blob: de19222c52b16f9c501967d08435a2fc83cd387f [file] [log] [blame]
Misha Brukmanc501f552004-03-01 17:47:27 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman76307852003-11-08 01:05:38 +00003<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencercb84e432004-08-26 20:44:00 +00006 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
Misha Brukman76307852003-11-08 01:05:38 +000010 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattner757528b0b2004-05-23 21:06:01 +000012
Misha Brukman76307852003-11-08 01:05:38 +000013<body>
Chris Lattner757528b0b2004-05-23 21:06:01 +000014
Chris Lattner48b383b02003-11-25 01:02:51 +000015<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner2f7c9632001-06-06 20:29:01 +000016<ol>
Misha Brukman76307852003-11-08 01:05:38 +000017 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000020 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Chris Lattnerd79749a2004-12-09 16:36:40 +000023 <li><a href="#linkage">Linkage Types</a></li>
Chris Lattner0132aff2005-05-06 22:57:40 +000024 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000025 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +000026 <li><a href="#functionstructure">Functions</a></li>
27 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000028 </ol>
29 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000030 <li><a href="#typesystem">Type System</a>
31 <ol>
Robert Bocchino820bc75b2006-02-17 21:18:08 +000032 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner48b383b02003-11-25 01:02:51 +000033 <ol>
Misha Brukman76307852003-11-08 01:05:38 +000034 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000035 </ol>
36 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000037 <li><a href="#t_derived">Derived Types</a>
38 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +000039 <li><a href="#t_array">Array Type</a></li>
Misha Brukman76307852003-11-08 01:05:38 +000040 <li><a href="#t_function">Function Type</a></li>
41 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000042 <li><a href="#t_struct">Structure Type</a></li>
Chris Lattnerc8cb6952004-08-12 19:12:28 +000043 <li><a href="#t_packed">Packed Type</a></li>
Chris Lattner37b6b092005-04-25 17:34:15 +000044 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000045 </ol>
46 </li>
47 </ol>
48 </li>
Chris Lattner6af02f32004-12-09 16:11:40 +000049 <li><a href="#constants">Constants</a>
Chris Lattner74d3f822004-12-09 17:30:23 +000050 <ol>
51 <li><a href="#simpleconstants">Simple Constants</a>
52 <li><a href="#aggregateconstants">Aggregate Constants</a>
53 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
54 <li><a href="#undefvalues">Undefined Values</a>
55 <li><a href="#constantexprs">Constant Expressions</a>
56 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +000057 </li>
Chris Lattner98f013c2006-01-25 23:47:57 +000058 <li><a href="#othervalues">Other Values</a>
59 <ol>
60 <li><a href="#inlineasm">Inline Assembler Expressions</a>
61 </ol>
62 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000063 <li><a href="#instref">Instruction Reference</a>
64 <ol>
65 <li><a href="#terminators">Terminator Instructions</a>
66 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +000067 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
68 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +000069 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
70 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000071 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner08b7d5b2004-10-16 18:04:13 +000072 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000073 </ol>
74 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000075 <li><a href="#binaryops">Binary Operations</a>
76 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +000077 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
78 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
79 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
80 <li><a href="#i_div">'<tt>div</tt>' Instruction</a></li>
81 <li><a href="#i_rem">'<tt>rem</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +000082 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000083 </ol>
84 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000085 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
86 <ol>
Misha Brukman76307852003-11-08 01:05:38 +000087 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000088 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +000089 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
90 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
91 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000092 </ol>
93 </li>
Chris Lattnerce83bff2006-04-08 23:07:04 +000094 <li><a href="#vectorops">Vector Operations</a>
95 <ol>
96 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
97 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
98 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Tanya Lattnerb138bbe2006-04-14 19:24:33 +000099 <li><a href="#i_vsetint">'<tt>vsetint</tt>' Instruction</a></li>
100 <li><a href="#i_vsetfp">'<tt>vsetfp</tt>' Instruction</a></li>
101 <li><a href="#i_vselect">'<tt>vselect</tt>' Instruction</a></li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000102 </ol>
103 </li>
Chris Lattner6ab66722006-08-15 00:45:58 +0000104 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000105 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000106 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
107 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
108 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino820bc75b2006-02-17 21:18:08 +0000109 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
110 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
111 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000112 </ol>
113 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000114 <li><a href="#otherops">Other Operations</a>
115 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000116 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000117 <li><a href="#i_cast">'<tt>cast .. to</tt>' Instruction</a></li>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000118 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000119 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattner33337472006-01-13 23:26:01 +0000120 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000121 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000122 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000123 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000124 </li>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000125 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000126 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000127 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
128 <ol>
129 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
130 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
131 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
132 </ol>
133 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000134 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
135 <ol>
136 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
137 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
138 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
139 </ol>
140 </li>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000141 <li><a href="#int_codegen">Code Generator Intrinsics</a>
142 <ol>
143 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
144 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
Chris Lattner2f0f0012006-01-13 02:03:13 +0000145 <li><a href="#i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
146 <li><a href="#i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
Chris Lattnerc8a2c222005-02-28 19:24:19 +0000147 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
Andrew Lenharthb4427912005-03-28 20:05:49 +0000148 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Andrew Lenharth01aa5632005-11-11 16:47:30 +0000149 <li><a href="#i_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswellaa1c3c12004-04-09 16:43:20 +0000150 </ol>
151 </li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000152 <li><a href="#int_libc">Standard C Library Intrinsics</a>
153 <ol>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000154 <li><a href="#i_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
155 <li><a href="#i_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
156 <li><a href="#i_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
Chris Lattner069b5bd2006-01-16 22:38:59 +0000157 <li><a href="#i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a></li>
158 <li><a href="#i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
Chris Lattner33b73f92006-09-08 06:34:02 +0000159 <li><a href="#i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000160 </ol>
161 </li>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000162 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000163 <ol>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000164 <li><a href="#i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattnerb748c672006-01-16 22:34:14 +0000165 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
166 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
167 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000168 </ol>
169 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000170 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000171 </ol>
172 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000173</ol>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000174
175<div class="doc_author">
176 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
177 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman76307852003-11-08 01:05:38 +0000178</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000179
Chris Lattner2f7c9632001-06-06 20:29:01 +0000180<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000181<div class="doc_section"> <a name="abstract">Abstract </a></div>
182<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000183
Misha Brukman76307852003-11-08 01:05:38 +0000184<div class="doc_text">
Chris Lattner48b383b02003-11-25 01:02:51 +0000185<p>This document is a reference manual for the LLVM assembly language.
186LLVM is an SSA based representation that provides type safety,
187low-level operations, flexibility, and the capability of representing
188'all' high-level languages cleanly. It is the common code
189representation used throughout all phases of the LLVM compilation
190strategy.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000191</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000192
Chris Lattner2f7c9632001-06-06 20:29:01 +0000193<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000194<div class="doc_section"> <a name="introduction">Introduction</a> </div>
195<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000196
Misha Brukman76307852003-11-08 01:05:38 +0000197<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000198
Chris Lattner48b383b02003-11-25 01:02:51 +0000199<p>The LLVM code representation is designed to be used in three
200different forms: as an in-memory compiler IR, as an on-disk bytecode
201representation (suitable for fast loading by a Just-In-Time compiler),
202and as a human readable assembly language representation. This allows
203LLVM to provide a powerful intermediate representation for efficient
204compiler transformations and analysis, while providing a natural means
205to debug and visualize the transformations. The three different forms
206of LLVM are all equivalent. This document describes the human readable
207representation and notation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000208
John Criswell4a3327e2005-05-13 22:25:59 +0000209<p>The LLVM representation aims to be light-weight and low-level
Chris Lattner48b383b02003-11-25 01:02:51 +0000210while being expressive, typed, and extensible at the same time. It
211aims to be a "universal IR" of sorts, by being at a low enough level
212that high-level ideas may be cleanly mapped to it (similar to how
213microprocessors are "universal IR's", allowing many source languages to
214be mapped to them). By providing type information, LLVM can be used as
215the target of optimizations: for example, through pointer analysis, it
216can be proven that a C automatic variable is never accessed outside of
217the current function... allowing it to be promoted to a simple SSA
218value instead of a memory location.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000219
Misha Brukman76307852003-11-08 01:05:38 +0000220</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000221
Chris Lattner2f7c9632001-06-06 20:29:01 +0000222<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000223<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000224
Misha Brukman76307852003-11-08 01:05:38 +0000225<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000226
Chris Lattner48b383b02003-11-25 01:02:51 +0000227<p>It is important to note that this document describes 'well formed'
228LLVM assembly language. There is a difference between what the parser
229accepts and what is considered 'well formed'. For example, the
230following instruction is syntactically okay, but not well formed:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000231
232<pre>
233 %x = <a href="#i_add">add</a> int 1, %x
234</pre>
235
Chris Lattner48b383b02003-11-25 01:02:51 +0000236<p>...because the definition of <tt>%x</tt> does not dominate all of
237its uses. The LLVM infrastructure provides a verification pass that may
238be used to verify that an LLVM module is well formed. This pass is
John Criswell4a3327e2005-05-13 22:25:59 +0000239automatically run by the parser after parsing input assembly and by
Chris Lattner48b383b02003-11-25 01:02:51 +0000240the optimizer before it outputs bytecode. The violations pointed out
241by the verifier pass indicate bugs in transformation passes or input to
242the parser.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000243
Chris Lattner48b383b02003-11-25 01:02:51 +0000244<!-- Describe the typesetting conventions here. --> </div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000245
Chris Lattner2f7c9632001-06-06 20:29:01 +0000246<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000247<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000248<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000249
Misha Brukman76307852003-11-08 01:05:38 +0000250<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000251
Chris Lattner48b383b02003-11-25 01:02:51 +0000252<p>LLVM uses three different forms of identifiers, for different
253purposes:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000254
Chris Lattner2f7c9632001-06-06 20:29:01 +0000255<ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000256 <li>Named values are represented as a string of characters with a '%' prefix.
257 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
258 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
259 Identifiers which require other characters in their names can be surrounded
260 with quotes. In this way, anything except a <tt>"</tt> character can be used
261 in a name.</li>
262
263 <li>Unnamed values are represented as an unsigned numeric value with a '%'
264 prefix. For example, %12, %2, %44.</li>
265
Reid Spencer8f08d802004-12-09 18:02:53 +0000266 <li>Constants, which are described in a <a href="#constants">section about
267 constants</a>, below.</li>
Misha Brukman76307852003-11-08 01:05:38 +0000268</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000269
270<p>LLVM requires that values start with a '%' sign for two reasons: Compilers
271don't need to worry about name clashes with reserved words, and the set of
272reserved words may be expanded in the future without penalty. Additionally,
273unnamed identifiers allow a compiler to quickly come up with a temporary
274variable without having to avoid symbol table conflicts.</p>
275
Chris Lattner48b383b02003-11-25 01:02:51 +0000276<p>Reserved words in LLVM are very similar to reserved words in other
277languages. There are keywords for different opcodes ('<tt><a
Chris Lattnerd79749a2004-12-09 16:36:40 +0000278href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
279href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
280href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>', etc...),
281and others. These reserved words cannot conflict with variable names, because
282none of them start with a '%' character.</p>
283
284<p>Here is an example of LLVM code to multiply the integer variable
285'<tt>%X</tt>' by 8:</p>
286
Misha Brukman76307852003-11-08 01:05:38 +0000287<p>The easy way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000288
289<pre>
290 %result = <a href="#i_mul">mul</a> uint %X, 8
291</pre>
292
Misha Brukman76307852003-11-08 01:05:38 +0000293<p>After strength reduction:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000294
295<pre>
296 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
297</pre>
298
Misha Brukman76307852003-11-08 01:05:38 +0000299<p>And the hard way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000300
301<pre>
302 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
303 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
304 %result = <a href="#i_add">add</a> uint %1, %1
305</pre>
306
Chris Lattner48b383b02003-11-25 01:02:51 +0000307<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
308important lexical features of LLVM:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000309
Chris Lattner2f7c9632001-06-06 20:29:01 +0000310<ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000311
312 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
313 line.</li>
314
315 <li>Unnamed temporaries are created when the result of a computation is not
316 assigned to a named value.</li>
317
Misha Brukman76307852003-11-08 01:05:38 +0000318 <li>Unnamed temporaries are numbered sequentially</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000319
Misha Brukman76307852003-11-08 01:05:38 +0000320</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000321
John Criswell02fdc6f2005-05-12 16:52:32 +0000322<p>...and it also shows a convention that we follow in this document. When
Chris Lattnerd79749a2004-12-09 16:36:40 +0000323demonstrating instructions, we will follow an instruction with a comment that
324defines the type and name of value produced. Comments are shown in italic
325text.</p>
326
Misha Brukman76307852003-11-08 01:05:38 +0000327</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000328
329<!-- *********************************************************************** -->
330<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
331<!-- *********************************************************************** -->
332
333<!-- ======================================================================= -->
334<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
335</div>
336
337<div class="doc_text">
338
339<p>LLVM programs are composed of "Module"s, each of which is a
340translation unit of the input programs. Each module consists of
341functions, global variables, and symbol table entries. Modules may be
342combined together with the LLVM linker, which merges function (and
343global variable) definitions, resolves forward declarations, and merges
344symbol table entries. Here is an example of the "hello world" module:</p>
345
346<pre><i>; Declare the string constant as a global constant...</i>
347<a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
348 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
349
350<i>; External declaration of the puts function</i>
351<a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
352
Chris Lattnerd2d29a02006-06-13 03:05:47 +0000353<i>; Global variable / Function body section separator</i>
354implementation
355
Chris Lattner6af02f32004-12-09 16:11:40 +0000356<i>; Definition of main function</i>
357int %main() { <i>; int()* </i>
358 <i>; Convert [13x sbyte]* to sbyte *...</i>
359 %cast210 = <a
360 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
361
362 <i>; Call puts function to write out the string to stdout...</i>
363 <a
364 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
365 <a
366 href="#i_ret">ret</a> int 0<br>}<br></pre>
367
368<p>This example is made up of a <a href="#globalvars">global variable</a>
369named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
370function, and a <a href="#functionstructure">function definition</a>
371for "<tt>main</tt>".</p>
372
Chris Lattnerd79749a2004-12-09 16:36:40 +0000373<p>In general, a module is made up of a list of global values,
374where both functions and global variables are global values. Global values are
375represented by a pointer to a memory location (in this case, a pointer to an
376array of char, and a pointer to a function), and have one of the following <a
377href="#linkage">linkage types</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000378
Chris Lattnerd2d29a02006-06-13 03:05:47 +0000379<p>Due to a limitation in the current LLVM assembly parser (it is limited by
380one-token lookahead), modules are split into two pieces by the "implementation"
381keyword. Global variable prototypes and definitions must occur before the
382keyword, and function definitions must occur after it. Function prototypes may
383occur either before or after it. In the future, the implementation keyword may
384become a noop, if the parser gets smarter.</p>
385
Chris Lattnerd79749a2004-12-09 16:36:40 +0000386</div>
387
388<!-- ======================================================================= -->
389<div class="doc_subsection">
390 <a name="linkage">Linkage Types</a>
391</div>
392
393<div class="doc_text">
394
395<p>
396All Global Variables and Functions have one of the following types of linkage:
397</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000398
399<dl>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000400
Chris Lattner6af02f32004-12-09 16:11:40 +0000401 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000402
403 <dd>Global values with internal linkage are only directly accessible by
404 objects in the current module. In particular, linking code into a module with
405 an internal global value may cause the internal to be renamed as necessary to
406 avoid collisions. Because the symbol is internal to the module, all
407 references can be updated. This corresponds to the notion of the
408 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
Chris Lattner6af02f32004-12-09 16:11:40 +0000409 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000410
Chris Lattner6af02f32004-12-09 16:11:40 +0000411 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000412
413 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
414 the twist that linking together two modules defining the same
415 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
416 is typically used to implement inline functions. Unreferenced
417 <tt>linkonce</tt> globals are allowed to be discarded.
