47 +0000[diff] [blame]2057iterators come in constant flavors as <tt>Value::const_use_iterator</tt>
2058and <tt>Value::const_op_iterator</tt>. They automatically arise when
2059calling <tt>use/op_begin()</tt> on <tt>const Value*</tt>s or
2060<tt>const User*</tt>s respectively. Upon dereferencing, they return
Gabor Greif853daf42010-03-26 19:40:38 +0000[diff] [blame]2061<tt>const Use*</tt>s. Otherwise the above patterns remain unchanged.</p>
2062
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2063</div>
2064
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2065<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2066<h4>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2067 <a name="iterate_preds">Iterating over predecessors &amp;
2068successors of blocks</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2069</h4>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2070
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2071<div>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2072
2073<p>Iterating over the predecessors and successors of a block is quite easy
2074with the routines defined in <tt>"llvm/Support/CFG.h"</tt>. Just use code like
2075this to iterate over all predecessors of BB:</p>
2076
2077<div class="doc_code">
2078<pre>
2079#include "llvm/Support/CFG.h"
2080BasicBlock *BB = ...;
2081
2082for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
2083 BasicBlock *Pred = *PI;
2084 // <i>...</i>
2085}
2086</pre>
2087</div>
2088
2089<p>Similarly, to iterate over successors use
2090succ_iterator/succ_begin/succ_end.</p>
2091
2092</div>
2093
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2094</div>
Chris Lattner67793d82008-01-03 16:56:04 +0000[diff] [blame]2095
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2096<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2097<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2098 <a name="simplechanges">Making simple changes</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2099</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2100
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2101<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2102
2103<p>There are some primitive transformation operations present in the LLVM
Joel Stanley3ce479c2002-09-11 22:32:24 +0000[diff] [blame]2104infrastructure that are worth knowing about. When performing
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2105transformations, it's fairly common to manipulate the contents of basic
2106blocks. This section describes some of the common methods for doing so
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2107and gives example code.</p>
2108
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2109<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2110<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2111 <a name="schanges_creating">Creating and inserting new
2112 <tt>Instruction</tt>s</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2113</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2114
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2115<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2116
2117<p><i>Instantiating Instructions</i></p>
2118
Chris Lattner4dd45ff2004-05-23 21:06:58 +0000[diff] [blame]2119<p>Creation of <tt>Instruction</tt>s is straight-forward: simply call the
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2120constructor for the kind of instruction to instantiate and provide the necessary
2121parameters. For example, an <tt>AllocaInst</tt> only <i>requires</i> a
2122(const-ptr-to) <tt>Type</tt>. Thus:</p>
2123
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2124<div class="doc_code">
2125<pre>
Nick Lewyckyd93847a2007-12-03 01:52:52 +0000[diff] [blame]2126AllocaInst* ai = new AllocaInst(Type::Int32Ty);
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2127</pre>
2128</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2129
2130<p>will create an <tt>AllocaInst</tt> instance that represents the allocation of
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]2131one integer in the current stack frame, at run time. Each <tt>Instruction</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2132subclass is likely to have varying default parameters which change the semantics
2133of the instruction, so refer to the <a
Misha Brukman18a94d62004-06-03 23:35:54 +0000[diff] [blame]2134href="/doxygen/classllvm_1_1Instruction.html">doxygen documentation for the subclass of
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2135Instruction</a> that you're interested in instantiating.</p>
2136
2137<p><i>Naming values</i></p>
2138
2139<p>It is very useful to name the values of instructions when you're able to, as
2140this facilitates the debugging of your transformations. If you end up looking
2141at generated LLVM machine code, you definitely want to have logical names
2142associated with the results of instructions! By supplying a value for the
2143<tt>Name</tt> (default) parameter of the <tt>Instruction</tt> constructor, you
2144associate a logical name with the result of the instruction's execution at
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]2145run time. For example, say that I'm writing a transformation that dynamically
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2146allocates space for an integer on the stack, and that integer is going to be
2147used as some kind of index by some other code. To accomplish this, I place an
2148<tt>AllocaInst</tt> at the first point in the first <tt>BasicBlock</tt> of some
2149<tt>Function</tt>, and I'm intending to use it within the same
2150<tt>Function</tt>. I might do:</p>
2151
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2152<div class="doc_code">
2153<pre>
Nick Lewyckyd93847a2007-12-03 01:52:52 +0000[diff] [blame]2154AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2155</pre>
2156</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2157
2158<p>where <tt>indexLoc</tt> is now the logical name of the instruction's
Reid Spencer31882b82007-02-03 21:06:43 +0000[diff] [blame]2159execution value, which is a pointer to an integer on the run time stack.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2160
2161<p><i>Inserting instructions</i></p>
2162
2163<p>There are essentially two ways to insert an <tt>Instruction</tt>
2164into an existing sequence of instructions that form a <tt>BasicBlock</tt>:</p>
2165
Joel Stanleyd9ee6c02002-09-18 03:17:23 +0000[diff] [blame]2166<ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2167 <li>Insertion into an explicit instruction list
2168
2169 <p>Given a <tt>BasicBlock* pb</tt>, an <tt>Instruction* pi</tt> within that
2170 <tt>BasicBlock</tt>, and a newly-created instruction we wish to insert
2171 before <tt>*pi</tt>, we do the following: </p>
2172
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2173<div class="doc_code">
2174<pre>
2175BasicBlock *pb = ...;
2176Instruction *pi = ...;
2177Instruction *newInst = new Instruction(...);
2178
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2179pb-&gt;getInstList().insert(pi, newInst); // <i>Inserts newInst before pi in pb</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2180</pre>
2181</div>
Alkis Evlogimenos67150522004-05-27 00:57:51 +0000[diff] [blame]2182
2183 <p>Appending to the end of a <tt>BasicBlock</tt> is so common that
2184 the <tt>Instruction</tt> class and <tt>Instruction</tt>-derived
2185 classes provide constructors which take a pointer to a
2186 <tt>BasicBlock</tt> to be appended to. For example code that
2187 looked like: </p>
2188
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2189<div class="doc_code">
2190<pre>
2191BasicBlock *pb = ...;
2192Instruction *newInst = new Instruction(...);
2193
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2194pb-&gt;getInstList().push_back(newInst); // <i>Appends newInst to pb</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2195</pre>
2196</div>
Alkis Evlogimenos67150522004-05-27 00:57:51 +0000[diff] [blame]2197
2198 <p>becomes: </p>
2199
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2200<div class="doc_code">
2201<pre>
2202BasicBlock *pb = ...;
2203Instruction *newInst = new Instruction(..., pb);
2204</pre>
2205</div>
Alkis Evlogimenos67150522004-05-27 00:57:51 +0000[diff] [blame]2206
2207 <p>which is much cleaner, especially if you are creating
2208 long instruction streams.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2209
2210 <li>Insertion into an implicit instruction list
2211
2212 <p><tt>Instruction</tt> instances that are already in <tt>BasicBlock</tt>s
2213 are implicitly associated with an existing instruction list: the instruction
2214 list of the enclosing basic block. Thus, we could have accomplished the same
2215 thing as the above code without being given a <tt>BasicBlock</tt> by doing:
2216 </p>
2217
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2218<div class="doc_code">
2219<pre>
2220Instruction *pi = ...;
2221Instruction *newInst = new Instruction(...);
2222
2223pi-&gt;getParent()-&gt;getInstList().insert(pi, newInst);
2224</pre>
2225</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2226
2227 <p>In fact, this sequence of steps occurs so frequently that the
2228 <tt>Instruction</tt> class and <tt>Instruction</tt>-derived classes provide
2229 constructors which take (as a default parameter) a pointer to an
2230 <tt>Instruction</tt> which the newly-created <tt>Instruction</tt> should
2231 precede. That is, <tt>Instruction</tt> constructors are capable of
2232 inserting the newly-created instance into the <tt>BasicBlock</tt> of a
2233 provided instruction, immediately before that instruction. Using an
2234 <tt>Instruction</tt> constructor with a <tt>insertBefore</tt> (default)
2235 parameter, the above code becomes:</p>
2236
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2237<div class="doc_code">
2238<pre>
2239Instruction* pi = ...;
2240Instruction* newInst = new Instruction(..., pi);
2241</pre>
2242</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2243
2244 <p>which is much cleaner, especially if you're creating a lot of
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2245 instructions and adding them to <tt>BasicBlock</tt>s.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2246</ul>
2247
2248</div>
2249
2250<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2251<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2252 <a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2253</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2254
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2255<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2256
2257<p>Deleting an instruction from an existing sequence of instructions that form a
Chris Lattner669ce552011-03-24 16:13:31 +0000[diff] [blame]2258<a href="#BasicBlock"><tt>BasicBlock</tt></a> is very straight-forward: just
2259call the instruction's eraseFromParent() method. For example:</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2260
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2261<div class="doc_code">
2262<pre>
2263<a href="#Instruction">Instruction</a> *I = .. ;
Chris Lattner0c463092008-02-15 22:57:17 +0000[diff] [blame]2264I-&gt;eraseFromParent();
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2265</pre>
2266</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2267
Chris Lattner669ce552011-03-24 16:13:31 +0000[diff] [blame]2268<p>This unlinks the instruction from its containing basic block and deletes
2269it. If you'd just like to unlink the instruction from its containing basic
2270block but not delete it, you can use the <tt>removeFromParent()</tt> method.</p>
2271
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2272</div>
2273
2274<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2275<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2276 <a name="schanges_replacing">Replacing an <tt>Instruction</tt> with another
2277 <tt>Value</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2278</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2279
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2280<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2281
2282<p><i>Replacing individual instructions</i></p>
2283
2284<p>Including "<a href="/doxygen/BasicBlockUtils_8h-source.html">llvm/Transforms/Utils/BasicBlockUtils.h</a>"
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2285permits use of two very useful replace functions: <tt>ReplaceInstWithValue</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2286and <tt>ReplaceInstWithInst</tt>.</p>
2287
NAKAMURA Takumi64835132011-04-18 01:17:51 +0000[diff] [blame]2288<h5><a name="schanges_deleting">Deleting <tt>Instruction</tt>s</a></h5>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2289
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2290<ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2291 <li><tt>ReplaceInstWithValue</tt>
2292
Nick Lewycky57368a52008-09-15 06:31:52 +0000[diff] [blame]2293 <p>This function replaces all uses of a given instruction with a value,
2294 and then removes the original instruction. The following example
2295 illustrates the replacement of the result of a particular
Chris Lattnerf7648e72005-01-17 00:12:04 +0000[diff] [blame]2296 <tt>AllocaInst</tt> that allocates memory for a single integer with a null
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2297 pointer to an integer.</p>
2298
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2299<div class="doc_code">
2300<pre>
2301AllocaInst* instToReplace = ...;
2302BasicBlock::iterator ii(instToReplace);
2303
2304ReplaceInstWithValue(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Daniel Dunbarb9e03a62008-10-03 22:17:25 +0000[diff] [blame]2305 Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2306</pre></div></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2307
2308 <li><tt>ReplaceInstWithInst</tt>
2309
2310 <p>This function replaces a particular instruction with another
Nick Lewycky57368a52008-09-15 06:31:52 +0000[diff] [blame]2311 instruction, inserting the new instruction into the basic block at the
2312 location where the old instruction was, and replacing any uses of the old
2313 instruction with the new instruction. The following example illustrates
2314 the replacement of one <tt>AllocaInst</tt> with another.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2315
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2316<div class="doc_code">
2317<pre>
2318AllocaInst* instToReplace = ...;
2319BasicBlock::iterator ii(instToReplace);
2320
2321ReplaceInstWithInst(instToReplace-&gt;getParent()-&gt;getInstList(), ii,
Nick Lewyckyd93847a2007-12-03 01:52:52 +0000[diff] [blame]2322 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2323</pre></div></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2324</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2325
2326<p><i>Replacing multiple uses of <tt>User</tt>s and <tt>Value</tt>s</i></p>
2327
2328<p>You can use <tt>Value::replaceAllUsesWith</tt> and
2329<tt>User::replaceUsesOfWith</tt> to change more than one use at a time. See the
Chris Lattnerbc5cc2e2007-01-04 22:01:45 +0000[diff] [blame]2330doxygen documentation for the <a href="/doxygen/classllvm_1_1Value.html">Value Class</a>
Misha Brukmanfdcb3c22004-06-03 23:29:12 +0000[diff] [blame]2331and <a href="/doxygen/classllvm_1_1User.html">User Class</a>, respectively, for more
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]2332information.</p>
2333
2334<!-- Value::replaceAllUsesWith User::replaceUsesOfWith Point out:
2335include/llvm/Transforms/Utils/ especially BasicBlockUtils.h with:
2336ReplaceInstWithValue, ReplaceInstWithInst -->
2337
2338</div>
2339
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2340<!--_______________________________________________________________________-->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2341<h4>
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2342 <a name="schanges_deletingGV">Deleting <tt>GlobalVariable</tt>s</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2343</h4>
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2344
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2345<div>
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2346
Tanya Lattner872bf1b2007-06-20 20:46:37 +0000[diff] [blame]2347<p>Deleting a global variable from a module is just as easy as deleting an
2348Instruction. First, you must have a pointer to the global variable that you wish
2349 to delete. You use this pointer to erase it from its parent, the module.
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2350 For example:</p>
2351
2352<div class="doc_code">
2353<pre>
2354<a href="#GlobalVariable">GlobalVariable</a> *GV = .. ;
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2355
Tanya Lattner872bf1b2007-06-20 20:46:37 +0000[diff] [blame]2356GV-&gt;eraseFromParent();
Tanya Lattner1a08cf32007-06-20 18:33:15 +0000[diff] [blame]2357</pre>
2358</div>
2359
2360</div>
2361
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2362</div>
2363
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2364<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2365<h3>
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2366 <a name="create_types">How to Create Types</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2367</h3>
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2368
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2369<div>
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2370
2371<p>In generating IR, you may need some complex types. If you know these types
Misha Brukman3268c952009-05-01 20:40:51 +0000[diff] [blame]2372statically, you can use <tt>TypeBuilder&lt;...&gt;::get()</tt>, defined
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2373in <tt>llvm/Support/TypeBuilder.h</tt>, to retrieve them. <tt>TypeBuilder</tt>
2374has two forms depending on whether you're building types for cross-compilation
Misha Brukman3268c952009-05-01 20:40:51 +0000[diff] [blame]2375or native library use. <tt>TypeBuilder&lt;T, true&gt;</tt> requires
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2376that <tt>T</tt> be independent of the host environment, meaning that it's built
2377out of types from
2378the <a href="/doxygen/namespacellvm_1_1types.html"><tt>llvm::types</tt></a>
2379namespace and pointers, functions, arrays, etc. built of
Misha Brukman3268c952009-05-01 20:40:51 +0000[diff] [blame]2380those. <tt>TypeBuilder&lt;T, false&gt;</tt> additionally allows native C types
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2381whose size may depend on the host compiler. For example,</p>
2382
2383<div class="doc_code">
2384<pre>
Misha Brukman3268c952009-05-01 20:40:51 +0000[diff] [blame]2385FunctionType *ft = TypeBuilder&lt;types::i&lt;8&gt;(types::i&lt;32&gt;*), true&gt;::get();
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2386</pre>
2387</div>
2388
2389<p>is easier to read and write than the equivalent</p>
2390
2391<div class="doc_code">
2392<pre>
Owen Andersoneb105f92009-06-16 17:40:28 +0000[diff] [blame]2393std::vector&lt;const Type*&gt; params;
Jeffrey Yasskin590dc3c2009-04-30 22:33:41 +0000[diff] [blame]2394params.push_back(PointerType::getUnqual(Type::Int32Ty));
2395FunctionType *ft = FunctionType::get(Type::Int8Ty, params, false);
2396</pre>
2397</div>
2398
2399<p>See the <a href="/doxygen/TypeBuilder_8h-source.html#l00001">class
2400comment</a> for more details.</p>
2401
2402</div>
2403
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2404</div>
2405
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]2406<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2407<h2>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2408 <a name="threading">Threads and LLVM</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2409</h2>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2410<!-- *********************************************************************** -->
2411
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2412<div>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2413<p>
2414This section describes the interaction of the LLVM APIs with multithreading,
2415both on the part of client applications, and in the JIT, in the hosted
2416application.
2417</p>
2418
2419<p>
2420Note that LLVM's support for multithreading is still relatively young. Up
2421through version 2.5, the execution of threaded hosted applications was
2422supported, but not threaded client access to the APIs. While this use case is
2423now supported, clients <em>must</em> adhere to the guidelines specified below to
2424ensure proper operation in multithreaded mode.
2425</p>
2426
2427<p>
2428Note that, on Unix-like platforms, LLVM requires the presence of GCC's atomic
2429intrinsics in order to support threaded operation. If you need a
2430multhreading-capable LLVM on a platform without a suitably modern system
2431compiler, consider compiling LLVM and LLVM-GCC in single-threaded mode, and
2432using the resultant compiler to build a copy of LLVM with multithreading
2433support.
2434</p>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2435
2436<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2437<h3>
Owen Anderson4741b572009-06-16 18:04:19 +0000[diff] [blame]2438 <a name="startmultithreaded">Entering and Exiting Multithreaded Mode</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2439</h3>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2440
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2441<div>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2442
2443<p>
2444In order to properly protect its internal data structures while avoiding
Owen Anderson4741b572009-06-16 18:04:19 +0000[diff] [blame]2445excessive locking overhead in the single-threaded case, the LLVM must intialize
2446certain data structures necessary to provide guards around its internals. To do
2447so, the client program must invoke <tt>llvm_start_multithreaded()</tt> before
2448making any concurrent LLVM API calls. To subsequently tear down these
2449structures, use the <tt>llvm_stop_multithreaded()</tt> call. You can also use
2450the <tt>llvm_is_multithreaded()</tt> call to check the status of multithreaded
2451mode.
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2452</p>
2453
2454<p>
Owen Anderson4741b572009-06-16 18:04:19 +0000[diff] [blame]2455Note that both of these calls must be made <em>in isolation</em>. That is to
2456say that no other LLVM API calls may be executing at any time during the
2457execution of <tt>llvm_start_multithreaded()</tt> or <tt>llvm_stop_multithreaded
2458</tt>. It's is the client's responsibility to enforce this isolation.
2459</p>
2460
2461<p>
2462The return value of <tt>llvm_start_multithreaded()</tt> indicates the success or
2463failure of the initialization. Failure typically indicates that your copy of
2464LLVM was built without multithreading support, typically because GCC atomic
2465intrinsics were not found in your system compiler. In this case, the LLVM API
2466will not be safe for concurrent calls. However, it <em>will</em> be safe for
Jeffrey Yasskin9fb8ce82010-01-29 19:10:38 +0000[diff] [blame]2467hosting threaded applications in the JIT, though <a href="#jitthreading">care
2468must be taken</a> to ensure that side exits and the like do not accidentally
2469result in concurrent LLVM API calls.
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2470</p>
2471</div>
2472
2473<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2474<h3>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2475 <a name="shutdown">Ending Execution with <tt>llvm_shutdown()</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2476</h3>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2477
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2478<div>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2479<p>
2480When you are done using the LLVM APIs, you should call <tt>llvm_shutdown()</tt>
Owen Anderson4741b572009-06-16 18:04:19 +0000[diff] [blame]2481to deallocate memory used for internal structures. This will also invoke
2482<tt>llvm_stop_multithreaded()</tt> if LLVM is operating in multithreaded mode.
2483As such, <tt>llvm_shutdown()</tt> requires the same isolation guarantees as
2484<tt>llvm_stop_multithreaded()</tt>.
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2485</p>
2486
2487<p>
2488Note that, if you use scope-based shutdown, you can use the
2489<tt>llvm_shutdown_obj</tt> class, which calls <tt>llvm_shutdown()</tt> in its
2490destructor.
2491</div>
2492
2493<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2494<h3>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2495 <a name="managedstatic">Lazy Initialization with <tt>ManagedStatic</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2496</h3>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2497
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2498<div>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2499<p>
2500<tt>ManagedStatic</tt> is a utility class in LLVM used to implement static
2501initialization of static resources, such as the global type tables. Before the
2502invocation of <tt>llvm_shutdown()</tt>, it implements a simple lazy
2503initialization scheme. Once <tt>llvm_start_multithreaded()</tt> returns,
2504however, it uses double-checked locking to implement thread-safe lazy
2505initialization.
2506</p>
2507
2508<p>
2509Note that, because no other threads are allowed to issue LLVM API calls before
2510<tt>llvm_start_multithreaded()</tt> returns, it is possible to have
2511<tt>ManagedStatic</tt>s of <tt>llvm::sys::Mutex</tt>s.
2512</p>
Owen Anderson4741b572009-06-16 18:04:19 +0000[diff] [blame]2513
2514<p>
2515The <tt>llvm_acquire_global_lock()</tt> and <tt>llvm_release_global_lock</tt>
2516APIs provide access to the global lock used to implement the double-checked
2517locking for lazy initialization. These should only be used internally to LLVM,
2518and only if you know what you're doing!
2519</p>
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2520</div>
2521
Owen Andersonfe7bbc92009-08-19 17:58:52 +0000[diff] [blame]2522<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2523<h3>
Owen Andersonfe7bbc92009-08-19 17:58:52 +0000[diff] [blame]2524 <a name="llvmcontext">Achieving Isolation with <tt>LLVMContext</tt></a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2525</h3>
Owen Andersonfe7bbc92009-08-19 17:58:52 +0000[diff] [blame]2526
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2527<div>
Owen Andersonfe7bbc92009-08-19 17:58:52 +0000[diff] [blame]2528<p>
2529<tt>LLVMContext</tt> is an opaque class in the LLVM API which clients can use
2530to operate multiple, isolated instances of LLVM concurrently within the same
2531address space. For instance, in a hypothetical compile-server, the compilation
2532of an individual translation unit is conceptually independent from all the
2533others, and it would be desirable to be able to compile incoming translation
2534units concurrently on independent server threads. Fortunately,
2535<tt>LLVMContext</tt> exists to enable just this kind of scenario!
2536</p>
2537
2538<p>
2539Conceptually, <tt>LLVMContext</tt> provides isolation. Every LLVM entity
2540(<tt>Module</tt>s, <tt>Value</tt>s, <tt>Type</tt>s, <tt>Constant</tt>s, etc.)
Chris Lattnerf6bb2a72009-08-20 03:10:14 +0000[diff] [blame]2541in LLVM's in-memory IR belongs to an <tt>LLVMContext</tt>. Entities in
Owen Andersonfe7bbc92009-08-19 17:58:52 +0000[diff] [blame]2542different contexts <em>cannot</em> interact with each other: <tt>Module</tt>s in
2543different contexts cannot be linked together, <tt>Function</tt>s cannot be added
2544to <tt>Module</tt>s in different contexts, etc. What this means is that is is
2545safe to compile on multiple threads simultaneously, as long as no two threads
2546operate on entities within the same context.
2547</p>
2548
2549<p>
2550In practice, very few places in the API require the explicit specification of a
2551<tt>LLVMContext</tt>, other than the <tt>Type</tt> creation/lookup APIs.