Chris Lattner6af02f32004-12-09 16:11:40 +0000418 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000419
Chris Lattner6af02f32004-12-09 16:11:40 +0000420 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000421
422 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
423 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
424 used to implement constructs in C such as "<tt>int X;</tt>" at global scope.
Chris Lattner6af02f32004-12-09 16:11:40 +0000425 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000426
Chris Lattner6af02f32004-12-09 16:11:40 +0000427 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000428
429 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
430 pointer to array type. When two global variables with appending linkage are
431 linked together, the two global arrays are appended together. This is the
432 LLVM, typesafe, equivalent of having the system linker append together
433 "sections" with identical names when .o files are linked.
Chris Lattner6af02f32004-12-09 16:11:40 +0000434 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000435
Chris Lattner6af02f32004-12-09 16:11:40 +0000436 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000437
438 <dd>If none of the above identifiers are used, the global is externally
439 visible, meaning that it participates in linkage and can be used to resolve
440 external symbol references.
Chris Lattner6af02f32004-12-09 16:11:40 +0000441 </dd>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000442
443 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
444
445 <dd>"<tt>extern_weak</tt>" TBD
446 </dd>
447
448 <p>
449 The next two types of linkage are targeted for Microsoft Windows platform
450 only. They are designed to support importing (exporting) symbols from (to)
451 DLLs.
452 </p>
453
454 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
455
456 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
457 or variable via a global pointer to a pointer that is set up by the DLL
458 exporting the symbol. On Microsoft Windows targets, the pointer name is
459 formed by combining <code>_imp__</code> and the function or variable name.
460 </dd>
461
462 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
463
464 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
465 pointer to a pointer in a DLL, so that it can be referenced with the
466 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
467 name is formed by combining <code>_imp__</code> and the function or variable
468 name.
469 </dd>
470
Chris Lattner6af02f32004-12-09 16:11:40 +0000471</dl>
472
Chris Lattner6af02f32004-12-09 16:11:40 +0000473<p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
474variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
475variable and was linked with this one, one of the two would be renamed,
476preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
477external (i.e., lacking any linkage declarations), they are accessible
478outside of the current module. It is illegal for a function <i>declaration</i>
479to have any linkage type other than "externally visible".</a></p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000480
Chris Lattner6af02f32004-12-09 16:11:40 +0000481</div>
482
483<!-- ======================================================================= -->
484<div class="doc_subsection">
Chris Lattner0132aff2005-05-06 22:57:40 +0000485 <a name="callingconv">Calling Conventions</a>
486</div>
487
488<div class="doc_text">
489
490<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
491and <a href="#i_invoke">invokes</a> can all have an optional calling convention
492specified for the call. The calling convention of any pair of dynamic
493caller/callee must match, or the behavior of the program is undefined. The
494following calling conventions are supported by LLVM, and more may be added in
495the future:</p>
496
497<dl>
498 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
499
500 <dd>This calling convention (the default if no other calling convention is
501 specified) matches the target C calling conventions. This calling convention
John Criswell02fdc6f2005-05-12 16:52:32 +0000502 supports varargs function calls and tolerates some mismatch in the declared
Chris Lattner0132aff2005-05-06 22:57:40 +0000503 prototype and implemented declaration of the function (as does normal C).
504 </dd>
505
Chris Lattner95ff1952006-05-19 21:15:36 +0000506 <dt><b>"<tt>csretcc</tt>" - The C struct return calling convention</b>:</dt>
507
508 <dd>This calling convention matches the target C calling conventions, except
509 that functions with this convention are required to take a pointer as their
510 first argument, and the return type of the function must be void. This is
511 used for C functions that return aggregates by-value. In this case, the
512 function has been transformed to take a pointer to the struct as the first
513 argument to the function. For targets where the ABI specifies specific
514 behavior for structure-return calls, the calling convention can be used to
515 distinguish between struct return functions and other functions that take a
516 pointer to a struct as the first argument.
517 </dd>
518
Chris Lattner0132aff2005-05-06 22:57:40 +0000519 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
520
521 <dd>This calling convention attempts to make calls as fast as possible
522 (e.g. by passing things in registers). This calling convention allows the
523 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattnerc792eb32005-05-06 23:08:23 +0000524 without having to conform to an externally specified ABI. Implementations of
525 this convention should allow arbitrary tail call optimization to be supported.
526 This calling convention does not support varargs and requires the prototype of
527 all callees to exactly match the prototype of the function definition.
Chris Lattner0132aff2005-05-06 22:57:40 +0000528 </dd>
529
530 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
531
532 <dd>This calling convention attempts to make code in the caller as efficient
533 as possible under the assumption that the call is not commonly executed. As
534 such, these calls often preserve all registers so that the call does not break
535 any live ranges in the caller side. This calling convention does not support
536 varargs and requires the prototype of all callees to exactly match the
537 prototype of the function definition.
538 </dd>
539
Chris Lattner573f64e2005-05-07 01:46:40 +0000540 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000541
542 <dd>Any calling convention may be specified by number, allowing
543 target-specific calling conventions to be used. Target specific calling
544 conventions start at 64.
545 </dd>
Chris Lattner573f64e2005-05-07 01:46:40 +0000546</dl>
Chris Lattner0132aff2005-05-06 22:57:40 +0000547
548<p>More calling conventions can be added/defined on an as-needed basis, to
549support pascal conventions or any other well-known target-independent
550convention.</p>
551
552</div>
553
554<!-- ======================================================================= -->
555<div class="doc_subsection">
Chris Lattner6af02f32004-12-09 16:11:40 +0000556 <a name="globalvars">Global Variables</a>
557</div>
558
559<div class="doc_text">
560
Chris Lattner5d5aede2005-02-12 19:30:21 +0000561<p>Global variables define regions of memory allocated at compilation time
Chris Lattner662c8722005-11-12 00:45:07 +0000562instead of run-time. Global variables may optionally be initialized, may have
563an explicit section to be placed in, and may
Chris Lattner54611b42005-11-06 08:02:57 +0000564have an optional explicit alignment specified. A
John Criswell4c0cf7f2005-10-24 16:17:18 +0000565variable may be defined as a global "constant," which indicates that the
Chris Lattner5d5aede2005-02-12 19:30:21 +0000566contents of the variable will <b>never</b> be modified (enabling better
567optimization, allowing the global data to be placed in the read-only section of
568an executable, etc). Note that variables that need runtime initialization
John Criswell4c0cf7f2005-10-24 16:17:18 +0000569cannot be marked "constant" as there is a store to the variable.</p>
Chris Lattner5d5aede2005-02-12 19:30:21 +0000570
571<p>
572LLVM explicitly allows <em>declarations</em> of global variables to be marked
573constant, even if the final definition of the global is not. This capability
574can be used to enable slightly better optimization of the program, but requires
575the language definition to guarantee that optimizations based on the
576'constantness' are valid for the translation units that do not include the
577definition.
578</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000579
580<p>As SSA values, global variables define pointer values that are in
581scope (i.e. they dominate) all basic blocks in the program. Global
582variables always define a pointer to their "content" type because they
583describe a region of memory, and all memory objects in LLVM are
584accessed through pointers.</p>
585
Chris Lattner662c8722005-11-12 00:45:07 +0000586<p>LLVM allows an explicit section to be specified for globals. If the target
587supports it, it will emit globals to the section specified.</p>
588
Chris Lattner54611b42005-11-06 08:02:57 +0000589<p>An explicit alignment may be specified for a global. If not present, or if
590the alignment is set to zero, the alignment of the global is set by the target
591to whatever it feels convenient. If an explicit alignment is specified, the
592global is forced to have at least that much alignment. All alignments must be
593a power of 2.</p>
594
Chris Lattner6af02f32004-12-09 16:11:40 +0000595</div>
596
597
598<!-- ======================================================================= -->
599<div class="doc_subsection">
600 <a name="functionstructure">Functions</a>
601</div>
602
603<div class="doc_text">
604
Chris Lattner0132aff2005-05-06 22:57:40 +0000605<p>LLVM function definitions consist of an optional <a href="#linkage">linkage
606type</a>, an optional <a href="#callingconv">calling convention</a>, a return
Chris Lattner662c8722005-11-12 00:45:07 +0000607type, a function name, a (possibly empty) argument list, an optional section,
608an optional alignment, an opening curly brace,
Chris Lattner0132aff2005-05-06 22:57:40 +0000609a list of basic blocks, and a closing curly brace. LLVM function declarations
610are defined with the "<tt>declare</tt>" keyword, an optional <a
Chris Lattner54611b42005-11-06 08:02:57 +0000611href="#callingconv">calling convention</a>, a return type, a function name,
612a possibly empty list of arguments, and an optional alignment.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000613
614<p>A function definition contains a list of basic blocks, forming the CFG for
615the function. Each basic block may optionally start with a label (giving the
616basic block a symbol table entry), contains a list of instructions, and ends
617with a <a href="#terminators">terminator</a> instruction (such as a branch or
618function return).</p>
619
John Criswell02fdc6f2005-05-12 16:52:32 +0000620<p>The first basic block in a program is special in two ways: it is immediately
Chris Lattner6af02f32004-12-09 16:11:40 +0000621executed on entrance to the function, and it is not allowed to have predecessor
622basic blocks (i.e. there can not be any branches to the entry block of a
623function). Because the block can have no predecessors, it also cannot have any
624<a href="#i_phi">PHI nodes</a>.</p>
625
626<p>LLVM functions are identified by their name and type signature. Hence, two
627functions with the same name but different parameter lists or return values are
Chris Lattner455fc8c2005-03-07 22:13:59 +0000628considered different functions, and LLVM will resolve references to each
Chris Lattner6af02f32004-12-09 16:11:40 +0000629appropriately.</p>
630
Chris Lattner662c8722005-11-12 00:45:07 +0000631<p>LLVM allows an explicit section to be specified for functions. If the target
632supports it, it will emit functions to the section specified.</p>
633
Chris Lattner54611b42005-11-06 08:02:57 +0000634<p>An explicit alignment may be specified for a function. If not present, or if
635the alignment is set to zero, the alignment of the function is set by the target
636to whatever it feels convenient. If an explicit alignment is specified, the
637function is forced to have at least that much alignment. All alignments must be
638a power of 2.</p>
639
Chris Lattner6af02f32004-12-09 16:11:40 +0000640</div>
641
Chris Lattner91c15c42006-01-23 23:23:47 +0000642<!-- ======================================================================= -->
643<div class="doc_subsection">
Chris Lattner93564892006-04-08 04:40:53 +0000644 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner91c15c42006-01-23 23:23:47 +0000645</div>
646
647<div class="doc_text">
648<p>
649Modules may contain "module-level inline asm" blocks, which corresponds to the
650GCC "file scope inline asm" blocks. These blocks are internally concatenated by
651LLVM and treated as a single unit, but may be separated in the .ll file if
652desired. The syntax is very simple:
653</p>
654
655<div class="doc_code"><pre>
Chris Lattnera1280ad2006-01-24 00:37:20 +0000656 module asm "inline asm code goes here"
657 module asm "more can go here"
Chris Lattner91c15c42006-01-23 23:23:47 +0000658</pre></div>
659
660<p>The strings can contain any character by escaping non-printable characters.
661 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
662 for the number.
663</p>
664
665<p>
666 The inline asm code is simply printed to the machine code .s file when
667 assembly code is generated.
668</p>
669</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000670
671
Chris Lattner2f7c9632001-06-06 20:29:01 +0000672<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000673<div class="doc_section"> <a name="typesystem">Type System</a> </div>
674<!-- *********************************************************************** -->
Chris Lattner6af02f32004-12-09 16:11:40 +0000675
Misha Brukman76307852003-11-08 01:05:38 +0000676<div class="doc_text">
Chris Lattner6af02f32004-12-09 16:11:40 +0000677
Misha Brukman76307852003-11-08 01:05:38 +0000678<p>The LLVM type system is one of the most important features of the
Chris Lattner48b383b02003-11-25 01:02:51 +0000679intermediate representation. Being typed enables a number of
680optimizations to be performed on the IR directly, without having to do
681extra analyses on the side before the transformation. A strong type
682system makes it easier to read the generated code and enables novel
683analyses and transformations that are not feasible to perform on normal
684three address code representations.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000685
686</div>
687
Chris Lattner2f7c9632001-06-06 20:29:01 +0000688<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000689<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000690<div class="doc_text">
John Criswell417228d2004-04-09 16:48:45 +0000691<p>The primitive types are the fundamental building blocks of the LLVM
Chris Lattner455fc8c2005-03-07 22:13:59 +0000692system. The current set of primitive types is as follows:</p>
Misha Brukmanc501f552004-03-01 17:47:27 +0000693
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000694<table class="layout">
695 <tr class="layout">
696 <td class="left">
697 <table>
Chris Lattner48b383b02003-11-25 01:02:51 +0000698 <tbody>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000699 <tr><th>Type</th><th>Description</th></tr>
700 <tr><td><tt>void</tt></td><td>No value</td></tr>
Misha Brukman36c6bc12005-04-22 18:02:52 +0000701 <tr><td><tt>ubyte</tt></td><td>Unsigned 8-bit value</td></tr>
702 <tr><td><tt>ushort</tt></td><td>Unsigned 16-bit value</td></tr>
703 <tr><td><tt>uint</tt></td><td>Unsigned 32-bit value</td></tr>
704 <tr><td><tt>ulong</tt></td><td>Unsigned 64-bit value</td></tr>
705 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000706 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
Chris Lattner48b383b02003-11-25 01:02:51 +0000707 </tbody>
708 </table>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000709 </td>
710 <td class="right">
711 <table>
Chris Lattner48b383b02003-11-25 01:02:51 +0000712 <tbody>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000713 <tr><th>Type</th><th>Description</th></tr>
714 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
Misha Brukman36c6bc12005-04-22 18:02:52 +0000715 <tr><td><tt>sbyte</tt></td><td>Signed 8-bit value</td></tr>
716 <tr><td><tt>short</tt></td><td>Signed 16-bit value</td></tr>
717 <tr><td><tt>int</tt></td><td>Signed 32-bit value</td></tr>
718 <tr><td><tt>long</tt></td><td>Signed 64-bit value</td></tr>
719 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
Chris Lattner48b383b02003-11-25 01:02:51 +0000720 </tbody>
721 </table>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000722 </td>
723 </tr>
Misha Brukman76307852003-11-08 01:05:38 +0000724</table>
Misha Brukman76307852003-11-08 01:05:38 +0000725</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000726
Chris Lattner2f7c9632001-06-06 20:29:01 +0000727<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000728<div class="doc_subsubsection"> <a name="t_classifications">Type
729Classifications</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000730<div class="doc_text">
Chris Lattner48b383b02003-11-25 01:02:51 +0000731<p>These different primitive types fall into a few useful
732classifications:</p>
Misha Brukmanc501f552004-03-01 17:47:27 +0000733
734<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +0000735 <tbody>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000736 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner48b383b02003-11-25 01:02:51 +0000737 <tr>
738 <td><a name="t_signed">signed</a></td>
739 <td><tt>sbyte, short, int, long, float, double</tt></td>
740 </tr>
741 <tr>
742 <td><a name="t_unsigned">unsigned</a></td>
743 <td><tt>ubyte, ushort, uint, ulong</tt></td>
744 </tr>
745 <tr>
746 <td><a name="t_integer">integer</a></td>
747 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
748 </tr>
749 <tr>
750 <td><a name="t_integral">integral</a></td>
Misha Brukman20f9a622004-08-12 20:16:08 +0000751 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt>
752 </td>
Chris Lattner48b383b02003-11-25 01:02:51 +0000753 </tr>
754 <tr>
755 <td><a name="t_floating">floating point</a></td>
756 <td><tt>float, double</tt></td>
757 </tr>
758 <tr>
759 <td><a name="t_firstclass">first class</a></td>
Misha Brukman20f9a622004-08-12 20:16:08 +0000760 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long,<br>
761 float, double, <a href="#t_pointer">pointer</a>,
762 <a href="#t_packed">packed</a></tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +0000763 </tr>
764 </tbody>
Misha Brukman76307852003-11-08 01:05:38 +0000765</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000766
Chris Lattner48b383b02003-11-25 01:02:51 +0000767<p>The <a href="#t_firstclass">first class</a> types are perhaps the
768most important. Values of these types are the only ones which can be
769produced by instructions, passed as arguments, or used as operands to
770instructions. This means that all structures and arrays must be
771manipulated either by pointer or by component.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000772</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000773
Chris Lattner2f7c9632001-06-06 20:29:01 +0000774<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000775<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000776
Misha Brukman76307852003-11-08 01:05:38 +0000777<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +0000778
Chris Lattner48b383b02003-11-25 01:02:51 +0000779<p>The real power in LLVM comes from the derived types in the system.