2552Because every <tt>Type</tt> carries a reference to its owning context, most
2553other entities can determine what context they belong to by looking at their
2554own <tt>Type</tt>. If you are adding new entities to LLVM IR, please try to
2555maintain this interface design.
2556</p>
2557
2558<p>
2559For clients that do <em>not</em> require the benefits of isolation, LLVM
2560provides a convenience API <tt>getGlobalContext()</tt>. This returns a global,
2561lazily initialized <tt>LLVMContext</tt> that may be used in situations where
2562isolation is not a concern.
2563</p>
2564</div>
2565
Jeffrey Yasskin9fb8ce82010-01-29 19:10:38 +0000[diff] [blame]2566<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2567<h3>
Jeffrey Yasskin9fb8ce82010-01-29 19:10:38 +0000[diff] [blame]2568 <a name="jitthreading">Threads and the JIT</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2569</h3>
Jeffrey Yasskin9fb8ce82010-01-29 19:10:38 +0000[diff] [blame]2570
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2571<div>
Jeffrey Yasskin9fb8ce82010-01-29 19:10:38 +0000[diff] [blame]2572<p>
2573LLVM's "eager" JIT compiler is safe to use in threaded programs. Multiple
2574threads can call <tt>ExecutionEngine::getPointerToFunction()</tt> or
2575<tt>ExecutionEngine::runFunction()</tt> concurrently, and multiple threads can
2576run code output by the JIT concurrently. The user must still ensure that only
2577one thread accesses IR in a given <tt>LLVMContext</tt> while another thread
2578might be modifying it. One way to do that is to always hold the JIT lock while
2579accessing IR outside the JIT (the JIT <em>modifies</em> the IR by adding
2580<tt>CallbackVH</tt>s). Another way is to only
2581call <tt>getPointerToFunction()</tt> from the <tt>LLVMContext</tt>'s thread.
2582</p>
2583
2584<p>When the JIT is configured to compile lazily (using
2585<tt>ExecutionEngine::DisableLazyCompilation(false)</tt>), there is currently a
2586<a href="http://llvm.org/bugs/show_bug.cgi?id=5184">race condition</a> in
2587updating call sites after a function is lazily-jitted. It's still possible to
2588use the lazy JIT in a threaded program if you ensure that only one thread at a
2589time can call any particular lazy stub and that the JIT lock guards any IR
2590access, but we suggest using only the eager JIT in threaded programs.
2591</p>
2592</div>
2593
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2594</div>
2595
Owen Andersonf0ffb772009-06-16 01:17:16 +0000[diff] [blame]2596<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2597<h2>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2598 <a name="advanced">Advanced Topics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2599</h2>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2600<!-- *********************************************************************** -->
2601
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2602<div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2603<p>
2604This section describes some of the advanced or obscure API's that most clients
2605do not need to be aware of. These API's tend manage the inner workings of the
2606LLVM system, and only need to be accessed in unusual circumstances.
2607</p>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2608
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2609<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2610<h3>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2611 <a name="TypeResolve">LLVM Type Resolution</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2612</h3>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2613
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2614<div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2615
2616<p>
2617The LLVM type system has a very simple goal: allow clients to compare types for
2618structural equality with a simple pointer comparison (aka a shallow compare).
2619This goal makes clients much simpler and faster, and is used throughout the LLVM
2620system.
2621</p>
2622
2623<p>
2624Unfortunately achieving this goal is not a simple matter. In particular,
2625recursive types and late resolution of opaque types makes the situation very
2626difficult to handle. Fortunately, for the most part, our implementation makes
2627most clients able to be completely unaware of the nasty internal details. The
2628primary case where clients are exposed to the inner workings of it are when
Gabor Greifa54634a2007-07-06 22:07:22 +0000[diff] [blame]2629building a recursive type. In addition to this case, the LLVM bitcode reader,
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2630assembly parser, and linker also have to be aware of the inner workings of this
2631system.
2632</p>
2633
Chris Lattnerd1039cc2005-04-25 15:47:57 +0000[diff] [blame]2634<p>
2635For our purposes below, we need three concepts. First, an "Opaque Type" is
2636exactly as defined in the <a href="LangRef.html#t_opaque">language
2637reference</a>. Second an "Abstract Type" is any type which includes an
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]2638opaque type as part of its type graph (for example "<tt>{ opaque, i32 }</tt>").
2639Third, a concrete type is a type that is not an abstract type (e.g. "<tt>{ i32,
Chris Lattnerd1039cc2005-04-25 15:47:57 +0000[diff] [blame]2640float }</tt>").
2641</p>
2642
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2643<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2644<h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2645 <a name="BuildRecType">Basic Recursive Type Construction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2646</h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2647
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2648<div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2649
2650<p>
2651Because the most common question is "how do I build a recursive type with LLVM",
2652we answer it now and explain it as we go. Here we include enough to cause this
2653to be emitted to an output .ll file:
2654</p>
2655
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2656<div class="doc_code">
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2657<pre>
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]2658%mylist = type { %mylist*, i32 }
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2659</pre>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2660</div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2661
2662<p>
2663To build this, use the following LLVM APIs:
2664</p>
2665
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2666<div class="doc_code">
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2667<pre>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2668// <i>Create the initial outer struct</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2669<a href="#PATypeHolder">PATypeHolder</a> StructTy = OpaqueType::get();
2670std::vector&lt;const Type*&gt; Elts;
Daniel Dunbarb9e03a62008-10-03 22:17:25 +0000[diff] [blame]2671Elts.push_back(PointerType::getUnqual(StructTy));
Nick Lewyckyd93847a2007-12-03 01:52:52 +0000[diff] [blame]2672Elts.push_back(Type::Int32Ty);
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2673StructType *NewSTy = StructType::get(Elts);
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2674
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]2675// <i>At this point, NewSTy = "{ opaque*, i32 }". Tell VMCore that</i>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2676// <i>the struct and the opaque type are actually the same.</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2677cast&lt;OpaqueType&gt;(StructTy.get())-&gt;<a href="#refineAbstractTypeTo">refineAbstractTypeTo</a>(NewSTy);
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2678
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2679// <i>NewSTy is potentially invalidated, but StructTy (a <a href="#PATypeHolder">PATypeHolder</a>) is</i>
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2680// <i>kept up-to-date</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2681NewSTy = cast&lt;StructType&gt;(StructTy.get());
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2682
Bill Wendlingf21825f2006-10-11 18:00:22 +0000[diff] [blame]2683// <i>Add a name for the type to the module symbol table (optional)</i>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2684MyModule-&gt;addTypeName("mylist", NewSTy);
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2685</pre>
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]2686</div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2687
2688<p>
2689This code shows the basic approach used to build recursive types: build a
2690non-recursive type using 'opaque', then use type unification to close the cycle.
2691The type unification step is performed by the <tt><a
Chris Lattner694e1102007-02-03 03:06:52 +0000[diff] [blame]2692href="#refineAbstractTypeTo">refineAbstractTypeTo</a></tt> method, which is
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2693described next. After that, we describe the <a
2694href="#PATypeHolder">PATypeHolder class</a>.
2695</p>
2696
2697</div>
2698
2699<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2700<h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2701 <a name="refineAbstractTypeTo">The <tt>refineAbstractTypeTo</tt> method</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2702</h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2703
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2704<div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2705<p>
2706The <tt>refineAbstractTypeTo</tt> method starts the type unification process.
2707While this method is actually a member of the DerivedType class, it is most
2708often used on OpaqueType instances. Type unification is actually a recursive
2709process. After unification, types can become structurally isomorphic to
2710existing types, and all duplicates are deleted (to preserve pointer equality).
2711</p>
2712
2713<p>
2714In the example above, the OpaqueType object is definitely deleted.
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]2715Additionally, if there is an "{ \2*, i32}" type already created in the system,
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2716the pointer and struct type created are <b>also</b> deleted. Obviously whenever
2717a type is deleted, any "Type*" pointers in the program are invalidated. As
2718such, it is safest to avoid having <i>any</i> "Type*" pointers to abstract types
2719live across a call to <tt>refineAbstractTypeTo</tt> (note that non-abstract
2720types can never move or be deleted). To deal with this, the <a
2721href="#PATypeHolder">PATypeHolder</a> class is used to maintain a stable
2722reference to a possibly refined type, and the <a
2723href="#AbstractTypeUser">AbstractTypeUser</a> class is used to update more
2724complex datastructures.
2725</p>
2726
2727</div>
2728
2729<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2730<h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2731 <a name="PATypeHolder">The PATypeHolder Class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2732</h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2733
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2734<div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2735<p>
2736PATypeHolder is a form of a "smart pointer" for Type objects. When VMCore
2737happily goes about nuking types that become isomorphic to existing types, it
2738automatically updates all PATypeHolder objects to point to the new type. In the
2739example above, this allows the code to maintain a pointer to the resultant
2740resolved recursive type, even though the Type*'s are potentially invalidated.
2741</p>
2742
2743<p>
2744PATypeHolder is an extremely light-weight object that uses a lazy union-find
2745implementation to update pointers. For example the pointer from a Value to its
2746Type is maintained by PATypeHolder objects.
2747</p>
2748
2749</div>
2750
2751<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2752<h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2753 <a name="AbstractTypeUser">The AbstractTypeUser Class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2754</h4>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2755
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2756<div>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2757
2758<p>
2759Some data structures need more to perform more complex updates when types get
Chris Lattner406fb4f2007-02-16 04:37:31 +0000[diff] [blame]2760resolved. To support this, a class can derive from the AbstractTypeUser class.
2761This class
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2762allows it to get callbacks when certain types are resolved. To register to get
2763callbacks for a particular type, the DerivedType::{add/remove}AbstractTypeUser
Chris Lattnerd1039cc2005-04-25 15:47:57 +0000[diff] [blame]2764methods can be called on a type. Note that these methods only work for <i>
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]2765 abstract</i> types. Concrete types (those that do not include any opaque
2766objects) can never be refined.
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2767</p>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2768</div>
2769
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2770</div>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2771
2772<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2773<h3>
Chris Lattner406fb4f2007-02-16 04:37:31 +0000[diff] [blame]2774 <a name="SymbolTable">The <tt>ValueSymbolTable</tt> and
2775 <tt>TypeSymbolTable</tt> classes</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2776</h3>
Chris Lattnerfdfb25a2005-04-23 17:27:36 +0000[diff] [blame]2777
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2778<div>
Chris Lattner406fb4f2007-02-16 04:37:31 +0000[diff] [blame]2779<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html">
2780ValueSymbolTable</a></tt> class provides a symbol table that the <a
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2781href="#Function"><tt>Function</tt></a> and <a href="#Module">
Chris Lattner406fb4f2007-02-16 04:37:31 +0000[diff] [blame]2782<tt>Module</tt></a> classes use for naming value definitions. The symbol table
2783can provide a name for any <a href="#Value"><tt>Value</tt></a>.
2784The <tt><a href="http://llvm.org/doxygen/classllvm_1_1TypeSymbolTable.html">
2785TypeSymbolTable</a></tt> class is used by the <tt>Module</tt> class to store
2786names for types.</p>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2787
Reid Spencer9e224a22007-01-07 00:41:39 +0000[diff] [blame]2788<p>Note that the <tt>SymbolTable</tt> class should not be directly accessed
2789by most clients. It should only be used when iteration over the symbol table
2790names themselves are required, which is very special purpose. Note that not
2791all LLVM
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2792<tt><a href="#Value">Value</a></tt>s have names, and those without names (i.e. they have
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2793an empty name) do not exist in the symbol table.
2794</p>
2795
Chris Lattner406fb4f2007-02-16 04:37:31 +0000[diff] [blame]2796<p>These symbol tables support iteration over the values/types in the symbol
2797table with <tt>begin/end/iterator</tt> and supports querying to see if a
2798specific name is in the symbol table (with <tt>lookup</tt>). The
2799<tt>ValueSymbolTable</tt> class exposes no public mutator methods, instead,
2800simply call <tt>setName</tt> on a value, which will autoinsert it into the
2801appropriate symbol table. For types, use the Module::addTypeName method to
2802insert entries into the symbol table.</p>
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2803
Chris Lattnerc6ce71d2005-04-23 16:10:52 +0000[diff] [blame]2804</div>
2805
2806
2807
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2808<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2809<h3>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2810 <a name="UserLayout">The <tt>User</tt> and owned <tt>Use</tt> classes' memory layout</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2811</h3>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2812
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2813<div>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2814<p>The <tt><a href="http://llvm.org/doxygen/classllvm_1_1User.html">
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]2815User</a></tt> class provides a basis for expressing the ownership of <tt>User</tt>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2816towards other <tt><a href="http://llvm.org/doxygen/classllvm_1_1Value.html">
2817Value</a></tt>s. The <tt><a href="http://llvm.org/doxygen/classllvm_1_1Use.html">
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2818Use</a></tt> helper class is employed to do the bookkeeping and to facilitate <i>O(1)</i>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2819addition and removal.</p>
2820
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2821<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2822<h4>
2823 <a name="Use2User">
2824 Interaction and relationship between <tt>User</tt> and <tt>Use</tt> objects
2825 </a>
2826</h4>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2827
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2828<div>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2829<p>
2830A subclass of <tt>User</tt> can choose between incorporating its <tt>Use</tt> objects
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2831or refer to them out-of-line by means of a pointer. A mixed variant
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2832(some <tt>Use</tt>s inline others hung off) is impractical and breaks the invariant
2833that the <tt>Use</tt> objects belonging to the same <tt>User</tt> form a contiguous array.
2834</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2835
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2836<p>
2837We have 2 different layouts in the <tt>User</tt> (sub)classes:
2838<ul>
2839<li><p>Layout a)
2840The <tt>Use</tt> object(s) are inside (resp. at fixed offset) of the <tt>User</tt>
2841object and there are a fixed number of them.</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2842
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2843<li><p>Layout b)
2844The <tt>Use</tt> object(s) are referenced by a pointer to an
2845array from the <tt>User</tt> object and there may be a variable
2846number of them.</p>
2847</ul>
2848<p>
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2849As of v2.4 each layout still possesses a direct pointer to the
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2850start of the array of <tt>Use</tt>s. Though not mandatory for layout a),
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2851we stick to this redundancy for the sake of simplicity.
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2852The <tt>User</tt> object also stores the number of <tt>Use</tt> objects it
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2853has. (Theoretically this information can also be calculated
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2854given the scheme presented below.)</p>
2855<p>
2856Special forms of allocation operators (<tt>operator new</tt>)
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2857enforce the following memory layouts:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2858
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2859<ul>
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2860<li><p>Layout a) is modelled by prepending the <tt>User</tt> object by the <tt>Use[]</tt> array.</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2861
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2862<pre>
2863...---.---.---.---.-------...
2864 | P | P | P | P | User
2865'''---'---'---'---'-------'''
2866</pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2867
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2868<li><p>Layout b) is modelled by pointing at the <tt>Use[]</tt> array.</p>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2869<pre>
2870.-------...
2871| User
2872'-------'''
2873 |
2874 v
2875 .---.---.---.---...
2876 | P | P | P | P |
2877 '---'---'---'---'''
2878</pre>
2879</ul>
2880<i>(In the above figures '<tt>P</tt>' stands for the <tt>Use**</tt> that
2881 is stored in each <tt>Use</tt> object in the member <tt>Use::Prev</tt>)</i>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2882
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2883</div>
2884
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2885<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2886<h4>
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]2887 <a name="Waymarking">The waymarking algorithm</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2888</h4>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2889
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2890<div>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2891<p>
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2892Since the <tt>Use</tt> objects are deprived of the direct (back)pointer to
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2893their <tt>User</tt> objects, there must be a fast and exact method to
2894recover it. This is accomplished by the following scheme:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2895
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]2896A bit-encoding in the 2 LSBits (least significant bits) of the <tt>Use::Prev</tt> allows to find the
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2897start of the <tt>User</tt> object:
2898<ul>
2899<li><tt>00</tt> &mdash;&gt; binary digit 0</li>
2900<li><tt>01</tt> &mdash;&gt; binary digit 1</li>
2901<li><tt>10</tt> &mdash;&gt; stop and calculate (<tt>s</tt>)</li>
2902<li><tt>11</tt> &mdash;&gt; full stop (<tt>S</tt>)</li>
2903</ul>
2904<p>
2905Given a <tt>Use*</tt>, all we have to do is to walk till we get
2906a stop and we either have a <tt>User</tt> immediately behind or
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2907we have to walk to the next stop picking up digits
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2908and calculating the offset:</p>
2909<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2910.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
2911| 1 | s | 1 | 0 | 1 | 0 | s | 1 | 1 | 0 | s | 1 | 1 | s | 1 | S | User (or User*)
2912'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
2913 |+15 |+10 |+6 |+3 |+1
2914 | | | | |__>
2915 | | | |__________>
2916 | | |______________________>
2917 | |______________________________________>
2918 |__________________________________________________________>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2919</pre>
2920<p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2921Only the significant number of bits need to be stored between the
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2922stops, so that the <i>worst case is 20 memory accesses</i> when there are
29231000 <tt>Use</tt> objects associated with a <tt>User</tt>.</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2924
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2925</div>
2926
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2927<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2928<h4>
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]2929 <a name="ReferenceImpl">Reference implementation</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]2930</h4>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2931
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]2932<div>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2933<p>
2934The following literate Haskell fragment demonstrates the concept:</p>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2935
2936<div class="doc_code">
2937<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2938> import Test.QuickCheck
2939>
2940> digits :: Int -> [Char] -> [Char]
2941> digits 0 acc = '0' : acc
2942> digits 1 acc = '1' : acc
2943> digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
2944>
2945> dist :: Int -> [Char] -> [Char]
2946> dist 0 [] = ['S']
2947> dist 0 acc = acc
2948> dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
2949> dist n acc = dist (n - 1) $ dist 1 acc
2950>
2951> takeLast n ss = reverse $ take n $ reverse ss
2952>
2953> test = takeLast 40 $ dist 20 []
2954>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2955</pre>
2956</div>
2957<p>
2958Printing &lt;test&gt; gives: <tt>"1s100000s11010s10100s1111s1010s110s11s1S"</tt></p>
2959<p>
2960The reverse algorithm computes the length of the string just by examining
2961a certain prefix:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2962
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2963<div class="doc_code">
2964<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2965> pref :: [Char] -> Int
2966> pref "S" = 1
2967> pref ('s':'1':rest) = decode 2 1 rest
2968> pref (_:rest) = 1 + pref rest
2969>
2970> decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
2971> decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
2972> decode walk acc _ = walk + acc
2973>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2974</pre>
2975</div>
2976<p>
2977Now, as expected, printing &lt;pref test&gt; gives <tt>40</tt>.</p>
2978<p>
2979We can <i>quickCheck</i> this with following property:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2980
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2981<div class="doc_code">
2982<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2983> testcase = dist 2000 []
2984> testcaseLength = length testcase
2985>
2986> identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
2987> where arr = takeLast n testcase
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2988>
2989</pre>
2990</div>
2991<p>
2992As expected &lt;quickCheck identityProp&gt; gives:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2993
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2994<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]2995*Main> quickCheck identityProp
2996OK, passed 100 tests.
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]2997</pre>
2998<p>
2999Let's be a bit more exhaustive:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]3000
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3001<div class="doc_code">
3002<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]3003>
3004> deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
3005>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3006</pre>
3007</div>
3008<p>
3009And here is the result of &lt;deepCheck identityProp&gt;:</p>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]3010
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3011<pre>
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]3012*Main> deepCheck identityProp
3013OK, passed 500 tests.
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]3014</pre>
3015
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3016</div>
3017
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3018<!-- ______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3019<h4>
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]3020 <a name="Tagging">Tagging considerations</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3021</h4>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3022
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3023<div>
3024
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3025<p>
3026To maintain the invariant that the 2 LSBits of each <tt>Use**</tt> in <tt>Use</tt>
3027never change after being set up, setters of <tt>Use::Prev</tt> must re-tag the
3028new <tt>Use**</tt> on every modification. Accordingly getters must strip the
3029tag bits.</p>
3030<p>
Gabor Greif845eb8e2008-06-25 00:10:22 +0000[diff] [blame]3031For layout b) instead of the <tt>User</tt> we find a pointer (<tt>User*</tt> with LSBit set).
3032Following this pointer brings us to the <tt>User</tt>. A portable trick ensures
3033that the first bytes of <tt>User</tt> (if interpreted as a pointer) never has
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]3034the LSBit set. (Portability is relying on the fact that all known compilers place the
3035<tt>vptr</tt> in the first word of the instances.)</p>
Gabor Greif095502e2008-06-18 13:44:57 +0000[diff] [blame]3036
Gabor Greif5c28a6c2008-06-16 21:06:12 +0000[diff] [blame]3037</div>
3038
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3039</div>
3040
3041</div>
3042
3043<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3044<h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3045 <a name="coreclasses">The Core LLVM Class Hierarchy Reference </a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3046</h2>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3047<!-- *********************************************************************** -->
3048
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3049<div>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3050<p><tt>#include "<a href="/doxygen/Type_8h-source.html">llvm/Type.h</a>"</tt>
3051<br>doxygen info: <a href="/doxygen/classllvm_1_1Type.html">Type Class</a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3052
3053<p>The Core LLVM classes are the primary means of representing the program
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3054being inspected or transformed. The core LLVM classes are defined in
3055header files in the <tt>include/llvm/</tt> directory, and implemented in
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3056the <tt>lib/VMCore</tt> directory.</p>
3057
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3058<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3059<h3>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3060 <a name="Type">The <tt>Type</tt> class and Derived Types</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3061</h3>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3062
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3063<div>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3064
3065 <p><tt>Type</tt> is a superclass of all type classes. Every <tt>Value</tt> has
3066 a <tt>Type</tt>. <tt>Type</tt> cannot be instantiated directly but only
3067 through its subclasses. Certain primitive types (<tt>VoidType</tt>,
3068 <tt>LabelType</tt>, <tt>FloatType</tt> and <tt>DoubleType</tt>) have hidden
3069 subclasses. They are hidden because they offer no useful functionality beyond
3070 what the <tt>Type</tt> class offers except to distinguish themselves from
3071 other subclasses of <tt>Type</tt>.</p>
3072 <p>All other types are subclasses of <tt>DerivedType</tt>. Types can be
3073 named, but this is not a requirement. There exists exactly
3074 one instance of a given shape at any one time. This allows type equality to
3075 be performed with address equality of the Type Instance. That is, given two
3076 <tt>Type*</tt> values, the types are identical if the pointers are identical.