780This is what allows a programmer to represent arrays, functions,
781pointers, and other useful types. Note that these derived types may be
782recursive: For example, it is possible to have a two dimensional array.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000783
Misha Brukman76307852003-11-08 01:05:38 +0000784</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000785
Chris Lattner2f7c9632001-06-06 20:29:01 +0000786<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000787<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000788
Misha Brukman76307852003-11-08 01:05:38 +0000789<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +0000790
Chris Lattner2f7c9632001-06-06 20:29:01 +0000791<h5>Overview:</h5>
Chris Lattner74d3f822004-12-09 17:30:23 +0000792
Misha Brukman76307852003-11-08 01:05:38 +0000793<p>The array type is a very simple derived type that arranges elements
Chris Lattner48b383b02003-11-25 01:02:51 +0000794sequentially in memory. The array type requires a size (number of
795elements) and an underlying data type.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000796
Chris Lattner590645f2002-04-14 06:13:44 +0000797<h5>Syntax:</h5>
Chris Lattner74d3f822004-12-09 17:30:23 +0000798
799<pre>
800 [&lt;# elements&gt; x &lt;elementtype&gt;]
801</pre>
802
John Criswell02fdc6f2005-05-12 16:52:32 +0000803<p>The number of elements is a constant integer value; elementtype may
Chris Lattner48b383b02003-11-25 01:02:51 +0000804be any type with a size.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000805
Chris Lattner590645f2002-04-14 06:13:44 +0000806<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000807<table class="layout">
808 <tr class="layout">
809 <td class="left">
810 <tt>[40 x int ]</tt><br/>
811 <tt>[41 x int ]</tt><br/>
812 <tt>[40 x uint]</tt><br/>
813 </td>
814 <td class="left">
815 Array of 40 integer values.<br/>
816 Array of 41 integer values.<br/>
817 Array of 40 unsigned integer values.<br/>
818 </td>
819 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000820</table>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000821<p>Here are some examples of multidimensional arrays:</p>
822<table class="layout">
823 <tr class="layout">
824 <td class="left">
825 <tt>[3 x [4 x int]]</tt><br/>
826 <tt>[12 x [10 x float]]</tt><br/>
827 <tt>[2 x [3 x [4 x uint]]]</tt><br/>
828 </td>
829 <td class="left">
John Criswell4a3327e2005-05-13 22:25:59 +0000830 3x4 array of integer values.<br/>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000831 12x10 array of single precision floating point values.<br/>
832 2x3x4 array of unsigned integer values.<br/>
833 </td>
834 </tr>
835</table>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000836
John Criswell4c0cf7f2005-10-24 16:17:18 +0000837<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
838length array. Normally, accesses past the end of an array are undefined in
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000839LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
840As a special case, however, zero length arrays are recognized to be variable
841length. This allows implementation of 'pascal style arrays' with the LLVM
842type "{ int, [0 x float]}", for example.</p>
843
Misha Brukman76307852003-11-08 01:05:38 +0000844</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000845
Chris Lattner2f7c9632001-06-06 20:29:01 +0000846<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000847<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000848<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000849<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000850<p>The function type can be thought of as a function signature. It
851consists of a return type and a list of formal parameter types.
John Criswella0d50d22003-11-25 21:45:46 +0000852Function types are usually used to build virtual function tables
Chris Lattner48b383b02003-11-25 01:02:51 +0000853(which are structures of pointers to functions), for indirect function
854calls, and when defining a function.</p>
John Criswella0d50d22003-11-25 21:45:46 +0000855<p>
856The return type of a function type cannot be an aggregate type.
857</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000858<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000859<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
John Criswell4c0cf7f2005-10-24 16:17:18 +0000860<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Misha Brukman20f9a622004-08-12 20:16:08 +0000861specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner5ed60612003-09-03 00:41:47 +0000862which indicates that the function takes a variable number of arguments.
863Variable argument functions can access their arguments with the <a
Chris Lattner48b383b02003-11-25 01:02:51 +0000864 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000865<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000866<table class="layout">
867 <tr class="layout">
868 <td class="left">
869 <tt>int (int)</tt> <br/>
870 <tt>float (int, int *) *</tt><br/>
871 <tt>int (sbyte *, ...)</tt><br/>
872 </td>
873 <td class="left">
874 function taking an <tt>int</tt>, returning an <tt>int</tt><br/>
875 <a href="#t_pointer">Pointer</a> to a function that takes an
Misha Brukman20f9a622004-08-12 20:16:08 +0000876 <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000877 returning <tt>float</tt>.<br/>
878 A vararg function that takes at least one <a href="#t_pointer">pointer</a>
879 to <tt>sbyte</tt> (signed char in C), which returns an integer. This is
880 the signature for <tt>printf</tt> in LLVM.<br/>
881 </td>
882 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000883</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000884
Misha Brukman76307852003-11-08 01:05:38 +0000885</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000886<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000887<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000888<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000889<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000890<p>The structure type is used to represent a collection of data members
891together in memory. The packing of the field types is defined to match
892the ABI of the underlying processor. The elements of a structure may
893be any type that has a size.</p>
894<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
895and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
896field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
897instruction.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000898<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000899<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000900<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000901<table class="layout">
902 <tr class="layout">
903 <td class="left">
904 <tt>{ int, int, int }</tt><br/>
905 <tt>{ float, int (int) * }</tt><br/>
906 </td>
907 <td class="left">
908 a triple of three <tt>int</tt> values<br/>
909 A pair, where the first element is a <tt>float</tt> and the second element
910 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
911 that takes an <tt>int</tt>, returning an <tt>int</tt>.<br/>
912 </td>
913 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000914</table>
Misha Brukman76307852003-11-08 01:05:38 +0000915</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000916
Chris Lattner2f7c9632001-06-06 20:29:01 +0000917<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000918<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000919<div class="doc_text">
Chris Lattner590645f2002-04-14 06:13:44 +0000920<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000921<p>As in many languages, the pointer type represents a pointer or
922reference to another object, which must live in memory.</p>
Chris Lattner590645f2002-04-14 06:13:44 +0000923<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000924<pre> &lt;type&gt; *<br></pre>
Chris Lattner590645f2002-04-14 06:13:44 +0000925<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000926<table class="layout">
927 <tr class="layout">
928 <td class="left">
929 <tt>[4x int]*</tt><br/>
930 <tt>int (int *) *</tt><br/>
931 </td>
932 <td class="left">
933 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
934 four <tt>int</tt> values<br/>
935 A <a href="#t_pointer">pointer</a> to a <a
Chris Lattner16fb0032005-02-19 02:22:14 +0000936 href="#t_function">function</a> that takes an <tt>int*</tt>, returning an
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000937 <tt>int</tt>.<br/>
938 </td>
939 </tr>
Misha Brukman76307852003-11-08 01:05:38 +0000940</table>
Misha Brukman76307852003-11-08 01:05:38 +0000941</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000942
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000943<!-- _______________________________________________________________________ -->
944<div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000945<div class="doc_text">
Chris Lattner37b6b092005-04-25 17:34:15 +0000946
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000947<h5>Overview:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +0000948
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000949<p>A packed type is a simple derived type that represents a vector
950of elements. Packed types are used when multiple primitive data
951are operated in parallel using a single instruction (SIMD).
952A packed type requires a size (number of
Chris Lattner330ce692005-11-10 01:44:22 +0000953elements) and an underlying primitive data type. Vectors must have a power
954of two length (1, 2, 4, 8, 16 ...). Packed types are
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000955considered <a href="#t_firstclass">first class</a>.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +0000956
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000957<h5>Syntax:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +0000958
959<pre>
960 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
961</pre>
962
John Criswell4a3327e2005-05-13 22:25:59 +0000963<p>The number of elements is a constant integer value; elementtype may
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000964be any integral or floating point type.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +0000965
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000966<h5>Examples:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +0000967
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000968<table class="layout">
969 <tr class="layout">
970 <td class="left">
971 <tt>&lt;4 x int&gt;</tt><br/>
972 <tt>&lt;8 x float&gt;</tt><br/>
973 <tt>&lt;2 x uint&gt;</tt><br/>
974 </td>
975 <td class="left">
976 Packed vector of 4 integer values.<br/>
977 Packed vector of 8 floating-point values.<br/>
978 Packed vector of 2 unsigned integer values.<br/>
979 </td>
980 </tr>
981</table>
Misha Brukman76307852003-11-08 01:05:38 +0000982</div>
983
Chris Lattner37b6b092005-04-25 17:34:15 +0000984<!-- _______________________________________________________________________ -->
985<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
986<div class="doc_text">
987
988<h5>Overview:</h5>
989
990<p>Opaque types are used to represent unknown types in the system. This
991corresponds (for example) to the C notion of a foward declared structure type.
992In LLVM, opaque types can eventually be resolved to any type (not just a
993structure type).</p>
994
995<h5>Syntax:</h5>
996
997<pre>
998 opaque
999</pre>
1000
1001<h5>Examples:</h5>
1002
1003<table class="layout">
1004 <tr class="layout">
1005 <td class="left">
1006 <tt>opaque</tt>
1007 </td>
1008 <td class="left">
1009 An opaque type.<br/>
1010 </td>
1011 </tr>
1012</table>
1013</div>
1014
1015
Chris Lattner74d3f822004-12-09 17:30:23 +00001016<!-- *********************************************************************** -->
1017<div class="doc_section"> <a name="constants">Constants</a> </div>
1018<!-- *********************************************************************** -->
1019
1020<div class="doc_text">
1021
1022<p>LLVM has several different basic types of constants. This section describes
1023them all and their syntax.</p>
1024
1025</div>
1026
1027<!-- ======================================================================= -->
Reid Spencer8f08d802004-12-09 18:02:53 +00001028<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001029
1030<div class="doc_text">
1031
1032<dl>
1033 <dt><b>Boolean constants</b></dt>
1034
1035 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1036 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
1037 </dd>
1038
1039 <dt><b>Integer constants</b></dt>
1040
Reid Spencer8f08d802004-12-09 18:02:53 +00001041 <dd>Standard integers (such as '4') are constants of the <a
Chris Lattner74d3f822004-12-09 17:30:23 +00001042 href="#t_integer">integer</a> type. Negative numbers may be used with signed
1043 integer types.
1044 </dd>
1045
1046 <dt><b>Floating point constants</b></dt>
1047
1048 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1049 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattner74d3f822004-12-09 17:30:23 +00001050 notation (see below). Floating point constants must have a <a
1051 href="#t_floating">floating point</a> type. </dd>
1052
1053 <dt><b>Null pointer constants</b></dt>
1054
John Criswelldfe6a862004-12-10 15:51:16 +00001055 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Chris Lattner74d3f822004-12-09 17:30:23 +00001056 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1057
1058</dl>
1059
John Criswelldfe6a862004-12-10 15:51:16 +00001060<p>The one non-intuitive notation for constants is the optional hexadecimal form
Chris Lattner74d3f822004-12-09 17:30:23 +00001061of floating point constants. For example, the form '<tt>double
10620x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
10634.5e+15</tt>'. The only time hexadecimal floating point constants are required
Reid Spencer8f08d802004-12-09 18:02:53 +00001064(and the only time that they are generated by the disassembler) is when a
1065floating point constant must be emitted but it cannot be represented as a
1066decimal floating point number. For example, NaN's, infinities, and other
1067special values are represented in their IEEE hexadecimal format so that
1068assembly and disassembly do not cause any bits to change in the constants.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001069
1070</div>
1071
1072<!-- ======================================================================= -->
1073<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1074</div>
1075
1076<div class="doc_text">
Chris Lattner455fc8c2005-03-07 22:13:59 +00001077<p>Aggregate constants arise from aggregation of simple constants
1078and smaller aggregate constants.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001079
1080<dl>
1081 <dt><b>Structure constants</b></dt>
1082
1083 <dd>Structure constants are represented with notation similar to structure
1084 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner455fc8c2005-03-07 22:13:59 +00001085 (<tt>{}</tt>)). For example: "<tt>{ int 4, float 17.0, int* %G }</tt>",
1086 where "<tt>%G</tt>" is declared as "<tt>%G = external global int</tt>". Structure constants
1087 must have <a href="#t_struct">structure type</a>, and the number and
Chris Lattner74d3f822004-12-09 17:30:23 +00001088 types of elements must match those specified by the type.
1089 </dd>
1090
1091 <dt><b>Array constants</b></dt>
1092
1093 <dd>Array constants are represented with notation similar to array type
1094 definitions (a comma separated list of elements, surrounded by square brackets
John Criswelldfe6a862004-12-10 15:51:16 +00001095 (<tt>[]</tt>)). For example: "<tt>[ int 42, int 11, int 74 ]</tt>". Array
Chris Lattner74d3f822004-12-09 17:30:23 +00001096 constants must have <a href="#t_array">array type</a>, and the number and
1097 types of elements must match those specified by the type.
1098 </dd>
1099
1100 <dt><b>Packed constants</b></dt>
1101
1102 <dd>Packed constants are represented with notation similar to packed type
1103 definitions (a comma separated list of elements, surrounded by
John Criswelldfe6a862004-12-10 15:51:16 +00001104 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; int 42,
Chris Lattner74d3f822004-12-09 17:30:23 +00001105 int 11, int 74, int 100 &gt;</tt>". Packed constants must have <a
1106 href="#t_packed">packed type</a>, and the number and types of elements must
1107 match those specified by the type.