3077 </p>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3078
3079<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3080<h4>
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]3081 <a name="m_Type">Important Public Methods</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3082</h4>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3083
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3084<div>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3085
3086<ul>
Duncan Sands9dff9be2010-02-15 16:12:20 +0000[diff] [blame]3087 <li><tt>bool isIntegerTy() const</tt>: Returns true for any integer type.</li>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3088
Duncan Sands9dff9be2010-02-15 16:12:20 +0000[diff] [blame]3089 <li><tt>bool isFloatingPointTy()</tt>: Return true if this is one of the five
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3090 floating point types.</li>
3091
3092 <li><tt>bool isAbstract()</tt>: Return true if the type is abstract (contains
3093 an OpaqueType anywhere in its definition).</li>
3094
3095 <li><tt>bool isSized()</tt>: Return true if the type has known size. Things
3096 that don't have a size are abstract types, labels and void.</li>
3097
3098</ul>
3099</div>
3100
3101<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3102<h4>
Gabor Greif2231c2c2009-01-05 16:05:32 +0000[diff] [blame]3103 <a name="derivedtypes">Important Derived Types</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3104</h4>
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3105<div>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3106<dl>
3107 <dt><tt>IntegerType</tt></dt>
3108 <dd>Subclass of DerivedType that represents integer types of any bit width.
3109 Any bit width between <tt>IntegerType::MIN_INT_BITS</tt> (1) and
3110 <tt>IntegerType::MAX_INT_BITS</tt> (~8 million) can be represented.
3111 <ul>
3112 <li><tt>static const IntegerType* get(unsigned NumBits)</tt>: get an integer
3113 type of a specific bit width.</li>
3114 <li><tt>unsigned getBitWidth() const</tt>: Get the bit width of an integer
3115 type.</li>
3116 </ul>
3117 </dd>
3118 <dt><tt>SequentialType</tt></dt>
3119 <dd>This is subclassed by ArrayType and PointerType
3120 <ul>
3121 <li><tt>const Type * getElementType() const</tt>: Returns the type of each
3122 of the elements in the sequential type. </li>
3123 </ul>
3124 </dd>
3125 <dt><tt>ArrayType</tt></dt>
3126 <dd>This is a subclass of SequentialType and defines the interface for array
3127 types.
3128 <ul>
3129 <li><tt>unsigned getNumElements() const</tt>: Returns the number of
3130 elements in the array. </li>
3131 </ul>
3132 </dd>
3133 <dt><tt>PointerType</tt></dt>
Chris Lattner099213b2007-02-03 03:05:57 +0000[diff] [blame]3134 <dd>Subclass of SequentialType for pointer types.</dd>
Reid Spencerd84d35b2007-02-15 02:26:10 +0000[diff] [blame]3135 <dt><tt>VectorType</tt></dt>
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]3136 <dd>Subclass of SequentialType for vector types. A
3137 vector type is similar to an ArrayType but is distinguished because it is
Benjamin Kramer0f420382009-10-12 14:46:08 +0000[diff] [blame]3138 a first class type whereas ArrayType is not. Vector types are used for
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3139 vector operations and are usually small vectors of of an integer or floating
3140 point type.</dd>
3141 <dt><tt>StructType</tt></dt>
3142 <dd>Subclass of DerivedTypes for struct types.</dd>
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]3143 <dt><tt><a name="FunctionType">FunctionType</a></tt></dt>
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3144 <dd>Subclass of DerivedTypes for function types.
3145 <ul>
Dan Gohman39027c42010-03-30 20:04:57 +0000[diff] [blame]3146 <li><tt>bool isVarArg() const</tt>: Returns true if it's a vararg
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3147 function</li>
3148 <li><tt> const Type * getReturnType() const</tt>: Returns the
3149 return type of the function.</li>
3150 <li><tt>const Type * getParamType (unsigned i)</tt>: Returns
3151 the type of the ith parameter.</li>
3152 <li><tt> const unsigned getNumParams() const</tt>: Returns the
3153 number of formal parameters.</li>
3154 </ul>
3155 </dd>
3156 <dt><tt>OpaqueType</tt></dt>
3157 <dd>Sublcass of DerivedType for abstract types. This class
3158 defines no content and is used as a placeholder for some other type. Note
3159 that OpaqueType is used (temporarily) during type resolution for forward
3160 references of types. Once the referenced type is resolved, the OpaqueType
3161 is replaced with the actual type. OpaqueType can also be used for data
3162 abstraction. At link time opaque types can be resolved to actual types
3163 of the same name.</dd>
3164</dl>
3165</div>
3166
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3167</div>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3168
3169<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3170<h3>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3171 <a name="Module">The <tt>Module</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3172</h3>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3173
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3174<div>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3175
3176<p><tt>#include "<a
3177href="/doxygen/Module_8h-source.html">llvm/Module.h</a>"</tt><br> doxygen info:
3178<a href="/doxygen/classllvm_1_1Module.html">Module Class</a></p>
3179
3180<p>The <tt>Module</tt> class represents the top level structure present in LLVM
3181programs. An LLVM module is effectively either a translation unit of the
3182original program or a combination of several translation units merged by the
3183linker. The <tt>Module</tt> class keeps track of a list of <a
3184href="#Function"><tt>Function</tt></a>s, a list of <a
3185href="#GlobalVariable"><tt>GlobalVariable</tt></a>s, and a <a
3186href="#SymbolTable"><tt>SymbolTable</tt></a>. Additionally, it contains a few
3187helpful member functions that try to make common operations easy.</p>
3188
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3189<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3190<h4>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3191 <a name="m_Module">Important Public Members of the <tt>Module</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3192</h4>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3193
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3194<div>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3195
3196<ul>
3197 <li><tt>Module::Module(std::string name = "")</tt></li>
3198</ul>
3199
3200<p>Constructing a <a href="#Module">Module</a> is easy. You can optionally
3201provide a name for it (probably based on the name of the translation unit).</p>
3202
3203<ul>
3204 <li><tt>Module::iterator</tt> - Typedef for function list iterator<br>
3205 <tt>Module::const_iterator</tt> - Typedef for const_iterator.<br>
3206
3207 <tt>begin()</tt>, <tt>end()</tt>
3208 <tt>size()</tt>, <tt>empty()</tt>
3209
3210 <p>These are forwarding methods that make it easy to access the contents of
3211 a <tt>Module</tt> object's <a href="#Function"><tt>Function</tt></a>
3212 list.</p></li>
3213
3214 <li><tt>Module::FunctionListType &amp;getFunctionList()</tt>
3215
3216 <p> Returns the list of <a href="#Function"><tt>Function</tt></a>s. This is
3217 necessary to use when you need to update the list or perform a complex
3218 action that doesn't have a forwarding method.</p>
3219
3220 <p><!-- Global Variable --></p></li>
3221</ul>
3222
3223<hr>
3224
3225<ul>
3226 <li><tt>Module::global_iterator</tt> - Typedef for global variable list iterator<br>
3227
3228 <tt>Module::const_global_iterator</tt> - Typedef for const_iterator.<br>
3229
3230 <tt>global_begin()</tt>, <tt>global_end()</tt>
3231 <tt>global_size()</tt>, <tt>global_empty()</tt>
3232
3233 <p> These are forwarding methods that make it easy to access the contents of
3234 a <tt>Module</tt> object's <a
3235 href="#GlobalVariable"><tt>GlobalVariable</tt></a> list.</p></li>
3236
3237 <li><tt>Module::GlobalListType &amp;getGlobalList()</tt>
3238
3239 <p>Returns the list of <a
3240 href="#GlobalVariable"><tt>GlobalVariable</tt></a>s. This is necessary to
3241 use when you need to update the list or perform a complex action that
3242 doesn't have a forwarding method.</p>
3243
3244 <p><!-- Symbol table stuff --> </p></li>
3245</ul>
3246
3247<hr>
3248
3249<ul>
3250 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
3251
3252 <p>Return a reference to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3253 for this <tt>Module</tt>.</p>
3254
3255 <p><!-- Convenience methods --></p></li>
3256</ul>
3257
3258<hr>
3259
3260<ul>
3261 <li><tt><a href="#Function">Function</a> *getFunction(const std::string
3262 &amp;Name, const <a href="#FunctionType">FunctionType</a> *Ty)</tt>
3263
3264 <p>Look up the specified function in the <tt>Module</tt> <a
3265 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, return
3266 <tt>null</tt>.</p></li>
3267
3268 <li><tt><a href="#Function">Function</a> *getOrInsertFunction(const
3269 std::string &amp;Name, const <a href="#FunctionType">FunctionType</a> *T)</tt>
3270
3271 <p>Look up the specified function in the <tt>Module</tt> <a
3272 href="#SymbolTable"><tt>SymbolTable</tt></a>. If it does not exist, add an
3273 external declaration for the function and return it.</p></li>
3274
3275 <li><tt>std::string getTypeName(const <a href="#Type">Type</a> *Ty)</tt>
3276
3277 <p>If there is at least one entry in the <a
3278 href="#SymbolTable"><tt>SymbolTable</tt></a> for the specified <a
3279 href="#Type"><tt>Type</tt></a>, return it. Otherwise return the empty
3280 string.</p></li>
3281
3282 <li><tt>bool addTypeName(const std::string &amp;Name, const <a
3283 href="#Type">Type</a> *Ty)</tt>
3284
3285 <p>Insert an entry in the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
3286 mapping <tt>Name</tt> to <tt>Ty</tt>. If there is already an entry for this
3287 name, true is returned and the <a
3288 href="#SymbolTable"><tt>SymbolTable</tt></a> is not modified.</p></li>
3289</ul>
3290
3291</div>
3292
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3293</div>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3294
Reid Spencer2759b182007-01-12 17:26:25 +0000[diff] [blame]3295<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3296<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3297 <a name="Value">The <tt>Value</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3298</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3299
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3300<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3301
3302<p><tt>#include "<a href="/doxygen/Value_8h-source.html">llvm/Value.h</a>"</tt>
3303<br>
Chris Lattnerbc5cc2e2007-01-04 22:01:45 +0000[diff] [blame]3304doxygen info: <a href="/doxygen/classllvm_1_1Value.html">Value Class</a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3305
3306<p>The <tt>Value</tt> class is the most important class in the LLVM Source
3307base. It represents a typed value that may be used (among other things) as an
3308operand to an instruction. There are many different types of <tt>Value</tt>s,
3309such as <a href="#Constant"><tt>Constant</tt></a>s,<a
3310href="#Argument"><tt>Argument</tt></a>s. Even <a
3311href="#Instruction"><tt>Instruction</tt></a>s and <a
3312href="#Function"><tt>Function</tt></a>s are <tt>Value</tt>s.</p>
3313
3314<p>A particular <tt>Value</tt> may be used many times in the LLVM representation
3315for a program. For example, an incoming argument to a function (represented
3316with an instance of the <a href="#Argument">Argument</a> class) is "used" by
3317every instruction in the function that references the argument. To keep track
3318of this relationship, the <tt>Value</tt> class keeps a list of all of the <a
3319href="#User"><tt>User</tt></a>s that is using it (the <a
3320href="#User"><tt>User</tt></a> class is a base class for all nodes in the LLVM
3321graph that can refer to <tt>Value</tt>s). This use list is how LLVM represents
3322def-use information in the program, and is accessible through the <tt>use_</tt>*
3323methods, shown below.</p>
3324
3325<p>Because LLVM is a typed representation, every LLVM <tt>Value</tt> is typed,
3326and this <a href="#Type">Type</a> is available through the <tt>getType()</tt>
3327method. In addition, all LLVM values can be named. The "name" of the
3328<tt>Value</tt> is a symbolic string printed in the LLVM code:</p>
3329
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]3330<div class="doc_code">
3331<pre>
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]3332%<b>foo</b> = add i32 1, 2
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]3333</pre>
3334</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3335
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]3336<p><a name="nameWarning">The name of this instruction is "foo".</a> <b>NOTE</b>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3337that the name of any value may be missing (an empty string), so names should
3338<b>ONLY</b> be used for debugging (making the source code easier to read,
3339debugging printouts), they should not be used to keep track of values or map
3340between them. For this purpose, use a <tt>std::map</tt> of pointers to the
3341<tt>Value</tt> itself instead.</p>
3342
3343<p>One important aspect of LLVM is that there is no distinction between an SSA
3344variable and the operation that produces it. Because of this, any reference to
3345the value produced by an instruction (or the value available as an incoming
Chris Lattner0b41ebe2004-03-18 14:58:55 +0000[diff] [blame]3346argument, for example) is represented as a direct pointer to the instance of
3347the class that
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3348represents this value. Although this may take some getting used to, it
3349simplifies the representation and makes it easier to manipulate.</p>
3350
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3351<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3352<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3353 <a name="m_Value">Important Public Members of the <tt>Value</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3354</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3355
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3356<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3357
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3358<ul>
3359 <li><tt>Value::use_iterator</tt> - Typedef for iterator over the
3360use-list<br>
Gabor Greif108abc62010-03-26 19:59:25 +0000[diff] [blame]3361 <tt>Value::const_use_iterator</tt> - Typedef for const_iterator over
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3362the use-list<br>
3363 <tt>unsigned use_size()</tt> - Returns the number of users of the
3364value.<br>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]3365 <tt>bool use_empty()</tt> - Returns true if there are no users.<br>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3366 <tt>use_iterator use_begin()</tt> - Get an iterator to the start of
3367the use-list.<br>
3368 <tt>use_iterator use_end()</tt> - Get an iterator to the end of the
3369use-list.<br>
3370 <tt><a href="#User">User</a> *use_back()</tt> - Returns the last
3371element in the list.
3372 <p> These methods are the interface to access the def-use
3373information in LLVM. As with all other iterators in LLVM, the naming
3374conventions follow the conventions defined by the <a href="#stl">STL</a>.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3375 </li>
3376 <li><tt><a href="#Type">Type</a> *getType() const</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3377 <p>This method returns the Type of the Value.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3378 </li>
3379 <li><tt>bool hasName() const</tt><br>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]3380 <tt>std::string getName() const</tt><br>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3381 <tt>void setName(const std::string &amp;Name)</tt>
3382 <p> This family of methods is used to access and assign a name to a <tt>Value</tt>,
3383be aware of the <a href="#nameWarning">precaution above</a>.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3384 </li>
3385 <li><tt>void replaceAllUsesWith(Value *V)</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3386
3387 <p>This method traverses the use list of a <tt>Value</tt> changing all <a
3388 href="#User"><tt>User</tt>s</a> of the current value to refer to
3389 "<tt>V</tt>" instead. For example, if you detect that an instruction always
3390 produces a constant value (for example through constant folding), you can
3391 replace all uses of the instruction with the constant like this:</p>
3392
Bill Wendling34ab0672006-10-11 06:30:10 +0000[diff] [blame]3393<div class="doc_code">
3394<pre>
3395Inst-&gt;replaceAllUsesWith(ConstVal);
3396</pre>
3397</div>
3398
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3399</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3400
3401</div>
3402
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3403</div>
3404
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3405<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3406<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3407 <a name="User">The <tt>User</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3408</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3409
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3410<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3411
3412<p>
3413<tt>#include "<a href="/doxygen/User_8h-source.html">llvm/User.h</a>"</tt><br>
Misha Brukmanfdcb3c22004-06-03 23:29:12 +0000[diff] [blame]3414doxygen info: <a href="/doxygen/classllvm_1_1User.html">User Class</a><br>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3415Superclass: <a href="#Value"><tt>Value</tt></a></p>
3416
3417<p>The <tt>User</tt> class is the common base class of all LLVM nodes that may
3418refer to <a href="#Value"><tt>Value</tt></a>s. It exposes a list of "Operands"
3419that are all of the <a href="#Value"><tt>Value</tt></a>s that the User is
3420referring to. The <tt>User</tt> class itself is a subclass of
3421<tt>Value</tt>.</p>
3422
3423<p>The operands of a <tt>User</tt> point directly to the LLVM <a
3424href="#Value"><tt>Value</tt></a> that it refers to. Because LLVM uses Static
3425Single Assignment (SSA) form, there can only be one definition referred to,
3426allowing this direct connection. This connection provides the use-def
3427information in LLVM.</p>
3428
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3429<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3430<h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3431 <a name="m_User">Important Public Members of the <tt>User</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3432</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3433
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3434<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3435
3436<p>The <tt>User</tt> class exposes the operand list in two ways: through
3437an index access interface and through an iterator based interface.</p>
3438
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3439<ul>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3440 <li><tt>Value *getOperand(unsigned i)</tt><br>
3441 <tt>unsigned getNumOperands()</tt>
3442 <p> These two methods expose the operands of the <tt>User</tt> in a
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3443convenient form for direct access.</p></li>
3444
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3445 <li><tt>User::op_iterator</tt> - Typedef for iterator over the operand
3446list<br>
Chris Lattnerf7648e72005-01-17 00:12:04 +0000[diff] [blame]3447 <tt>op_iterator op_begin()</tt> - Get an iterator to the start of
3448the operand list.<br>
3449 <tt>op_iterator op_end()</tt> - Get an iterator to the end of the
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3450operand list.
3451 <p> Together, these methods make up the iterator based interface to
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3452the operands of a <tt>User</tt>.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3453</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3454
3455</div>
3456
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3457</div>
3458
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3459<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3460<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3461 <a name="Instruction">The <tt>Instruction</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3462</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3463
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3464<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3465
3466<p><tt>#include "</tt><tt><a
3467href="/doxygen/Instruction_8h-source.html">llvm/Instruction.h</a>"</tt><br>
Misha Brukman18a94d62004-06-03 23:35:54 +0000[diff] [blame]3468doxygen info: <a href="/doxygen/classllvm_1_1Instruction.html">Instruction Class</a><br>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3469Superclasses: <a href="#User"><tt>User</tt></a>, <a
3470href="#Value"><tt>Value</tt></a></p>
3471
3472<p>The <tt>Instruction</tt> class is the common base class for all LLVM
3473instructions. It provides only a few methods, but is a very commonly used
3474class. The primary data tracked by the <tt>Instruction</tt> class itself is the
3475opcode (instruction type) and the parent <a
3476href="#BasicBlock"><tt>BasicBlock</tt></a> the <tt>Instruction</tt> is embedded
3477into. To represent a specific type of instruction, one of many subclasses of
3478<tt>Instruction</tt> are used.</p>
3479
3480<p> Because the <tt>Instruction</tt> class subclasses the <a
3481href="#User"><tt>User</tt></a> class, its operands can be accessed in the same
3482way as for other <a href="#User"><tt>User</tt></a>s (with the
3483<tt>getOperand()</tt>/<tt>getNumOperands()</tt> and
3484<tt>op_begin()</tt>/<tt>op_end()</tt> methods).</p> <p> An important file for
3485the <tt>Instruction</tt> class is the <tt>llvm/Instruction.def</tt> file. This
3486file contains some meta-data about the various different types of instructions
3487in LLVM. It describes the enum values that are used as opcodes (for example
Reid Spencerc6be3862006-12-19 19:47:19 +0000[diff] [blame]3488<tt>Instruction::Add</tt> and <tt>Instruction::ICmp</tt>), as well as the
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3489concrete sub-classes of <tt>Instruction</tt> that implement the instruction (for
3490example <tt><a href="#BinaryOperator">BinaryOperator</a></tt> and <tt><a
Reid Spencerc6be3862006-12-19 19:47:19 +0000[diff] [blame]3491href="#CmpInst">CmpInst</a></tt>). Unfortunately, the use of macros in
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3492this file confuses doxygen, so these enum values don't show up correctly in the
Misha Brukman18a94d62004-06-03 23:35:54 +0000[diff] [blame]3493<a href="/doxygen/classllvm_1_1Instruction.html">doxygen output</a>.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3494
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3495<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3496<h4>
3497 <a name="s_Instruction">
3498 Important Subclasses of the <tt>Instruction</tt> class
3499 </a>
3500</h4>
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3501<div>
Reid Spencerc6be3862006-12-19 19:47:19 +0000[diff] [blame]3502 <ul>
3503 <li><tt><a name="BinaryOperator">BinaryOperator</a></tt>
3504 <p>This subclasses represents all two operand instructions whose operands
3505 must be the same type, except for the comparison instructions.</p></li>
3506 <li><tt><a name="CastInst">CastInst</a></tt>
3507 <p>This subclass is the parent of the 12 casting instructions. It provides
3508 common operations on cast instructions.</p>
3509 <li><tt><a name="CmpInst">CmpInst</a></tt>
3510 <p>This subclass respresents the two comparison instructions,
3511 <a href="LangRef.html#i_icmp">ICmpInst</a> (integer opreands), and
3512 <a href="LangRef.html#i_fcmp">FCmpInst</a> (floating point operands).</p>
3513 <li><tt><a name="TerminatorInst">TerminatorInst</a></tt>
3514 <p>This subclass is the parent of all terminator instructions (those which
3515 can terminate a block).</p>
3516 </ul>
3517 </div>
3518
3519<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3520<h4>
3521 <a name="m_Instruction">
3522 Important Public Members of the <tt>Instruction</tt> class
3523 </a>
3524</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3525
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3526<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3527
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3528<ul>
3529 <li><tt><a href="#BasicBlock">BasicBlock</a> *getParent()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3530 <p>Returns the <a href="#BasicBlock"><tt>BasicBlock</tt></a> that
3531this <tt>Instruction</tt> is embedded into.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3532 <li><tt>bool mayWriteToMemory()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3533 <p>Returns true if the instruction writes to memory, i.e. it is a
3534 <tt>call</tt>,<tt>free</tt>,<tt>invoke</tt>, or <tt>store</tt>.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3535 <li><tt>unsigned getOpcode()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3536 <p>Returns the opcode for the <tt>Instruction</tt>.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3537 <li><tt><a href="#Instruction">Instruction</a> *clone() const</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3538 <p>Returns another instance of the specified instruction, identical
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3539in all ways to the original except that the instruction has no parent
3540(ie it's not embedded into a <a href="#BasicBlock"><tt>BasicBlock</tt></a>),
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3541and it has no name</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3542</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3543
3544</div>
3545
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3546</div>
3547
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3548<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3549<h3>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3550 <a name="Constant">The <tt>Constant</tt> class and subclasses</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3551</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3552
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3553<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3554
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3555<p>Constant represents a base class for different types of constants. It
3556is subclassed by ConstantInt, ConstantArray, etc. for representing
3557the various types of Constants. <a href="#GlobalValue">GlobalValue</a> is also
3558a subclass, which represents the address of a global variable or function.
3559</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3560
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3561<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3562<h4>Important Subclasses of Constant</h4>
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3563<div>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3564<ul>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3565 <li>ConstantInt : This subclass of Constant represents an integer constant of
3566 any width.
3567 <ul>
Reid Spencerbbd97b12007-03-01 21:05:33 +0000[diff] [blame]3568 <li><tt>const APInt&amp; getValue() const</tt>: Returns the underlying
3569 value of this constant, an APInt value.</li>
3570 <li><tt>int64_t getSExtValue() const</tt>: Converts the underlying APInt
3571 value to an int64_t via sign extension. If the value (not the bit width)
3572 of the APInt is too large to fit in an int64_t, an assertion will result.
3573 For this reason, use of this method is discouraged.</li>
3574 <li><tt>uint64_t getZExtValue() const</tt>: Converts the underlying APInt
3575 value to a uint64_t via zero extension. IF the value (not the bit width)
3576 of the APInt is too large to fit in a uint64_t, an assertion will result.