1108 </dd>
1109
1110 <dt><b>Zero initialization</b></dt>
1111
1112 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1113 value to zero of <em>any</em> type, including scalar and aggregate types.
1114 This is often used to avoid having to print large zero initializers (e.g. for
John Criswell4c0cf7f2005-10-24 16:17:18 +00001115 large arrays) and is always exactly equivalent to using explicit zero
Chris Lattner74d3f822004-12-09 17:30:23 +00001116 initializers.
1117 </dd>
1118</dl>
1119
1120</div>
1121
1122<!-- ======================================================================= -->
1123<div class="doc_subsection">
1124 <a name="globalconstants">Global Variable and Function Addresses</a>
1125</div>
1126
1127<div class="doc_text">
1128
1129<p>The addresses of <a href="#globalvars">global variables</a> and <a
1130href="#functionstructure">functions</a> are always implicitly valid (link-time)
John Criswelldfe6a862004-12-10 15:51:16 +00001131constants. These constants are explicitly referenced when the <a
1132href="#identifiers">identifier for the global</a> is used and always have <a
Chris Lattner74d3f822004-12-09 17:30:23 +00001133href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1134file:</p>
1135
1136<pre>
1137 %X = global int 17
1138 %Y = global int 42
1139 %Z = global [2 x int*] [ int* %X, int* %Y ]
1140</pre>
1141
1142</div>
1143
1144<!-- ======================================================================= -->
Reid Spencer641f5c92004-12-09 18:13:12 +00001145<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001146<div class="doc_text">
Reid Spencer641f5c92004-12-09 18:13:12 +00001147 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
John Criswell4a3327e2005-05-13 22:25:59 +00001148 no specific value. Undefined values may be of any type and be used anywhere
Reid Spencer641f5c92004-12-09 18:13:12 +00001149 a constant is permitted.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001150
Reid Spencer641f5c92004-12-09 18:13:12 +00001151 <p>Undefined values indicate to the compiler that the program is well defined
1152 no matter what value is used, giving the compiler more freedom to optimize.
1153 </p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001154</div>
1155
1156<!-- ======================================================================= -->
1157<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1158</div>
1159
1160<div class="doc_text">
1161
1162<p>Constant expressions are used to allow expressions involving other constants
1163to be used as constants. Constant expressions may be of any <a
John Criswell4a3327e2005-05-13 22:25:59 +00001164href="#t_firstclass">first class</a> type and may involve any LLVM operation
Chris Lattner74d3f822004-12-09 17:30:23 +00001165that does not have side effects (e.g. load and call are not supported). The
1166following is the syntax for constant expressions:</p>
1167
1168<dl>
1169 <dt><b><tt>cast ( CST to TYPE )</tt></b></dt>
1170
1171 <dd>Cast a constant to another type.</dd>
1172
1173 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1174
1175 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1176 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1177 instruction, the index list may have zero or more indexes, which are required
1178 to make sense for the type of "CSTPTR".</dd>
1179
Robert Bocchino7e97a6d2006-01-10 19:31:34 +00001180 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1181
1182 <dd>Perform the <a href="#i_select">select operation</a> on
1183 constants.
1184
1185 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1186
1187 <dd>Perform the <a href="#i_extractelement">extractelement
1188 operation</a> on constants.
1189
Robert Bocchinof72fdfe2006-01-15 20:48:27 +00001190 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1191
1192 <dd>Perform the <a href="#i_insertelement">insertelement
1193 operation</a> on constants.
1194
Chris Lattner016a0e52006-04-08 00:13:41 +00001195
1196 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1197
1198 <dd>Perform the <a href="#i_shufflevector">shufflevector
1199 operation</a> on constants.
1200
Chris Lattner74d3f822004-12-09 17:30:23 +00001201 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1202
Reid Spencer641f5c92004-12-09 18:13:12 +00001203 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1204 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
Chris Lattner74d3f822004-12-09 17:30:23 +00001205 binary</a> operations. The constraints on operands are the same as those for
1206 the corresponding instruction (e.g. no bitwise operations on floating point
John Criswell02fdc6f2005-05-12 16:52:32 +00001207 values are allowed).</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00001208</dl>
Chris Lattner74d3f822004-12-09 17:30:23 +00001209</div>
Chris Lattnerb1652612004-03-08 16:49:10 +00001210
Chris Lattner2f7c9632001-06-06 20:29:01 +00001211<!-- *********************************************************************** -->
Chris Lattner98f013c2006-01-25 23:47:57 +00001212<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1213<!-- *********************************************************************** -->
1214
1215<!-- ======================================================================= -->
1216<div class="doc_subsection">
1217<a name="inlineasm">Inline Assembler Expressions</a>
1218</div>
1219
1220<div class="doc_text">
1221
1222<p>
1223LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1224Module-Level Inline Assembly</a>) through the use of a special value. This
1225value represents the inline assembler as a string (containing the instructions
1226to emit), a list of operand constraints (stored as a string), and a flag that
1227indicates whether or not the inline asm expression has side effects. An example
1228inline assembler expression is:
1229</p>
1230
1231<pre>
1232 int(int) asm "bswap $0", "=r,r"
1233</pre>
1234
1235<p>
1236Inline assembler expressions may <b>only</b> be used as the callee operand of
1237a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1238</p>
1239
1240<pre>
1241 %X = call int asm "<a href="#i_bswap">bswap</a> $0", "=r,r"(int %Y)
1242</pre>
1243
1244<p>
1245Inline asms with side effects not visible in the constraint list must be marked
1246as having side effects. This is done through the use of the
1247'<tt>sideeffect</tt>' keyword, like so:
1248</p>
1249
1250<pre>
1251 call void asm sideeffect "eieio", ""()
1252</pre>
1253
1254<p>TODO: The format of the asm and constraints string still need to be
1255documented here. Constraints on what can be done (e.g. duplication, moving, etc
1256need to be documented).
1257</p>
1258
1259</div>
1260
1261<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001262<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1263<!-- *********************************************************************** -->
Chris Lattner74d3f822004-12-09 17:30:23 +00001264
Misha Brukman76307852003-11-08 01:05:38 +00001265<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +00001266
Chris Lattner48b383b02003-11-25 01:02:51 +00001267<p>The LLVM instruction set consists of several different
1268classifications of instructions: <a href="#terminators">terminator
John Criswell4a3327e2005-05-13 22:25:59 +00001269instructions</a>, <a href="#binaryops">binary instructions</a>,
1270<a href="#bitwiseops">bitwise binary instructions</a>, <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001271 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1272instructions</a>.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001273
Misha Brukman76307852003-11-08 01:05:38 +00001274</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001275
Chris Lattner2f7c9632001-06-06 20:29:01 +00001276<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001277<div class="doc_subsection"> <a name="terminators">Terminator
1278Instructions</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001279
Misha Brukman76307852003-11-08 01:05:38 +00001280<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +00001281
Chris Lattner48b383b02003-11-25 01:02:51 +00001282<p>As mentioned <a href="#functionstructure">previously</a>, every
1283basic block in a program ends with a "Terminator" instruction, which
1284indicates which block should be executed after the current block is
1285finished. These terminator instructions typically yield a '<tt>void</tt>'
1286value: they produce control flow, not values (the one exception being
1287the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
John Criswelldfe6a862004-12-10 15:51:16 +00001288<p>There are six different terminator instructions: the '<a
Chris Lattner48b383b02003-11-25 01:02:51 +00001289 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1290instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001291the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1292 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1293 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001294
Misha Brukman76307852003-11-08 01:05:38 +00001295</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001296
Chris Lattner2f7c9632001-06-06 20:29:01 +00001297<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001298<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1299Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001300<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001301<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001302<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner590645f2002-04-14 06:13:44 +00001303 ret void <i>; Return from void function</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001304</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001305<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001306<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
John Criswell4a3327e2005-05-13 22:25:59 +00001307value) from a function back to the caller.</p>
John Criswell417228d2004-04-09 16:48:45 +00001308<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner48b383b02003-11-25 01:02:51 +00001309returns a value and then causes control flow, and one that just causes
1310control flow to occur.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001311<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001312<p>The '<tt>ret</tt>' instruction may return any '<a
1313 href="#t_firstclass">first class</a>' type. Notice that a function is
1314not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1315instruction inside of the function that returns a value that does not
1316match the return type of the function.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001317<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001318<p>When the '<tt>ret</tt>' instruction is executed, control flow
1319returns back to the calling function's context. If the caller is a "<a
John Criswell40db33f2004-06-25 15:16:57 +00001320 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner48b383b02003-11-25 01:02:51 +00001321the instruction after the call. If the caller was an "<a
1322 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
John Criswell02fdc6f2005-05-12 16:52:32 +00001323at the beginning of the "normal" destination block. If the instruction
Chris Lattner48b383b02003-11-25 01:02:51 +00001324returns a value, that value shall set the call or invoke instruction's
1325return value.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001326<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001327<pre> ret int 5 <i>; Return an integer value of 5</i>
Chris Lattner590645f2002-04-14 06:13:44 +00001328 ret void <i>; Return from a void function</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001329</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001330</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001331<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001332<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001333<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001334<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001335<pre> br bool &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001336</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001337<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001338<p>The '<tt>br</tt>' instruction is used to cause control flow to
1339transfer to a different basic block in the current function. There are
1340two forms of this instruction, corresponding to a conditional branch
1341and an unconditional branch.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001342<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001343<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1344single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1345unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1346value as a target.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001347<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001348<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1349argument is evaluated. If the value is <tt>true</tt>, control flows
1350to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1351control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001352<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001353<pre>Test:<br> %cond = <a href="#i_setcc">seteq</a> int %a, %b<br> br bool %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
1354 href="#i_ret">ret</a> int 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> int 0<br></pre>
Misha Brukman76307852003-11-08 01:05:38 +00001355</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001356<!-- _______________________________________________________________________ -->
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001357<div class="doc_subsubsection">
1358 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1359</div>
1360
Misha Brukman76307852003-11-08 01:05:38 +00001361<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001362<h5>Syntax:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001363
1364<pre>
1365 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1366</pre>
1367
Chris Lattner2f7c9632001-06-06 20:29:01 +00001368<h5>Overview:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001369
1370<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1371several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman76307852003-11-08 01:05:38 +00001372instruction, allowing a branch to occur to one of many possible
1373destinations.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001374
1375
Chris Lattner2f7c9632001-06-06 20:29:01 +00001376<h5>Arguments:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001377
1378<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1379comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1380an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1381table is not allowed to contain duplicate constant entries.</p>
1382
Chris Lattner2f7c9632001-06-06 20:29:01 +00001383<h5>Semantics:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001384
Chris Lattner48b383b02003-11-25 01:02:51 +00001385<p>The <tt>switch</tt> instruction specifies a table of values and
1386destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswellbcbb18c2004-06-25 16:05:06 +00001387table is searched for the given value. If the value is found, control flow is
1388transfered to the corresponding destination; otherwise, control flow is
1389transfered to the default destination.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001390
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001391<h5>Implementation:</h5>
1392
1393<p>Depending on properties of the target machine and the particular
1394<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswellbcbb18c2004-06-25 16:05:06 +00001395ways. For example, it could be generated as a series of chained conditional
1396branches or with a lookup table.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00001397
1398<h5>Example:</h5>
1399
1400<pre>
1401 <i>; Emulate a conditional br instruction</i>
1402 %Val = <a href="#i_cast">cast</a> bool %value to int
1403 switch int %Val, label %truedest [int 0, label %falsedest ]
1404
1405 <i>; Emulate an unconditional br instruction</i>
1406 switch uint 0, label %dest [ ]
1407
1408 <i>; Implement a jump table:</i>
1409 switch uint %val, label %otherwise [ uint 0, label %onzero
1410 uint 1, label %onone
1411 uint 2, label %ontwo ]
Chris Lattner2f7c9632001-06-06 20:29:01 +00001412</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001413</div>
Chris Lattner0132aff2005-05-06 22:57:40 +00001414
Chris Lattner2f7c9632001-06-06 20:29:01 +00001415<!-- _______________________________________________________________________ -->
Chris Lattner0132aff2005-05-06 22:57:40 +00001416<div class="doc_subsubsection">
1417 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1418</div>
1419
Misha Brukman76307852003-11-08 01:05:38 +00001420<div class="doc_text">
Chris Lattner0132aff2005-05-06 22:57:40 +00001421
Chris Lattner2f7c9632001-06-06 20:29:01 +00001422<h5>Syntax:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00001423
1424<pre>
1425 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; %&lt;function ptr val&gt;(&lt;function args&gt;)
Chris Lattner6b7a0082006-05-14 18:23:06 +00001426 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattner0132aff2005-05-06 22:57:40 +00001427</pre>
1428
Chris Lattnera8292f32002-05-06 22:08:29 +00001429<h5>Overview:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00001430
1431<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1432function, with the possibility of control flow transfer to either the
John Criswell02fdc6f2005-05-12 16:52:32 +00001433'<tt>normal</tt>' label or the
1434'<tt>exception</tt>' label. If the callee function returns with the
Chris Lattner0132aff2005-05-06 22:57:40 +00001435"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1436"normal" label. If the callee (or any indirect callees) returns with the "<a
John Criswell02fdc6f2005-05-12 16:52:32 +00001437href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1438continued at the dynamically nearest "exception" label.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00001439
Chris Lattner2f7c9632001-06-06 20:29:01 +00001440<h5>Arguments:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00001441
Misha Brukman76307852003-11-08 01:05:38 +00001442<p>This instruction requires several arguments:</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00001443
Chris Lattner2f7c9632001-06-06 20:29:01 +00001444<ol>
Chris Lattner0132aff2005-05-06 22:57:40 +00001445 <li>
John Criswell4a3327e2005-05-13 22:25:59 +00001446 The optional "cconv" marker indicates which <a href="callingconv">calling
Chris Lattner0132aff2005-05-06 22:57:40 +00001447 convention</a> the call should use. If none is specified, the call defaults
1448 to using C calling conventions.
1449 </li>
1450 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1451 function value being invoked. In most cases, this is a direct function
1452 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1453 an arbitrary pointer to function value.
1454 </li>
1455
1456 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1457 function to be invoked. </li>
1458
1459 <li>'<tt>function args</tt>': argument list whose types match the function
1460 signature argument types. If the function signature indicates the function
1461 accepts a variable number of arguments, the extra arguments can be
1462 specified. </li>
1463
1464 <li>'<tt>normal label</tt>': the label reached when the called function
1465 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1466
1467 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1468 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1469
Chris Lattner2f7c9632001-06-06 20:29:01 +00001470</ol>
Chris Lattner0132aff2005-05-06 22:57:40 +00001471
Chris Lattner2f7c9632001-06-06 20:29:01 +00001472<h5>Semantics:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00001473
Misha Brukman76307852003-11-08 01:05:38 +00001474<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattner0132aff2005-05-06 22:57:40 +00001475href="#i_call">call</a></tt>' instruction in most regards. The primary
1476difference is that it establishes an association with a label, which is used by
1477the runtime library to unwind the stack.</p>
1478
1479<p>This instruction is used in languages with destructors to ensure that proper
1480cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1481exception. Additionally, this is important for implementation of
1482'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1483
Chris Lattner2f7c9632001-06-06 20:29:01 +00001484<h5>Example:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00001485<pre>
1486 %retval = invoke int %Test(int 15) to label %Continue
Chris Lattner6b7a0082006-05-14 18:23:06 +00001487 unwind label %TestCleanup <i>; {int}:retval set</i>
Chris Lattner0132aff2005-05-06 22:57:40 +00001488 %retval = invoke <a href="#callingconv">coldcc</a> int %Test(int 15) to label %Continue
Chris Lattner6b7a0082006-05-14 18:23:06 +00001489 unwind label %TestCleanup <i>; {int}:retval set</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001490</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001491</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001492
1493
Chris Lattner5ed60612003-09-03 00:41:47 +00001494<!-- _______________________________________________________________________ -->
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001495
Chris Lattner48b383b02003-11-25 01:02:51 +00001496<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1497Instruction</a> </div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001498
Misha Brukman76307852003-11-08 01:05:38 +00001499<div class="doc_text">
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001500
Chris Lattner5ed60612003-09-03 00:41:47 +00001501<h5>Syntax:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001502<pre>
1503 unwind
1504</pre>
1505
Chris Lattner5ed60612003-09-03 00:41:47 +00001506<h5>Overview:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001507
1508<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1509at the first callee in the dynamic call stack which used an <a
1510href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1511primarily used to implement exception handling.</p>
1512
Chris Lattner5ed60612003-09-03 00:41:47 +00001513<h5>Semantics:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001514
1515<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1516immediately halt. The dynamic call stack is then searched for the first <a
1517href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1518execution continues at the "exceptional" destination block specified by the
1519<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1520dynamic call chain, undefined behavior results.</p>
Misha Brukman76307852003-11-08 01:05:38 +00001521</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00001522
1523<!-- _______________________________________________________________________ -->
1524
1525<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1526Instruction</a> </div>
1527
1528<div class="doc_text">
1529
1530<h5>Syntax:</h5>
1531<pre>
1532 unreachable
1533</pre>
1534
1535<h5>Overview:</h5>
1536
1537<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1538instruction is used to inform the optimizer that a particular portion of the
1539code is not reachable. This can be used to indicate that the code after a
1540no-return function cannot be reached, and other facts.</p>
1541
1542<h5>Semantics:</h5>
1543
1544<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1545</div>
1546
1547
1548
Chris Lattner2f7c9632001-06-06 20:29:01 +00001549<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001550<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001551<div class="doc_text">
Chris Lattner48b383b02003-11-25 01:02:51 +00001552<p>Binary operators are used to do most of the computation in a
1553program. They require two operands, execute an operation on them, and
John Criswelldfe6a862004-12-10 15:51:16 +00001554produce a single value. The operands might represent
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001555multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1556The result value of a binary operator is not
Chris Lattner48b383b02003-11-25 01:02:51 +00001557necessarily the same type as its operands.</p>
Misha Brukman76307852003-11-08 01:05:38 +00001558<p>There are several different binary operators:</p>
Misha Brukman76307852003-11-08 01:05:38 +00001559</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001560<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001561<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1562Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001563<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001564<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001565<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001566</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001567<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001568<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001569<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001570<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001571 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1572 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1573Both arguments must have identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001574<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001575<p>The value produced is the integer or floating point sum of the two
1576operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001577<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001578<pre> &lt;result&gt; = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001579</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001580</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001581<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001582<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1583Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001584<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001585<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001586<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001587</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001588<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001589<p>The '<tt>sub</tt>' instruction returns the difference of its two
1590operands.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001591<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1592instruction present in most other intermediate representations.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001593<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001594<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001595 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001596values.
1597This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1598Both arguments must have identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001599<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001600<p>The value produced is the integer or floating point difference of
1601the two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001602<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001603<pre> &lt;result&gt; = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001604 &lt;result&gt; = sub int 0, %val <i>; yields {int}:result = -%var</i>
1605</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001606</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001607<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001608<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1609Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001610<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001611<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001612<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001613</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001614<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001615<p>The '<tt>mul</tt>' instruction returns the product of its two
1616operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001617<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001618<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001619 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001620values.
1621This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1622Both arguments must have identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001623<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001624<p>The value produced is the integer or floating point product of the
Misha Brukman76307852003-11-08 01:05:38 +00001625two operands.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001626<p>There is no signed vs unsigned multiplication. The appropriate
1627action is taken based on the type of the operand.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001628<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001629<pre> &lt;result&gt; = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001630</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001631</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001632<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001633<div class="doc_subsubsection"> <a name="i_div">'<tt>div</tt>'
1634Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001635<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001636<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001637<pre> &lt;result&gt; = div &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1638</pre>
1639<h5>Overview:</h5>
1640<p>The '<tt>div</tt>' instruction returns the quotient of its two
1641operands.</p>
1642<h5>Arguments:</h5>
1643<p>The two arguments to the '<tt>div</tt>' instruction must be either <a
1644 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001645values.
1646This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1647Both arguments must have identical types.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001648<h5>Semantics:</h5>
1649<p>The value produced is the integer or floating point quotient of the
1650two operands.</p>
1651<h5>Example:</h5>
1652<pre> &lt;result&gt; = div int 4, %var <i>; yields {int}:result = 4 / %var</i>
1653</pre>
1654</div>
1655<!-- _______________________________________________________________________ -->
1656<div class="doc_subsubsection"> <a name="i_rem">'<tt>rem</tt>'
1657Instruction</a> </div>
1658<div class="doc_text">
1659<h5>Syntax:</h5>
1660<pre> &lt;result&gt; = rem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1661</pre>
1662<h5>Overview:</h5>
1663<p>The '<tt>rem</tt>' instruction returns the remainder from the
1664division of its two operands.</p>
1665<h5>Arguments:</h5>
1666<p>The two arguments to the '<tt>rem</tt>' instruction must be either <a
1667 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001668values.
1669This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1670Both arguments must have identical types.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001671<h5>Semantics:</h5>
1672<p>This returns the <i>remainder</i> of a division (where the result
1673has the same sign as the divisor), not the <i>modulus</i> (where the
1674result has the same sign as the dividend) of a value. For more
John Criswell4c0cf7f2005-10-24 16:17:18 +00001675information about the difference, see <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001676 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1677Math Forum</a>.</p>
1678<h5>Example:</h5>
1679<pre> &lt;result&gt; = rem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1680</pre>
Robert Bocchino820bc75b2006-02-17 21:18:08 +00001681
Chris Lattner48b383b02003-11-25 01:02:51 +00001682</div>
1683<!-- _______________________________________________________________________ -->
1684<div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1685Instructions</a> </div>
1686<div class="doc_text">
1687<h5>Syntax:</h5>
1688<pre> &lt;result&gt; = seteq &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001689 &lt;result&gt; = setne &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1690 &lt;result&gt; = setlt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1691 &lt;result&gt; = setgt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1692 &lt;result&gt; = setle &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1693 &lt;result&gt; = setge &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1694</pre>
Chris Lattner48b383b02003-11-25 01:02:51 +00001695<h5>Overview:</h5>
1696<p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1697value based on a comparison of their two operands.</p>
1698<h5>Arguments:</h5>
1699<p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1700be of <a href="#t_firstclass">first class</a> type (it is not possible
1701to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1702or '<tt>void</tt>' values, etc...). Both arguments must have identical
1703types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001704<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001705<p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1706value if both operands are equal.<br>
1707The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1708value if both operands are unequal.<br>
1709The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1710value if the first operand is less than the second operand.<br>
1711The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1712value if the first operand is greater than the second operand.<br>
1713The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1714value if the first operand is less than or equal to the second operand.<br>
1715The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1716value if the first operand is greater than or equal to the second
1717operand.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001718<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001719<pre> &lt;result&gt; = seteq int 4, 5 <i>; yields {bool}:result = false</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001720 &lt;result&gt; = setne float 4, 5 <i>; yields {bool}:result = true</i>
1721 &lt;result&gt; = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1722 &lt;result&gt; = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1723 &lt;result&gt; = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1724 &lt;result&gt; = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1725</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001726</div>
Robert Bocchino820bc75b2006-02-17 21:18:08 +00001727
Chris Lattner2f7c9632001-06-06 20:29:01 +00001728<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001729<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1730Operations</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001731<div class="doc_text">
Chris Lattner48b383b02003-11-25 01:02:51 +00001732<p>Bitwise binary operators are used to do various forms of
1733bit-twiddling in a program. They are generally very efficient
John Criswelldfe6a862004-12-10 15:51:16 +00001734instructions and can commonly be strength reduced from other
Chris Lattner48b383b02003-11-25 01:02:51 +00001735instructions. They require two operands, execute an operation on them,
1736and produce a single value. The resulting value of the bitwise binary
1737operators is always the same type as its first operand.</p>
Misha Brukman76307852003-11-08 01:05:38 +00001738</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001739<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001740<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1741Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001742<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001743<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001744<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001745</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001746<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001747<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1748its two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001749<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001750<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001751 href="#t_integral">integral</a> values. Both arguments must have
1752identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001753<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001754<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001755<p> </p>
Misha Brukmanc501f552004-03-01 17:47:27 +00001756<div style="align: center">
Misha Brukman76307852003-11-08 01:05:38 +00001757<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +00001758 <tbody>
1759 <tr>
1760 <td>In0</td>
1761 <td>In1</td>
1762 <td>Out</td>
1763 </tr>
1764 <tr>
1765 <td>0</td>
1766 <td>0</td>
1767 <td>0</td>
1768 </tr>
1769 <tr>
1770 <td>0</td>
1771 <td>1</td>
1772 <td>0</td>
1773 </tr>
1774 <tr>
1775 <td>1</td>
1776 <td>0</td>
1777 <td>0</td>
1778 </tr>
1779 <tr>
1780 <td>1</td>
1781 <td>1</td>
1782 <td>1</td>
1783 </tr>
1784 </tbody>
1785</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00001786</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001787<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001788<pre> &lt;result&gt; = and int 4, %var <i>; yields {int}:result = 4 &amp; %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001789 &lt;result&gt; = and int 15, 40 <i>; yields {int}:result = 8</i>
1790 &lt;result&gt; = and int 4, 8 <i>; yields {int}:result = 0</i>
1791</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001792</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001793<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001794<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001795<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001796<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001797<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001798</pre>
Chris Lattner48b383b02003-11-25 01:02:51 +00001799<h5>Overview:</h5>
1800<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1801or of its two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001802<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001803<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001804 href="#t_integral">integral</a> values. Both arguments must have
1805identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001806<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001807<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001808<p> </p>
Misha Brukmanc501f552004-03-01 17:47:27 +00001809<div style="align: center">
Chris Lattner48b383b02003-11-25 01:02:51 +00001810<table border="1" cellspacing="0" cellpadding="4">
1811 <tbody>
1812 <tr>
1813 <td>In0</td>
1814 <td>In1</td>
1815 <td>Out</td>
1816 </tr>
1817 <tr>
1818 <td>0</td>
1819 <td>0</td>
1820 <td>0</td>
1821 </tr>
1822 <tr>
1823 <td>0</td>
1824 <td>1</td>
1825 <td>1</td>
1826 </tr>
1827 <tr>
1828 <td>1</td>
1829 <td>0</td>
1830 <td>1</td>
1831 </tr>
1832 <tr>
1833 <td>1</td>
1834 <td>1</td>
1835 <td>1</td>
1836 </tr>
1837 </tbody>
1838</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00001839</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001840<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001841<pre> &lt;result&gt; = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001842 &lt;result&gt; = or int 15, 40 <i>; yields {int}:result = 47</i>
1843 &lt;result&gt; = or int 4, 8 <i>; yields {int}:result = 12</i>
1844</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001845</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001846<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001847<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1848Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001849<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001850<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001851<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001852</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001853<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001854<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
1855or of its two operands. The <tt>xor</tt> is used to implement the
1856"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001857<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001858<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001859 href="#t_integral">integral</a> values. Both arguments must have
1860identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001861<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001862<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00001863<p> </p>
Misha Brukmanc501f552004-03-01 17:47:27 +00001864<div style="align: center">
Chris Lattner48b383b02003-11-25 01:02:51 +00001865<table border="1" cellspacing="0" cellpadding="4">
1866 <tbody>
1867 <tr>
1868 <td>In0</td>
1869 <td>In1</td>
1870 <td>Out</td>
1871 </tr>
1872 <tr>
1873 <td>0</td>
1874 <td>0</td>
1875 <td>0</td>
1876 </tr>
1877 <tr>
1878 <td>0</td>
1879 <td>1</td>
1880 <td>1</td>
1881 </tr>
1882 <tr>
1883 <td>1</td>
1884 <td>0</td>
1885 <td>1</td>
1886 </tr>
1887 <tr>
1888 <td>1</td>
1889 <td>1</td>
1890 <td>0</td>
1891 </tr>
1892 </tbody>
1893</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00001894</div>
Chris Lattner48b383b02003-11-25 01:02:51 +00001895<p> </p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001896<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001897<pre> &lt;result&gt; = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001898 &lt;result&gt; = xor int 15, 40 <i>; yields {int}:result = 39</i>
1899 &lt;result&gt; = xor int 4, 8 <i>; yields {int}:result = 12</i>
Chris Lattner5ed60612003-09-03 00:41:47 +00001900 &lt;result&gt; = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001901</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001902</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001903<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001904<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
1905Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001906<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001907<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001908<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, ubyte &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001909</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001910<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001911<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
1912the left a specified number of bits.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001913<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001914<p>The first argument to the '<tt>shl</tt>' instruction must be an <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001915 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1916type.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001917<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001918<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001919<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001920<pre> &lt;result&gt; = shl int 4, ubyte %var <i>; yields {int}:result = 4 &lt;&lt; %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001921 &lt;result&gt; = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
1922 &lt;result&gt; = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
1923</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001924</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001925<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001926<div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
1927Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00001928<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +00001929<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001930<pre> &lt;result&gt; = shr &lt;ty&gt; &lt;var1&gt;, ubyte &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001931</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001932<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001933<p>The '<tt>shr</tt>' instruction returns the first operand shifted to
1934the right a specified number of bits.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001935<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001936<p>The first argument to the '<tt>shr</tt>' instruction must be an <a
Chris Lattner48b383b02003-11-25 01:02:51 +00001937 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
1938type.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001939<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001940<p>If the first argument is a <a href="#t_signed">signed</a> type, the
1941most significant bit is duplicated in the newly free'd bit positions.