Reid Spencer91643aa2007-03-02 01:31:31 +0000[diff] [blame]3577 For this reason, use of this method is discouraged.</li>
Reid Spencerbbd97b12007-03-01 21:05:33 +0000[diff] [blame]3578 <li><tt>static ConstantInt* get(const APInt&amp; Val)</tt>: Returns the
3579 ConstantInt object that represents the value provided by <tt>Val</tt>.
3580 The type is implied as the IntegerType that corresponds to the bit width
3581 of <tt>Val</tt>.</li>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3582 <li><tt>static ConstantInt* get(const Type *Ty, uint64_t Val)</tt>:
3583 Returns the ConstantInt object that represents the value provided by
3584 <tt>Val</tt> for integer type <tt>Ty</tt>.</li>
3585 </ul>
3586 </li>
3587 <li>ConstantFP : This class represents a floating point constant.
3588 <ul>
3589 <li><tt>double getValue() const</tt>: Returns the underlying value of
3590 this constant. </li>
3591 </ul>
3592 </li>
3593 <li>ConstantArray : This represents a constant array.
3594 <ul>
3595 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
3596 a vector of component constants that makeup this array. </li>
3597 </ul>
3598 </li>
3599 <li>ConstantStruct : This represents a constant struct.
3600 <ul>
3601 <li><tt>const std::vector&lt;Use&gt; &amp;getValues() const</tt>: Returns
3602 a vector of component constants that makeup this array. </li>
3603 </ul>
3604 </li>
3605 <li>GlobalValue : This represents either a global variable or a function. In
3606 either case, the value is a constant fixed address (after linking).
3607 </li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3608</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3609</div>
3610
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3611</div>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3612
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3613<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3614<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3615 <a name="GlobalValue">The <tt>GlobalValue</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3616</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3617
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3618<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3619
3620<p><tt>#include "<a
3621href="/doxygen/GlobalValue_8h-source.html">llvm/GlobalValue.h</a>"</tt><br>
Misha Brukmanfdcb3c22004-06-03 23:29:12 +0000[diff] [blame]3622doxygen info: <a href="/doxygen/classllvm_1_1GlobalValue.html">GlobalValue
3623Class</a><br>
Reid Spencer83ee73f2006-04-14 14:11:48 +0000[diff] [blame]3624Superclasses: <a href="#Constant"><tt>Constant</tt></a>,
3625<a href="#User"><tt>User</tt></a>, <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3626
3627<p>Global values (<a href="#GlobalVariable"><tt>GlobalVariable</tt></a>s or <a
3628href="#Function"><tt>Function</tt></a>s) are the only LLVM values that are
3629visible in the bodies of all <a href="#Function"><tt>Function</tt></a>s.
3630Because they are visible at global scope, they are also subject to linking with
3631other globals defined in different translation units. To control the linking
3632process, <tt>GlobalValue</tt>s know their linkage rules. Specifically,
3633<tt>GlobalValue</tt>s know whether they have internal or external linkage, as
Reid Spencerfa49f862004-07-18 13:10:31 +0000[diff] [blame]3634defined by the <tt>LinkageTypes</tt> enumeration.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3635
3636<p>If a <tt>GlobalValue</tt> has internal linkage (equivalent to being
3637<tt>static</tt> in C), it is not visible to code outside the current translation
3638unit, and does not participate in linking. If it has external linkage, it is
3639visible to external code, and does participate in linking. In addition to
3640linkage information, <tt>GlobalValue</tt>s keep track of which <a
3641href="#Module"><tt>Module</tt></a> they are currently part of.</p>
3642
3643<p>Because <tt>GlobalValue</tt>s are memory objects, they are always referred to
3644by their <b>address</b>. As such, the <a href="#Type"><tt>Type</tt></a> of a
3645global is always a pointer to its contents. It is important to remember this
3646when using the <tt>GetElementPtrInst</tt> instruction because this pointer must
3647be dereferenced first. For example, if you have a <tt>GlobalVariable</tt> (a
3648subclass of <tt>GlobalValue)</tt> that is an array of 24 ints, type <tt>[24 x
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]3649i32]</tt>, then the <tt>GlobalVariable</tt> is a pointer to that array. Although
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3650the address of the first element of this array and the value of the
3651<tt>GlobalVariable</tt> are the same, they have different types. The
Reid Spencerb9e5d102007-01-12 17:11:23 +0000[diff] [blame]3652<tt>GlobalVariable</tt>'s type is <tt>[24 x i32]</tt>. The first element's type
3653is <tt>i32.</tt> Because of this, accessing a global value requires you to
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3654dereference the pointer with <tt>GetElementPtrInst</tt> first, then its elements
3655can be accessed. This is explained in the <a href="LangRef.html#globalvars">LLVM
3656Language Reference Manual</a>.</p>
3657
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3658<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3659<h4>
3660 <a name="m_GlobalValue">
3661 Important Public Members of the <tt>GlobalValue</tt> class
3662 </a>
3663</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3664
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3665<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3666
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3667<ul>
3668 <li><tt>bool hasInternalLinkage() const</tt><br>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]3669 <tt>bool hasExternalLinkage() const</tt><br>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3670 <tt>void setInternalLinkage(bool HasInternalLinkage)</tt>
3671 <p> These methods manipulate the linkage characteristics of the <tt>GlobalValue</tt>.</p>
3672 <p> </p>
3673 </li>
3674 <li><tt><a href="#Module">Module</a> *getParent()</tt>
3675 <p> This returns the <a href="#Module"><tt>Module</tt></a> that the
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3676GlobalValue is currently embedded into.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3677</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3678
3679</div>
3680
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3681</div>
3682
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3683<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3684<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3685 <a name="Function">The <tt>Function</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3686</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3687
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3688<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3689
3690<p><tt>#include "<a
3691href="/doxygen/Function_8h-source.html">llvm/Function.h</a>"</tt><br> doxygen
Misha Brukman18a94d62004-06-03 23:35:54 +0000[diff] [blame]3692info: <a href="/doxygen/classllvm_1_1Function.html">Function Class</a><br>
Reid Spencer83ee73f2006-04-14 14:11:48 +0000[diff] [blame]3693Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3694<a href="#Constant"><tt>Constant</tt></a>,
3695<a href="#User"><tt>User</tt></a>,
3696<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3697
3698<p>The <tt>Function</tt> class represents a single procedure in LLVM. It is
Torok Edwin7630f102009-10-12 13:37:29 +0000[diff] [blame]3699actually one of the more complex classes in the LLVM hierarchy because it must
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3700keep track of a large amount of data. The <tt>Function</tt> class keeps track
Reid Spencer83ee73f2006-04-14 14:11:48 +0000[diff] [blame]3701of a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, a list of formal
3702<a href="#Argument"><tt>Argument</tt></a>s, and a
3703<a href="#SymbolTable"><tt>SymbolTable</tt></a>.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3704
3705<p>The list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s is the most
3706commonly used part of <tt>Function</tt> objects. The list imposes an implicit
3707ordering of the blocks in the function, which indicate how the code will be
Benjamin Kramer0f420382009-10-12 14:46:08 +0000[diff] [blame]3708laid out by the backend. Additionally, the first <a
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3709href="#BasicBlock"><tt>BasicBlock</tt></a> is the implicit entry node for the
3710<tt>Function</tt>. It is not legal in LLVM to explicitly branch to this initial
3711block. There are no implicit exit nodes, and in fact there may be multiple exit
3712nodes from a single <tt>Function</tt>. If the <a
3713href="#BasicBlock"><tt>BasicBlock</tt></a> list is empty, this indicates that
3714the <tt>Function</tt> is actually a function declaration: the actual body of the
3715function hasn't been linked in yet.</p>
3716
3717<p>In addition to a list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s, the
3718<tt>Function</tt> class also keeps track of the list of formal <a
3719href="#Argument"><tt>Argument</tt></a>s that the function receives. This
3720container manages the lifetime of the <a href="#Argument"><tt>Argument</tt></a>
3721nodes, just like the <a href="#BasicBlock"><tt>BasicBlock</tt></a> list does for
3722the <a href="#BasicBlock"><tt>BasicBlock</tt></a>s.</p>
3723
3724<p>The <a href="#SymbolTable"><tt>SymbolTable</tt></a> is a very rarely used
3725LLVM feature that is only used when you have to look up a value by name. Aside
3726from that, the <a href="#SymbolTable"><tt>SymbolTable</tt></a> is used
3727internally to make sure that there are not conflicts between the names of <a
3728href="#Instruction"><tt>Instruction</tt></a>s, <a
3729href="#BasicBlock"><tt>BasicBlock</tt></a>s, or <a
3730href="#Argument"><tt>Argument</tt></a>s in the function body.</p>
3731
Reid Spencerfa49f862004-07-18 13:10:31 +0000[diff] [blame]3732<p>Note that <tt>Function</tt> is a <a href="#GlobalValue">GlobalValue</a>
3733and therefore also a <a href="#Constant">Constant</a>. The value of the function
3734is its address (after linking) which is guaranteed to be constant.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3735
3736<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3737<h4>
3738 <a name="m_Function">
3739 Important Public Members of the <tt>Function</tt> class
3740 </a>
3741</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3742
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3743<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3744
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3745<ul>
3746 <li><tt>Function(const </tt><tt><a href="#FunctionType">FunctionType</a>
Chris Lattnere99bec72004-08-04 05:10:48 +0000[diff] [blame]3747 *Ty, LinkageTypes Linkage, const std::string &amp;N = "", Module* Parent = 0)</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3748
3749 <p>Constructor used when you need to create new <tt>Function</tt>s to add
3750 the the program. The constructor must specify the type of the function to
Chris Lattnere99bec72004-08-04 05:10:48 +0000[diff] [blame]3751 create and what type of linkage the function should have. The <a
3752 href="#FunctionType"><tt>FunctionType</tt></a> argument
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3753 specifies the formal arguments and return value for the function. The same
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]3754 <a href="#FunctionType"><tt>FunctionType</tt></a> value can be used to
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3755 create multiple functions. The <tt>Parent</tt> argument specifies the Module
3756 in which the function is defined. If this argument is provided, the function
3757 will automatically be inserted into that module's list of
3758 functions.</p></li>
3759
Chris Lattnerc09f2c22008-11-25 18:34:50 +0000[diff] [blame]3760 <li><tt>bool isDeclaration()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3761
3762 <p>Return whether or not the <tt>Function</tt> has a body defined. If the
3763 function is "external", it does not have a body, and thus must be resolved
3764 by linking with a function defined in a different translation unit.</p></li>
3765
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3766 <li><tt>Function::iterator</tt> - Typedef for basic block list iterator<br>
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]3767 <tt>Function::const_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3768
Chris Lattnerbc18ef02005-03-15 05:19:20 +0000[diff] [blame]3769 <tt>begin()</tt>, <tt>end()</tt>
3770 <tt>size()</tt>, <tt>empty()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3771
3772 <p>These are forwarding methods that make it easy to access the contents of
3773 a <tt>Function</tt> object's <a href="#BasicBlock"><tt>BasicBlock</tt></a>
3774 list.</p></li>
3775
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3776 <li><tt>Function::BasicBlockListType &amp;getBasicBlockList()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3777
3778 <p>Returns the list of <a href="#BasicBlock"><tt>BasicBlock</tt></a>s. This
3779 is necessary to use when you need to update the list or perform a complex
3780 action that doesn't have a forwarding method.</p></li>
3781
Chris Lattner5ce25cd2005-03-15 04:48:32 +0000[diff] [blame]3782 <li><tt>Function::arg_iterator</tt> - Typedef for the argument list
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3783iterator<br>
Chris Lattner5ce25cd2005-03-15 04:48:32 +0000[diff] [blame]3784 <tt>Function::const_arg_iterator</tt> - Typedef for const_iterator.<br>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3785
Chris Lattnerbc18ef02005-03-15 05:19:20 +0000[diff] [blame]3786 <tt>arg_begin()</tt>, <tt>arg_end()</tt>
Chris Lattner5ce25cd2005-03-15 04:48:32 +0000[diff] [blame]3787 <tt>arg_size()</tt>, <tt>arg_empty()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3788
3789 <p>These are forwarding methods that make it easy to access the contents of
3790 a <tt>Function</tt> object's <a href="#Argument"><tt>Argument</tt></a>
3791 list.</p></li>
3792
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3793 <li><tt>Function::ArgumentListType &amp;getArgumentList()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3794
3795 <p>Returns the list of <a href="#Argument"><tt>Argument</tt></a>s. This is
3796 necessary to use when you need to update the list or perform a complex
3797 action that doesn't have a forwarding method.</p></li>