1942If the first argument is unsigned, zero bits shall fill the empty
1943positions.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001944<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00001945<pre> &lt;result&gt; = shr int 4, ubyte %var <i>; yields {int}:result = 4 &gt;&gt; %var</i>
Chris Lattner33426d92003-06-18 21:30:51 +00001946 &lt;result&gt; = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001947 &lt;result&gt; = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
Chris Lattner33426d92003-06-18 21:30:51 +00001948 &lt;result&gt; = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
1949 &lt;result&gt; = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001950</pre>
Misha Brukman76307852003-11-08 01:05:38 +00001951</div>
Chris Lattner54611b42005-11-06 08:02:57 +00001952
Chris Lattner2f7c9632001-06-06 20:29:01 +00001953<!-- ======================================================================= -->
Chris Lattner54611b42005-11-06 08:02:57 +00001954<div class="doc_subsection">
Chris Lattnerce83bff2006-04-08 23:07:04 +00001955 <a name="vectorops">Vector Operations</a>
1956</div>
1957
1958<div class="doc_text">
1959
1960<p>LLVM supports several instructions to represent vector operations in a
1961target-independent manner. This instructions cover the element-access and
1962vector-specific operations needed to process vectors effectively. While LLVM
1963does directly support these vector operations, many sophisticated algorithms
1964will want to use target-specific intrinsics to take full advantage of a specific
1965target.</p>
1966
1967</div>
1968
1969<!-- _______________________________________________________________________ -->
1970<div class="doc_subsubsection">
1971 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
1972</div>
1973
1974<div class="doc_text">
1975
1976<h5>Syntax:</h5>
1977
1978<pre>
1979 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, uint &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
1980</pre>
1981
1982<h5>Overview:</h5>
1983
1984<p>
1985The '<tt>extractelement</tt>' instruction extracts a single scalar
1986element from a packed vector at a specified index.
1987</p>
1988
1989
1990<h5>Arguments:</h5>
1991
1992<p>
1993The first operand of an '<tt>extractelement</tt>' instruction is a
1994value of <a href="#t_packed">packed</a> type. The second operand is
1995an index indicating the position from which to extract the element.
1996The index may be a variable.</p>
1997
1998<h5>Semantics:</h5>
1999
2000<p>
2001The result is a scalar of the same type as the element type of
2002<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2003<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2004results are undefined.
2005</p>
2006
2007<h5>Example:</h5>
2008
2009<pre>
2010 %result = extractelement &lt;4 x int&gt; %vec, uint 0 <i>; yields int</i>
2011</pre>
2012</div>
2013
2014
2015<!-- _______________________________________________________________________ -->
2016<div class="doc_subsubsection">
2017 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2018</div>
2019
2020<div class="doc_text">
2021
2022<h5>Syntax:</h5>
2023
2024<pre>
2025 &lt;result&gt; = insertelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt, uint &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
2026</pre>
2027
2028<h5>Overview:</h5>
2029
2030<p>
2031The '<tt>insertelement</tt>' instruction inserts a scalar
2032element into a packed vector at a specified index.
2033</p>
2034
2035
2036<h5>Arguments:</h5>
2037
2038<p>
2039The first operand of an '<tt>insertelement</tt>' instruction is a
2040value of <a href="#t_packed">packed</a> type. The second operand is a
2041scalar value whose type must equal the element type of the first
2042operand. The third operand is an index indicating the position at
2043which to insert the value. The index may be a variable.</p>
2044
2045<h5>Semantics:</h5>
2046
2047<p>
2048The result is a packed vector of the same type as <tt>val</tt>. Its
2049element values are those of <tt>val</tt> except at position
2050<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2051exceeds the length of <tt>val</tt>, the results are undefined.
2052</p>
2053
2054<h5>Example:</h5>
2055
2056<pre>
2057 %result = insertelement &lt;4 x int&gt; %vec, int 1, uint 0 <i>; yields &lt;4 x int&gt;</i>
2058</pre>
2059</div>
2060
2061<!-- _______________________________________________________________________ -->
2062<div class="doc_subsubsection">
2063 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2064</div>
2065
2066<div class="doc_text">
2067
2068<h5>Syntax:</h5>
2069
2070<pre>
2071 &lt;result&gt; = shufflevector &lt;n x &lt;ty&gt;&gt; &lt;v1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;v2&gt;, &lt;n x uint&gt; &lt;mask&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
2072</pre>
2073
2074<h5>Overview:</h5>
2075
2076<p>
2077The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2078from two input vectors, returning a vector of the same type.
2079</p>
2080
2081<h5>Arguments:</h5>
2082
2083<p>
2084The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2085with types that match each other and types that match the result of the
2086instruction. The third argument is a shuffle mask, which has the same number
2087of elements as the other vector type, but whose element type is always 'uint'.
2088</p>
2089
2090<p>
2091The shuffle mask operand is required to be a constant vector with either
2092constant integer or undef values.
2093</p>
2094
2095<h5>Semantics:</h5>
2096
2097<p>
2098The elements of the two input vectors are numbered from left to right across
2099both of the vectors. The shuffle mask operand specifies, for each element of
2100the result vector, which element of the two input registers the result element
2101gets. The element selector may be undef (meaning "don't care") and the second
2102operand may be undef if performing a shuffle from only one vector.
2103</p>
2104
2105<h5>Example:</h5>
2106
2107<pre>
2108 %result = shufflevector &lt;4 x int&gt; %v1, &lt;4 x int&gt; %v2,
2109 &lt;4 x uint&gt; &lt;uint 0, uint 4, uint 1, uint 5&gt; <i>; yields &lt;4 x int&gt;</i>
2110 %result = shufflevector &lt;4 x int&gt; %v1, &lt;4 x int&gt; undef,
2111 &lt;4 x uint&gt; &lt;uint 0, uint 1, uint 2, uint 3&gt; <i>; yields &lt;4 x int&gt;</i> - Identity shuffle.
2112</pre>
2113</div>
2114
Tanya Lattnerb138bbe2006-04-14 19:24:33 +00002115
2116<!-- _______________________________________________________________________ -->
2117<div class="doc_subsubsection"> <a name="i_vsetint">'<tt>vsetint</tt>'
2118Instruction</a> </div>
2119<div class="doc_text">
2120<h5>Syntax:</h5>
2121<pre>&lt;result&gt; = vsetint &lt;op&gt;, &lt;n x &lt;ty&gt;&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields &lt;n x bool&gt;</i>
2122</pre>
2123
2124<h5>Overview:</h5>
2125
2126<p>The '<tt>vsetint</tt>' instruction takes two integer vectors and
2127returns a vector of boolean values representing, at each position, the
2128result of the comparison between the values at that position in the
2129two operands.</p>
2130
2131<h5>Arguments:</h5>
2132
2133<p>The arguments to a '<tt>vsetint</tt>' instruction are a comparison
2134operation and two value arguments. The value arguments must be of <a
2135href="#t_integral">integral</a> <a href="#t_packed">packed</a> type,
2136and they must have identical types. The operation argument must be
2137one of <tt>eq</tt>, <tt>ne</tt>, <tt>slt</tt>, <tt>sgt</tt>,
2138<tt>sle</tt>, <tt>sge</tt>, <tt>ult</tt>, <tt>ugt</tt>, <tt>ule</tt>,
2139<tt>uge</tt>, <tt>true</tt>, and <tt>false</tt>. The result is a
2140packed <tt>bool</tt> value with the same length as each operand.</p>
2141
2142<h5>Semantics:</h5>
2143
2144<p>The following table shows the semantics of '<tt>vsetint</tt>'. For
2145each position of the result, the comparison is done on the
2146corresponding positions of the two value arguments. Note that the
2147signedness of the comparison depends on the comparison opcode and
2148<i>not</i> on the signedness of the value operands. E.g., <tt>vsetint
2149slt <4 x unsigned> %x, %y</tt> does an elementwise <i>signed</i>
2150comparison of <tt>%x</tt> and <tt>%y</tt>.</p>
2151
2152<table border="1" cellspacing="0" cellpadding="4">
2153 <tbody>
2154 <tr><th>Operation</th><th>Result is true iff</th><th>Comparison is</th></tr>
2155 <tr><td><tt>eq</tt></td><td>var1 == var2</td><td>--</td></tr>
2156 <tr><td><tt>ne</tt></td><td>var1 != var2</td><td>--</td></tr>
2157 <tr><td><tt>slt</tt></td><td>var1 &lt; var2</td><td>signed</td></tr>
2158 <tr><td><tt>sgt</tt></td><td>var1 &gt; var2</td><td>signed</td></tr>
2159 <tr><td><tt>sle</tt></td><td>var1 &lt;= var2</td><td>signed</td></tr>
2160 <tr><td><tt>sge</tt></td><td>var1 &gt;= var2</td><td>signed</td></tr>
2161 <tr><td><tt>ult</tt></td><td>var1 &lt; var2</td><td>unsigned</td></tr>
2162 <tr><td><tt>ugt</tt></td><td>var1 &gt; var2</td><td>unsigned</td></tr>
2163 <tr><td><tt>ule</tt></td><td>var1 &lt;= var2</td><td>unsigned</td></tr>
2164 <tr><td><tt>uge</tt></td><td>var1 &gt;= var2</td><td>unsigned</td></tr>
2165 <tr><td><tt>true</tt></td><td>always</td><td>--</td></tr>
2166 <tr><td><tt>false</tt></td><td>never</td><td>--</td></tr>
2167 </tbody>
2168</table>
2169
2170<h5>Example:</h5>
2171<pre> &lt;result&gt; = vsetint eq &lt;2 x int&gt; &lt;int 0, int 1&gt;, &lt;int 1, int 0&gt; <i>; yields {&lt;2 x bool&gt;}:result = false, false</i>
2172 &lt;result&gt; = vsetint ne &lt;2 x int&gt; &lt;int 0, int 1&gt;, &lt;int 1, int 0&gt; <i>; yields {&lt;2 x bool&gt;}:result = true, true</i>
2173 &lt;result&gt; = vsetint slt &lt;2 x int&gt; &lt;int 0, int 1&gt;, &lt;int 1, int 0&gt; <i>; yields {&lt;2 x bool&gt;}:result = true, false</i>
2174 &lt;result&gt; = vsetint sgt &lt;2 x int&gt; &lt;int 0, int 1&gt;, &lt;int 1, int 0&gt; <i>; yields {&lt;2 x bool&gt;}:result = false, true</i>
2175 &lt;result&gt; = vsetint sle &lt;2 x int&gt; &lt;int 0, int 1&gt;, &lt;int 1, int 0&gt; <i>; yields {&lt;2 x bool&gt;}:result = true, false</i>
2176 &lt;result&gt; = vsetint sge &lt;2 x int&gt; &lt;int 0, int 1&gt;, &lt;int 1, int 0&gt; <i>; yields {&lt;2 x bool&gt;}:result = false, true</i>
2177</pre>
2178</div>
2179
2180<!-- _______________________________________________________________________ -->
2181<div class="doc_subsubsection"> <a name="i_vsetfp">'<tt>vsetfp</tt>'
2182Instruction</a> </div>
2183<div class="doc_text">
2184<h5>Syntax:</h5>
2185<pre>&lt;result&gt; = vsetfp &lt;op&gt;, &lt;n x &lt;ty&gt;&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields &lt;n x bool&gt;</i>
2186</pre>
2187
2188<h5>Overview:</h5>
2189
2190<p>The '<tt>vsetfp</tt>' instruction takes two floating point vector
2191arguments and returns a vector of boolean values representing, at each
2192position, the result of the comparison between the values at that
2193position in the two operands.</p>
2194
2195<h5>Arguments:</h5>
2196
2197<p>The arguments to a '<tt>vsetfp</tt>' instruction are a comparison
2198operation and two value arguments. The value arguments must be of <a
2199href="t_floating">floating point</a> <a href="#t_packed">packed</a>
2200type, and they must have identical types. The operation argument must
2201be one of <tt>eq</tt>, <tt>ne</tt>, <tt>lt</tt>, <tt>gt</tt>,
2202<tt>le</tt>, <tt>ge</tt>, <tt>oeq</tt>, <tt>one</tt>, <tt>olt</tt>,
2203<tt>ogt</tt>, <tt>ole</tt>, <tt>oge</tt>, <tt>ueq</tt>, <tt>une</tt>,
2204<tt>ult</tt>, <tt>ugt</tt>, <tt>ule</tt>, <tt>uge</tt>, <tt>o</tt>,
2205<tt>u</tt>, <tt>true</tt>, and <tt>false</tt>. The result is a packed
2206<tt>bool</tt> value with the same length as each operand.</p>
2207
2208<h5>Semantics:</h5>
2209
2210<p>The following table shows the semantics of '<tt>vsetfp</tt>' for
2211floating point types. If either operand is a floating point Not a
2212Number (NaN) value, the operation is unordered, and the value in the
2213first column below is produced at that position. Otherwise, the
2214operation is ordered, and the value in the second column is
2215produced.</p>
2216
2217<table border="1" cellspacing="0" cellpadding="4">
2218 <tbody>
2219 <tr><th>Operation</th><th>If unordered<th>Otherwise true iff</th></tr>
2220 <tr><td><tt>eq</tt></td><td>undefined</td><td>var1 == var2</td></tr>
2221 <tr><td><tt>ne</tt></td><td>undefined</td><td>var1 != var2</td></tr>
2222 <tr><td><tt>lt</tt></td><td>undefined</td><td>var1 &lt; var2</td></tr>
2223 <tr><td><tt>gt</tt></td><td>undefined</td><td>var1 &gt; var2</td></tr>
2224 <tr><td><tt>le</tt></td><td>undefined</td><td>var1 &lt;= var2</td></tr>
2225 <tr><td><tt>ge</tt></td><td>undefined</td><td>var1 &gt;= var2</td></tr>
2226 <tr><td><tt>oeq</tt></td><td>false</td><td>var1 == var2</td></tr>
2227 <tr><td><tt>one</tt></td><td>false</td><td>var1 != var2</td></tr>
2228 <tr><td><tt>olt</tt></td><td>false</td><td>var1 &lt; var2</td></tr>
2229 <tr><td><tt>ogt</tt></td><td>false</td><td>var1 &gt; var2</td></tr>
2230 <tr><td><tt>ole</tt></td><td>false</td><td>var1 &lt;= var2</td></tr>
2231 <tr><td><tt>oge</tt></td><td>false</td><td>var1 &gt;= var2</td></tr>
2232 <tr><td><tt>ueq</tt></td><td>true</td><td>var1 == var2</td></tr>
2233 <tr><td><tt>une</tt></td><td>true</td><td>var1 != var2</td></tr>
2234 <tr><td><tt>ult</tt></td><td>true</td><td>var1 &lt; var2</td></tr>
2235 <tr><td><tt>ugt</tt></td><td>true</td><td>var1 &gt; var2</td></tr>
2236 <tr><td><tt>ule</tt></td><td>true</td><td>var1 &lt;= var2</td></tr>
2237 <tr><td><tt>uge</tt></td><td>true</td><td>var1 &gt;= var2</td></tr>
2238 <tr><td><tt>o</tt></td><td>false</td><td>always</td></tr>
2239 <tr><td><tt>u</tt></td><td>true</td><td>never</td></tr>
2240 <tr><td><tt>true</tt></td><td>true</td><td>always</td></tr>
2241 <tr><td><tt>false</tt></td><td>false</td><td>never</td></tr>
2242 </tbody>
2243</table>
2244
2245<h5>Example:</h5>
2246<pre> &lt;result&gt; = vsetfp eq &lt;2 x float&gt; &lt;float 0.0, float 1.0&gt;, &lt;float 1.0, float 0.0&gt; <i>; yields {&lt;2 x bool&gt;}:result = false, false</i>
2247 &lt;result&gt; = vsetfp ne &lt;2 x float&gt; &lt;float 0.0, float 1.0&gt;, &lt;float 1.0, float 0.0&gt; <i>; yields {&lt;2 x bool&gt;}:result = true, true</i>
2248 &lt;result&gt; = vsetfp lt &lt;2 x float&gt; &lt;float 0.0, float 1.0&gt;, &lt;float 1.0, float 0.0&gt; <i>; yields {&lt;2 x bool&gt;}:result = true, false</i>
2249 &lt;result&gt; = vsetfp gt &lt;2 x float&gt; &lt;float 0.0, float 1.0&gt;, &lt;float 1.0, float 0.0&gt; <i>; yields {&lt;2 x bool&gt;}:result = false, true</i>
2250 &lt;result&gt; = vsetfp le &lt;2 x float&gt; &lt;float 0.0, float 1.0&gt;, &lt;float 1.0, float 0.0&gt; <i>; yields {&lt;2 x bool&gt;}:result = true, false</i>
2251 &lt;result&gt; = vsetfp ge &lt;2 x float&gt; &lt;float 0.0, float 1.0&gt;, &lt;float 1.0, float 0.0&gt; <i>; yields {&lt;2 x bool&gt;}:result = false, true</i>
2252</pre>
2253</div>
2254
2255<!-- _______________________________________________________________________ -->
2256<div class="doc_subsubsection">
2257 <a name="i_vselect">'<tt>vselect</tt>' Instruction</a>
2258</div>
2259
2260<div class="doc_text">
2261
2262<h5>Syntax:</h5>
2263
2264<pre>
2265 &lt;result&gt; = vselect &lt;n x bool&gt; &lt;cond&gt;, &lt;n x &lt;ty&gt;&gt; &lt;val1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;val2&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
2266</pre>
2267
2268<h5>Overview:</h5>
2269
2270<p>
2271The '<tt>vselect</tt>' instruction chooses one value at each position
2272of a vector based on a condition.
2273</p>
2274
2275
2276<h5>Arguments:</h5>
2277
2278<p>
2279The '<tt>vselect</tt>' instruction requires a <a
2280href="#t_packed">packed</a> <tt>bool</tt> value indicating the
2281condition at each vector position, and two values of the same packed
2282type. All three operands must have the same length. The type of the
2283result is the same as the type of the two value operands.</p>
2284
2285<h5>Semantics:</h5>
2286
2287<p>
2288At each position where the <tt>bool</tt> vector is true, that position
2289of the result gets its value from the first value argument; otherwise,
2290it gets its value from the second value argument.
2291</p>
2292
2293<h5>Example:</h5>
2294
2295<pre>
2296 %X = vselect bool &lt;2 x bool&gt; &lt;bool true, bool false&gt;, &lt;2 x ubyte&gt; &lt;ubyte 17, ubyte 17&gt;,
2297 &lt;2 x ubyte&gt; &lt;ubyte 42, ubyte 42&gt; <i>; yields &lt;2 x ubyte&gt;:17, 42</i>
2298</pre>
2299</div>
2300
2301
2302
Chris Lattnerce83bff2006-04-08 23:07:04 +00002303<!-- ======================================================================= -->
2304<div class="doc_subsection">
Chris Lattner6ab66722006-08-15 00:45:58 +00002305 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner54611b42005-11-06 08:02:57 +00002306</div>
2307
Misha Brukman76307852003-11-08 01:05:38 +00002308<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +00002309
Chris Lattner48b383b02003-11-25 01:02:51 +00002310<p>A key design point of an SSA-based representation is how it
2311represents memory. In LLVM, no memory locations are in SSA form, which
2312makes things very simple. This section describes how to read, write,
John Criswelldfe6a862004-12-10 15:51:16 +00002313allocate, and free memory in LLVM.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002314
Misha Brukman76307852003-11-08 01:05:38 +00002315</div>
Chris Lattner54611b42005-11-06 08:02:57 +00002316
Chris Lattner2f7c9632001-06-06 20:29:01 +00002317<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +00002318<div class="doc_subsubsection">
2319 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2320</div>
2321
Misha Brukman76307852003-11-08 01:05:38 +00002322<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +00002323
Chris Lattner2f7c9632001-06-06 20:29:01 +00002324<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002325
2326<pre>
2327 &lt;result&gt; = malloc &lt;type&gt;[, uint &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002328</pre>
Chris Lattner54611b42005-11-06 08:02:57 +00002329
Chris Lattner2f7c9632001-06-06 20:29:01 +00002330<h5>Overview:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002331
Chris Lattner48b383b02003-11-25 01:02:51 +00002332<p>The '<tt>malloc</tt>' instruction allocates memory from the system
2333heap and returns a pointer to it.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002334
Chris Lattner2f7c9632001-06-06 20:29:01 +00002335<h5>Arguments:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002336
2337<p>The '<tt>malloc</tt>' instruction allocates
2338<tt>sizeof(&lt;type&gt;)*NumElements</tt>
John Criswella92e5862004-02-24 16:13:56 +00002339bytes of memory from the operating system and returns a pointer of the
Chris Lattner54611b42005-11-06 08:02:57 +00002340appropriate type to the program. If "NumElements" is specified, it is the
2341number of elements allocated. If an alignment is specified, the value result
2342of the allocation is guaranteed to be aligned to at least that boundary. If
2343not specified, or if zero, the target can choose to align the allocation on any
2344convenient boundary.</p>
2345
Misha Brukman76307852003-11-08 01:05:38 +00002346<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002347
Chris Lattner2f7c9632001-06-06 20:29:01 +00002348<h5>Semantics:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002349
Chris Lattner48b383b02003-11-25 01:02:51 +00002350<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2351a pointer is returned.</p>
Misha Brukman76307852003-11-08 01:05:38 +00002352
Chris Lattner54611b42005-11-06 08:02:57 +00002353<h5>Example:</h5>
2354
2355<pre>
2356 %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
2357
2358 %size = <a href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
Chris Lattner590645f2002-04-14 06:13:44 +00002359 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
2360 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
Chris Lattner54611b42005-11-06 08:02:57 +00002361 %array3 = malloc int, uint 4, align 1024 <i>; yields {int*}:array3</i>
2362 %array4 = malloc int, align 1024 <i>; yields {int*}:array4</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002363</pre>
Misha Brukman76307852003-11-08 01:05:38 +00002364</div>
Chris Lattner54611b42005-11-06 08:02:57 +00002365
Chris Lattner2f7c9632001-06-06 20:29:01 +00002366<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +00002367<div class="doc_subsubsection">
2368 <a name="i_free">'<tt>free</tt>' Instruction</a>
2369</div>
2370
Misha Brukman76307852003-11-08 01:05:38 +00002371<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +00002372
Chris Lattner2f7c9632001-06-06 20:29:01 +00002373<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002374
2375<pre>
2376 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002377</pre>
Chris Lattner54611b42005-11-06 08:02:57 +00002378
Chris Lattner2f7c9632001-06-06 20:29:01 +00002379<h5>Overview:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002380
Chris Lattner48b383b02003-11-25 01:02:51 +00002381<p>The '<tt>free</tt>' instruction returns memory back to the unused
John Criswell4a3327e2005-05-13 22:25:59 +00002382memory heap to be reallocated in the future.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002383
Chris Lattner2f7c9632001-06-06 20:29:01 +00002384<h5>Arguments:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002385
Chris Lattner48b383b02003-11-25 01:02:51 +00002386<p>'<tt>value</tt>' shall be a pointer value that points to a value
2387that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2388instruction.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002389
Chris Lattner2f7c9632001-06-06 20:29:01 +00002390<h5>Semantics:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002391
John Criswelldfe6a862004-12-10 15:51:16 +00002392<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner48b383b02003-11-25 01:02:51 +00002393after this instruction executes.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002394
Chris Lattner2f7c9632001-06-06 20:29:01 +00002395<h5>Example:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002396
2397<pre>
2398 %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002399 free [4 x ubyte]* %array
2400</pre>
Misha Brukman76307852003-11-08 01:05:38 +00002401</div>
Chris Lattner54611b42005-11-06 08:02:57 +00002402
Chris Lattner2f7c9632001-06-06 20:29:01 +00002403<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +00002404<div class="doc_subsubsection">
2405 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2406</div>
2407
Misha Brukman76307852003-11-08 01:05:38 +00002408<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +00002409
Chris Lattner2f7c9632001-06-06 20:29:01 +00002410<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002411
2412<pre>
2413 &lt;result&gt; = alloca &lt;type&gt;[, uint &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002414</pre>
Chris Lattner54611b42005-11-06 08:02:57 +00002415
Chris Lattner2f7c9632001-06-06 20:29:01 +00002416<h5>Overview:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002417
Chris Lattner48b383b02003-11-25 01:02:51 +00002418<p>The '<tt>alloca</tt>' instruction allocates memory on the current
2419stack frame of the procedure that is live until the current function
2420returns to its caller.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002421
Chris Lattner2f7c9632001-06-06 20:29:01 +00002422<h5>Arguments:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002423
John Criswelldfe6a862004-12-10 15:51:16 +00002424<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
Chris Lattner48b383b02003-11-25 01:02:51 +00002425bytes of memory on the runtime stack, returning a pointer of the
Chris Lattner54611b42005-11-06 08:02:57 +00002426appropriate type to the program. If "NumElements" is specified, it is the
2427number of elements allocated. If an alignment is specified, the value result
2428of the allocation is guaranteed to be aligned to at least that boundary. If
2429not specified, or if zero, the target can choose to align the allocation on any
2430convenient boundary.</p>
2431
Misha Brukman76307852003-11-08 01:05:38 +00002432<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002433
Chris Lattner2f7c9632001-06-06 20:29:01 +00002434<h5>Semantics:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002435
John Criswell4a3327e2005-05-13 22:25:59 +00002436<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
Chris Lattner48b383b02003-11-25 01:02:51 +00002437memory is automatically released when the function returns. The '<tt>alloca</tt>'
2438instruction is commonly used to represent automatic variables that must
2439have an address available. When the function returns (either with the <tt><a
John Criswellc932bef2005-05-12 16:55:34 +00002440 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Misha Brukman76307852003-11-08 01:05:38 +00002441instructions), the memory is reclaimed.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00002442
Chris Lattner2f7c9632001-06-06 20:29:01 +00002443<h5>Example:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00002444
2445<pre>
2446 %ptr = alloca int <i>; yields {int*}:ptr</i>
Chris Lattner590645f2002-04-14 06:13:44 +00002447 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
Chris Lattner54611b42005-11-06 08:02:57 +00002448 %ptr = alloca int, uint 4, align 1024 <i>; yields {int*}:ptr</i>
2449 %ptr = alloca int, align 1024 <i>; yields {int*}:ptr</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002450</pre>
Misha Brukman76307852003-11-08 01:05:38 +00002451</div>
Chris Lattner54611b42005-11-06 08:02:57 +00002452
Chris Lattner2f7c9632001-06-06 20:29:01 +00002453<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002454<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2455Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00002456<div class="doc_text">
Chris Lattner095735d2002-05-06 03:03:22 +00002457<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002458<pre> &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;<br> &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;<br></pre>
Chris Lattner095735d2002-05-06 03:03:22 +00002459<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00002460<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner095735d2002-05-06 03:03:22 +00002461<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002462<p>The argument to the '<tt>load</tt>' instruction specifies the memory
John Criswell4c0cf7f2005-10-24 16:17:18 +00002463address from which to load. The pointer must point to a <a
Chris Lattner10ee9652004-06-03 22:57:15 +00002464 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
John Criswell4c0cf7f2005-10-24 16:17:18 +00002465marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
Chris Lattner48b383b02003-11-25 01:02:51 +00002466the number or order of execution of this <tt>load</tt> with other
2467volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2468instructions. </p>
Chris Lattner095735d2002-05-06 03:03:22 +00002469<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00002470<p>The location of memory pointed to is loaded.</p>
Chris Lattner095735d2002-05-06 03:03:22 +00002471<h5>Examples:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002472<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
2473 <a
2474 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner095735d2002-05-06 03:03:22 +00002475 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
2476</pre>
Misha Brukman76307852003-11-08 01:05:38 +00002477</div>
Chris Lattner095735d2002-05-06 03:03:22 +00002478<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002479<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2480Instruction</a> </div>
Chris Lattner095735d2002-05-06 03:03:22 +00002481<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002482<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; <i>; yields {void}</i>
Chris Lattner12d456c2003-09-08 18:27:49 +00002483 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; <i>; yields {void}</i>
Chris Lattner095735d2002-05-06 03:03:22 +00002484</pre>
Chris Lattner095735d2002-05-06 03:03:22 +00002485<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00002486<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner095735d2002-05-06 03:03:22 +00002487<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002488<p>There are two arguments to the '<tt>store</tt>' instruction: a value
John Criswell4c0cf7f2005-10-24 16:17:18 +00002489to store and an address in which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner48b383b02003-11-25 01:02:51 +00002490operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
John Criswell4a3327e2005-05-13 22:25:59 +00002491operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
Chris Lattner48b383b02003-11-25 01:02:51 +00002492optimizer is not allowed to modify the number or order of execution of
2493this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2494 href="#i_store">store</a></tt> instructions.</p>
2495<h5>Semantics:</h5>
2496<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2497at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner095735d2002-05-06 03:03:22 +00002498<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002499<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
2500 <a
2501 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner095735d2002-05-06 03:03:22 +00002502 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
2503</pre>
Chris Lattner095735d2002-05-06 03:03:22 +00002504<!-- _______________________________________________________________________ -->
Chris Lattner33fd7022004-04-05 01:30:49 +00002505<div class="doc_subsubsection">
2506 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2507</div>
2508
Misha Brukman76307852003-11-08 01:05:38 +00002509<div class="doc_text">
Chris Lattner590645f2002-04-14 06:13:44 +00002510<h5>Syntax:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00002511<pre>
2512 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2513</pre>
2514
Chris Lattner590645f2002-04-14 06:13:44 +00002515<h5>Overview:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00002516
2517<p>
2518The '<tt>getelementptr</tt>' instruction is used to get the address of a
2519subelement of an aggregate data structure.</p>
2520
Chris Lattner590645f2002-04-14 06:13:44 +00002521<h5>Arguments:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00002522
2523<p>This instruction takes a list of integer constants that indicate what
2524elements of the aggregate object to index to. The actual types of the arguments
2525provided depend on the type of the first pointer argument. The
2526'<tt>getelementptr</tt>' instruction is used to index down through the type
John Criswell88190562005-05-16 16:17:45 +00002527levels of a structure or to a specific index in an array. When indexing into a
2528structure, only <tt>uint</tt>
John Criswell4a3327e2005-05-13 22:25:59 +00002529integer constants are allowed. When indexing into an array or pointer,
Chris Lattner33fd7022004-04-05 01:30:49 +00002530<tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
2531
Chris Lattner48b383b02003-11-25 01:02:51 +00002532<p>For example, let's consider a C code fragment and how it gets
2533compiled to LLVM:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00002534
2535<pre>
2536 struct RT {
2537 char A;
2538 int B[10][20];
2539 char C;
2540 };
2541 struct ST {
2542 int X;
2543 double Y;
2544 struct RT Z;
2545 };
2546
2547 int *foo(struct ST *s) {
2548 return &amp;s[1].Z.B[5][13];
2549 }
2550</pre>
2551
Misha Brukman76307852003-11-08 01:05:38 +00002552<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00002553
2554<pre>
2555 %RT = type { sbyte, [10 x [20 x int]], sbyte }
2556 %ST = type { int, double, %RT }
2557
Brian Gaeke317ef962004-07-02 21:08:14 +00002558 implementation
2559
2560 int* %foo(%ST* %s) {
2561 entry:
2562 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
Chris Lattner33fd7022004-04-05 01:30:49 +00002563 ret int* %reg
2564 }
2565</pre>
2566
Chris Lattner590645f2002-04-14 06:13:44 +00002567<h5>Semantics:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00002568
2569<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
John Criswell4a3327e2005-05-13 22:25:59 +00002570on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
Chris Lattner10ee9652004-06-03 22:57:15 +00002571and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
2572<tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
Chris Lattner33fd7022004-04-05 01:30:49 +00002573types require <tt>uint</tt> <b>constants</b>.</p>
2574
Misha Brukman76307852003-11-08 01:05:38 +00002575<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Chris Lattner33fd7022004-04-05 01:30:49 +00002576type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
2577}</tt>' type, a structure. The second index indexes into the third element of
2578the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
2579sbyte }</tt>' type, another structure. The third index indexes into the second
2580element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
2581array. The two dimensions of the array are subscripted into, yielding an
John Criswell88190562005-05-16 16:17:45 +00002582'<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
Chris Lattner33fd7022004-04-05 01:30:49 +00002583to this element, thus computing a value of '<tt>int*</tt>' type.</p>
2584
Chris Lattner48b383b02003-11-25 01:02:51 +00002585<p>Note that it is perfectly legal to index partially through a
2586structure, returning a pointer to an inner element. Because of this,
2587the LLVM code for the given testcase is equivalent to:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00002588
2589<pre>
Chris Lattner455fc8c2005-03-07 22:13:59 +00002590 int* %foo(%ST* %s) {
Chris Lattner33fd7022004-04-05 01:30:49 +00002591 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
2592 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
2593 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
2594 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
2595 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
2596 ret int* %t5
2597 }
Chris Lattnera8292f32002-05-06 22:08:29 +00002598</pre>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00002599
2600<p>Note that it is undefined to access an array out of bounds: array and
2601pointer indexes must always be within the defined bounds of the array type.