3798
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3799 <li><tt><a href="#BasicBlock">BasicBlock</a> &amp;getEntryBlock()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3800
3801 <p>Returns the entry <a href="#BasicBlock"><tt>BasicBlock</tt></a> for the
3802 function. Because the entry block for the function is always the first
3803 block, this returns the first block of the <tt>Function</tt>.</p></li>
3804
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3805 <li><tt><a href="#Type">Type</a> *getReturnType()</tt><br>
3806 <tt><a href="#FunctionType">FunctionType</a> *getFunctionType()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3807
3808 <p>This traverses the <a href="#Type"><tt>Type</tt></a> of the
3809 <tt>Function</tt> and returns the return type of the function, or the <a
3810 href="#FunctionType"><tt>FunctionType</tt></a> of the actual
3811 function.</p></li>
3812
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3813 <li><tt><a href="#SymbolTable">SymbolTable</a> *getSymbolTable()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3814
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3815 <p> Return a pointer to the <a href="#SymbolTable"><tt>SymbolTable</tt></a>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3816 for this <tt>Function</tt>.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3817</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3818
3819</div>
3820
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3821</div>
3822
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3823<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3824<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3825 <a name="GlobalVariable">The <tt>GlobalVariable</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3826</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3827
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3828<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3829
3830<p><tt>#include "<a
3831href="/doxygen/GlobalVariable_8h-source.html">llvm/GlobalVariable.h</a>"</tt>
3832<br>
Tanya Lattnerf31fdf82004-06-22 08:02:25 +0000[diff] [blame]3833doxygen info: <a href="/doxygen/classllvm_1_1GlobalVariable.html">GlobalVariable
Reid Spencer83ee73f2006-04-14 14:11:48 +0000[diff] [blame]3834 Class</a><br>
3835Superclasses: <a href="#GlobalValue"><tt>GlobalValue</tt></a>,
3836<a href="#Constant"><tt>Constant</tt></a>,
3837<a href="#User"><tt>User</tt></a>,
3838<a href="#Value"><tt>Value</tt></a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3839
Benjamin Kramer0f420382009-10-12 14:46:08 +0000[diff] [blame]3840<p>Global variables are represented with the (surprise surprise)
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3841<tt>GlobalVariable</tt> class. Like functions, <tt>GlobalVariable</tt>s are also
3842subclasses of <a href="#GlobalValue"><tt>GlobalValue</tt></a>, and as such are
3843always referenced by their address (global values must live in memory, so their
Reid Spencer83ee73f2006-04-14 14:11:48 +0000[diff] [blame]3844"name" refers to their constant address). See
3845<a href="#GlobalValue"><tt>GlobalValue</tt></a> for more on this. Global
3846variables may have an initial value (which must be a
3847<a href="#Constant"><tt>Constant</tt></a>), and if they have an initializer,
3848they may be marked as "constant" themselves (indicating that their contents
3849never change at runtime).</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3850
3851<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3852<h4>
3853 <a name="m_GlobalVariable">
3854 Important Public Members of the <tt>GlobalVariable</tt> class
3855 </a>
3856</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3857
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3858<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3859
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3860<ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3861 <li><tt>GlobalVariable(const </tt><tt><a href="#Type">Type</a> *Ty, bool
3862 isConstant, LinkageTypes&amp; Linkage, <a href="#Constant">Constant</a>
3863 *Initializer = 0, const std::string &amp;Name = "", Module* Parent = 0)</tt>
3864
3865 <p>Create a new global variable of the specified type. If
3866 <tt>isConstant</tt> is true then the global variable will be marked as
3867 unchanging for the program. The Linkage parameter specifies the type of
Duncan Sands12da8ce2009-03-07 15:45:40 +0000[diff] [blame]3868 linkage (internal, external, weak, linkonce, appending) for the variable.
3869 If the linkage is InternalLinkage, WeakAnyLinkage, WeakODRLinkage,
3870 LinkOnceAnyLinkage or LinkOnceODRLinkage,&nbsp; then the resultant
3871 global variable will have internal linkage. AppendingLinkage concatenates
3872 together all instances (in different translation units) of the variable
3873 into a single variable but is only applicable to arrays. &nbsp;See
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3874 the <a href="LangRef.html#modulestructure">LLVM Language Reference</a> for
3875 further details on linkage types. Optionally an initializer, a name, and the
3876 module to put the variable into may be specified for the global variable as
3877 well.</p></li>
3878
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3879 <li><tt>bool isConstant() const</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3880
3881 <p>Returns true if this is a global variable that is known not to
3882 be modified at runtime.</p></li>
3883
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3884 <li><tt>bool hasInitializer()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3885
3886 <p>Returns true if this <tt>GlobalVariable</tt> has an intializer.</p></li>
3887
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3888 <li><tt><a href="#Constant">Constant</a> *getInitializer()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3889
Benjamin Kramer0f420382009-10-12 14:46:08 +0000[diff] [blame]3890 <p>Returns the initial value for a <tt>GlobalVariable</tt>. It is not legal
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3891 to call this method if there is no initializer.</p></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3892</ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3893
3894</div>
3895
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3896</div>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3897
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3898<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3899<h3>
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3900 <a name="BasicBlock">The <tt>BasicBlock</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3901</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3902
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3903<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3904
3905<p><tt>#include "<a
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3906href="/doxygen/BasicBlock_8h-source.html">llvm/BasicBlock.h</a>"</tt><br>
Stefanus Du Toit667cd9a2009-06-17 21:12:26 +0000[diff] [blame]3907doxygen info: <a href="/doxygen/classllvm_1_1BasicBlock.html">BasicBlock
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3908Class</a><br>
3909Superclass: <a href="#Value"><tt>Value</tt></a></p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3910
Nick Lewyckyb302f8d2011-02-17 02:19:22 +0000[diff] [blame]3911<p>This class represents a single entry single exit section of the code,
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3912commonly known as a basic block by the compiler community. The
3913<tt>BasicBlock</tt> class maintains a list of <a
3914href="#Instruction"><tt>Instruction</tt></a>s, which form the body of the block.
3915Matching the language definition, the last element of this list of instructions
3916is always a terminator instruction (a subclass of the <a
3917href="#TerminatorInst"><tt>TerminatorInst</tt></a> class).</p>
3918
3919<p>In addition to tracking the list of instructions that make up the block, the
3920<tt>BasicBlock</tt> class also keeps track of the <a
3921href="#Function"><tt>Function</tt></a> that it is embedded into.</p>
3922
3923<p>Note that <tt>BasicBlock</tt>s themselves are <a
3924href="#Value"><tt>Value</tt></a>s, because they are referenced by instructions
3925like branches and can go in the switch tables. <tt>BasicBlock</tt>s have type
3926<tt>label</tt>.</p>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3927
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3928<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3929<h4>
3930 <a name="m_BasicBlock">
3931 Important Public Members of the <tt>BasicBlock</tt> class
3932 </a>
3933</h4>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3934
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3935<div>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3936<ul>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3937
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3938<li><tt>BasicBlock(const std::string &amp;Name = "", </tt><tt><a
3939 href="#Function">Function</a> *Parent = 0)</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3940
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3941<p>The <tt>BasicBlock</tt> constructor is used to create new basic blocks for
3942insertion into a function. The constructor optionally takes a name for the new
3943block, and a <a href="#Function"><tt>Function</tt></a> to insert it into. If
3944the <tt>Parent</tt> parameter is specified, the new <tt>BasicBlock</tt> is
3945automatically inserted at the end of the specified <a
3946href="#Function"><tt>Function</tt></a>, if not specified, the BasicBlock must be
3947manually inserted into the <a href="#Function"><tt>Function</tt></a>.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3948
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3949<li><tt>BasicBlock::iterator</tt> - Typedef for instruction list iterator<br>
3950<tt>BasicBlock::const_iterator</tt> - Typedef for const_iterator.<br>
3951<tt>begin()</tt>, <tt>end()</tt>, <tt>front()</tt>, <tt>back()</tt>,
3952<tt>size()</tt>, <tt>empty()</tt>
3953STL-style functions for accessing the instruction list.
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3954
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3955<p>These methods and typedefs are forwarding functions that have the same
3956semantics as the standard library methods of the same names. These methods
3957expose the underlying instruction list of a basic block in a way that is easy to
3958manipulate. To get the full complement of container operations (including
3959operations to update the list), you must use the <tt>getInstList()</tt>
3960method.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3961
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3962<li><tt>BasicBlock::InstListType &amp;getInstList()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3963
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3964<p>This method is used to get access to the underlying container that actually
3965holds the Instructions. This method must be used when there isn't a forwarding
3966function in the <tt>BasicBlock</tt> class for the operation that you would like
3967to perform. Because there are no forwarding functions for "updating"
3968operations, you need to use this if you want to update the contents of a
3969<tt>BasicBlock</tt>.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3970
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3971<li><tt><a href="#Function">Function</a> *getParent()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3972
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3973<p> Returns a pointer to <a href="#Function"><tt>Function</tt></a> the block is
3974embedded into, or a null pointer if it is homeless.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3975
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3976<li><tt><a href="#TerminatorInst">TerminatorInst</a> *getTerminator()</tt>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3977
Chris Lattner9dd7a382007-02-03 20:02:25 +0000[diff] [blame]3978<p> Returns a pointer to the terminator instruction that appears at the end of
3979the <tt>BasicBlock</tt>. If there is no terminator instruction, or if the last
3980instruction in the block is not a terminator, then a null pointer is
3981returned.</p></li>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3982
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3983</ul>
3984
3985</div>
3986
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3987</div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3988
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3989<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3990<h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3991 <a name="Argument">The <tt>Argument</tt> class</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000[diff] [blame]3992</h3>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3993
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]3994<div>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3995
3996<p>This subclass of Value defines the interface for incoming formal
Chris Lattnerf7648e72005-01-17 00:12:04 +0000[diff] [blame]3997arguments to a function. A Function maintains a list of its formal
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]3998arguments. An argument has a pointer to the parent Function.</p>
3999
4000</div>
4001
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000[diff] [blame]4002</div>
4003
Chris Lattnerbcf337b2002-09-06 02:50:58 +0000[diff] [blame]4004<!-- *********************************************************************** -->
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]4005<hr>
4006<address>
4007 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +0000[diff] [blame]4008 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]4009 <a href="http://validator.w3.org/check/referer"><img
Gabor Greif687f35f2008-06-18 14:05:31 +0000[diff] [blame]4010 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01 Strict"></a>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]4011
4012 <a href="mailto:dhurjati@cs.uiuc.edu">Dinakar Dhurjati</a> and
4013 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
NAKAMURA Takumica46f5a2011-04-09 02:13:37 +0000[diff] [blame]4014 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Misha Brukman7f67e372004-01-15 00:14:41 +0000[diff] [blame]4015 Last modified: $Date$
4016</address>
4017
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4018</body>
4019</html>