2602The one exception for this rules is zero length arrays. These arrays are
2603defined to be accessible as variable length arrays, which requires access
2604beyond the zero'th element.</p>
2605
Chris Lattner6ab66722006-08-15 00:45:58 +00002606<p>The getelementptr instruction is often confusing. For some more insight
2607into how it works, see <a href="GetElementPtr.html">the getelementptr
2608FAQ</a>.</p>
2609
Chris Lattner590645f2002-04-14 06:13:44 +00002610<h5>Example:</h5>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00002611
Chris Lattner33fd7022004-04-05 01:30:49 +00002612<pre>
2613 <i>; yields [12 x ubyte]*:aptr</i>
2614 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
2615</pre>
2616
2617</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002618<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002619<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00002620<div class="doc_text">
John Criswell417228d2004-04-09 16:48:45 +00002621<p>The instructions in this category are the "miscellaneous"
Chris Lattner48b383b02003-11-25 01:02:51 +00002622instructions, which defy better classification.</p>
Misha Brukman76307852003-11-08 01:05:38 +00002623</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002624<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002625<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
2626Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +00002627<div class="doc_text">
Chris Lattner70de6632001-07-09 00:26:23 +00002628<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002629<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
Chris Lattner70de6632001-07-09 00:26:23 +00002630<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002631<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
2632the SSA graph representing the function.</p>
Chris Lattner70de6632001-07-09 00:26:23 +00002633<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002634<p>The type of the incoming values are specified with the first type
2635field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
2636as arguments, with one pair for each predecessor basic block of the
2637current block. Only values of <a href="#t_firstclass">first class</a>
2638type may be used as the value arguments to the PHI node. Only labels
2639may be used as the label arguments.</p>
2640<p>There must be no non-phi instructions between the start of a basic
2641block and the PHI instructions: i.e. PHI instructions must be first in
2642a basic block.</p>
Chris Lattner70de6632001-07-09 00:26:23 +00002643<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002644<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
2645value specified by the parameter, depending on which basic block we
2646came from in the last <a href="#terminators">terminator</a> instruction.</p>
Chris Lattnera8292f32002-05-06 22:08:29 +00002647<h5>Example:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002648<pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi uint [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add uint %indvar, 1<br> br label %Loop<br></pre>
Misha Brukman76307852003-11-08 01:05:38 +00002649</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002650
Chris Lattnera8292f32002-05-06 22:08:29 +00002651<!-- _______________________________________________________________________ -->
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002652<div class="doc_subsubsection">
2653 <a name="i_cast">'<tt>cast .. to</tt>' Instruction</a>
2654</div>
2655
Misha Brukman76307852003-11-08 01:05:38 +00002656<div class="doc_text">
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002657
Chris Lattnera8292f32002-05-06 22:08:29 +00002658<h5>Syntax:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002659
2660<pre>
2661 &lt;result&gt; = cast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Chris Lattnera8292f32002-05-06 22:08:29 +00002662</pre>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002663
Chris Lattnera8292f32002-05-06 22:08:29 +00002664<h5>Overview:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002665
2666<p>
2667The '<tt>cast</tt>' instruction is used as the primitive means to convert
2668integers to floating point, change data type sizes, and break type safety (by
2669casting pointers).
2670</p>
2671
2672
Chris Lattnera8292f32002-05-06 22:08:29 +00002673<h5>Arguments:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002674
2675<p>
2676The '<tt>cast</tt>' instruction takes a value to cast, which must be a first
2677class value, and a type to cast it to, which must also be a <a
2678href="#t_firstclass">first class</a> type.
2679</p>
2680
Chris Lattnera8292f32002-05-06 22:08:29 +00002681<h5>Semantics:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002682
2683<p>
2684This instruction follows the C rules for explicit casts when determining how the
2685data being cast must change to fit in its new container.
2686</p>
2687
2688<p>
2689When casting to bool, any value that would be considered true in the context of
2690a C '<tt>if</tt>' condition is converted to the boolean '<tt>true</tt>' values,
2691all else are '<tt>false</tt>'.
2692</p>
2693
2694<p>
2695When extending an integral value from a type of one signness to another (for
2696example '<tt>sbyte</tt>' to '<tt>ulong</tt>'), the value is sign-extended if the
2697<b>source</b> value is signed, and zero-extended if the source value is
2698unsigned. <tt>bool</tt> values are always zero extended into either zero or
2699one.
2700</p>
2701
Chris Lattner70de6632001-07-09 00:26:23 +00002702<h5>Example:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002703
2704<pre>
2705 %X = cast int 257 to ubyte <i>; yields ubyte:1</i>
Chris Lattnerd8f8ede2002-06-25 18:03:17 +00002706 %Y = cast int 123 to bool <i>; yields bool:true</i>
Chris Lattner70de6632001-07-09 00:26:23 +00002707</pre>
Misha Brukman76307852003-11-08 01:05:38 +00002708</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002709
2710<!-- _______________________________________________________________________ -->
2711<div class="doc_subsubsection">
2712 <a name="i_select">'<tt>select</tt>' Instruction</a>
2713</div>
2714
2715<div class="doc_text">
2716
2717<h5>Syntax:</h5>
2718
2719<pre>
2720 &lt;result&gt; = select bool &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
2721</pre>
2722
2723<h5>Overview:</h5>
2724
2725<p>
2726The '<tt>select</tt>' instruction is used to choose one value based on a
2727condition, without branching.
2728</p>
2729
2730
2731<h5>Arguments:</h5>
2732
2733<p>
2734The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
2735</p>
2736
2737<h5>Semantics:</h5>
2738
2739<p>
2740If the boolean condition evaluates to true, the instruction returns the first
John Criswell88190562005-05-16 16:17:45 +00002741value argument; otherwise, it returns the second value argument.
Chris Lattnerb53c28d2004-03-12 05:50:16 +00002742</p>
2743
2744<h5>Example:</h5>
2745
2746<pre>
2747 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
2748</pre>
2749</div>
2750
Robert Bocchinof72fdfe2006-01-15 20:48:27 +00002751
2752<!-- _______________________________________________________________________ -->
2753<div class="doc_subsubsection">
Chris Lattnere23c1392005-05-06 05:47:36 +00002754 <a name="i_call">'<tt>call</tt>' Instruction</a>
2755</div>
2756
Misha Brukman76307852003-11-08 01:05:38 +00002757<div class="doc_text">
Chris Lattnere23c1392005-05-06 05:47:36 +00002758
Chris Lattner2f7c9632001-06-06 20:29:01 +00002759<h5>Syntax:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00002760<pre>
Chris Lattner0132aff2005-05-06 22:57:40 +00002761 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt;* &lt;fnptrval&gt;(&lt;param list&gt;)
Chris Lattnere23c1392005-05-06 05:47:36 +00002762</pre>
2763
Chris Lattner2f7c9632001-06-06 20:29:01 +00002764<h5>Overview:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00002765
Misha Brukman76307852003-11-08 01:05:38 +00002766<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00002767
Chris Lattner2f7c9632001-06-06 20:29:01 +00002768<h5>Arguments:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00002769
Misha Brukman76307852003-11-08 01:05:38 +00002770<p>This instruction requires several arguments:</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00002771
Chris Lattnera8292f32002-05-06 22:08:29 +00002772<ol>
Chris Lattner48b383b02003-11-25 01:02:51 +00002773 <li>
Chris Lattner0132aff2005-05-06 22:57:40 +00002774 <p>The optional "tail" marker indicates whether the callee function accesses
2775 any allocas or varargs in the caller. If the "tail" marker is present, the
Chris Lattnere23c1392005-05-06 05:47:36 +00002776 function call is eligible for tail call optimization. Note that calls may
2777 be marked "tail" even if they do not occur before a <a
2778 href="#i_ret"><tt>ret</tt></a> instruction.
Chris Lattner48b383b02003-11-25 01:02:51 +00002779 </li>
2780 <li>
Chris Lattner0132aff2005-05-06 22:57:40 +00002781 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
2782 convention</a> the call should use. If none is specified, the call defaults
2783 to using C calling conventions.
2784 </li>
2785 <li>
Chris Lattnere23c1392005-05-06 05:47:36 +00002786 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
2787 being invoked. The argument types must match the types implied by this
John Criswell88190562005-05-16 16:17:45 +00002788 signature. This type can be omitted if the function is not varargs and
2789 if the function type does not return a pointer to a function.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00002790 </li>
2791 <li>
2792 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
2793 be invoked. In most cases, this is a direct function invocation, but
2794 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
John Criswell88190562005-05-16 16:17:45 +00002795 to function value.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00002796 </li>
2797 <li>
2798 <p>'<tt>function args</tt>': argument list whose types match the
Reid Spencerd845d162005-05-01 22:22:57 +00002799 function signature argument types. All arguments must be of
2800 <a href="#t_firstclass">first class</a> type. If the function signature
2801 indicates the function accepts a variable number of arguments, the extra
2802 arguments can be specified.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +00002803 </li>
Chris Lattnera8292f32002-05-06 22:08:29 +00002804</ol>
Chris Lattnere23c1392005-05-06 05:47:36 +00002805
Chris Lattner2f7c9632001-06-06 20:29:01 +00002806<h5>Semantics:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00002807
Chris Lattner48b383b02003-11-25 01:02:51 +00002808<p>The '<tt>call</tt>' instruction is used to cause control flow to
2809transfer to a specified function, with its incoming arguments bound to
2810the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
2811instruction in the called function, control flow continues with the
2812instruction after the function call, and the return value of the
2813function is bound to the result argument. This is a simpler case of
2814the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00002815
Chris Lattner2f7c9632001-06-06 20:29:01 +00002816<h5>Example:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00002817
2818<pre>
2819 %retval = call int %test(int %argc)
2820 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
2821 %X = tail call int %foo()
Chris Lattner0132aff2005-05-06 22:57:40 +00002822 %Y = tail call <a href="#callingconv">fastcc</a> int %foo()
Chris Lattnere23c1392005-05-06 05:47:36 +00002823</pre>
2824
Misha Brukman76307852003-11-08 01:05:38 +00002825</div>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002826
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002827<!-- _______________________________________________________________________ -->
Chris Lattner6a4a0492004-09-27 21:51:25 +00002828<div class="doc_subsubsection">
Chris Lattner33337472006-01-13 23:26:01 +00002829 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002830</div>
2831
Misha Brukman76307852003-11-08 01:05:38 +00002832<div class="doc_text">
Chris Lattner6a4a0492004-09-27 21:51:25 +00002833
Chris Lattner26ca62e2003-10-18 05:51:36 +00002834<h5>Syntax:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002835
2836<pre>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002837 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattner6a4a0492004-09-27 21:51:25 +00002838</pre>
2839
Chris Lattner26ca62e2003-10-18 05:51:36 +00002840<h5>Overview:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002841
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002842<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Chris Lattner6a4a0492004-09-27 21:51:25 +00002843the "variable argument" area of a function call. It is used to implement the
2844<tt>va_arg</tt> macro in C.</p>
2845
Chris Lattner26ca62e2003-10-18 05:51:36 +00002846<h5>Arguments:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002847
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002848<p>This instruction takes a <tt>va_list*</tt> value and the type of
2849the argument. It returns a value of the specified argument type and
Jeff Cohendc6bfea2005-11-11 02:15:27 +00002850increments the <tt>va_list</tt> to point to the next argument. Again, the
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002851actual type of <tt>va_list</tt> is target specific.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002852
Chris Lattner26ca62e2003-10-18 05:51:36 +00002853<h5>Semantics:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002854
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002855<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
2856type from the specified <tt>va_list</tt> and causes the
2857<tt>va_list</tt> to point to the next argument. For more information,
2858see the variable argument handling <a href="#int_varargs">Intrinsic
2859Functions</a>.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002860
2861<p>It is legal for this instruction to be called in a function which does not
2862take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman76307852003-11-08 01:05:38 +00002863function.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002864
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002865<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
John Criswell88190562005-05-16 16:17:45 +00002866href="#intrinsics">intrinsic function</a> because it takes a type as an
Chris Lattner6a4a0492004-09-27 21:51:25 +00002867argument.</p>
2868
Chris Lattner26ca62e2003-10-18 05:51:36 +00002869<h5>Example:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00002870
2871<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
2872
Misha Brukman76307852003-11-08 01:05:38 +00002873</div>
Chris Lattner941515c2004-01-06 05:31:32 +00002874
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002875<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002876<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
2877<!-- *********************************************************************** -->
Chris Lattner941515c2004-01-06 05:31:32 +00002878
Misha Brukman76307852003-11-08 01:05:38 +00002879<div class="doc_text">
Chris Lattnerfee11462004-02-12 17:01:32 +00002880
2881<p>LLVM supports the notion of an "intrinsic function". These functions have
John Criswell88190562005-05-16 16:17:45 +00002882well known names and semantics and are required to follow certain
Chris Lattnerfee11462004-02-12 17:01:32 +00002883restrictions. Overall, these instructions represent an extension mechanism for
2884the LLVM language that does not require changing all of the transformations in
2885LLVM to add to the language (or the bytecode reader/writer, the parser,
2886etc...).</p>
2887
John Criswell88190562005-05-16 16:17:45 +00002888<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
2889prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
Chris Lattnerfee11462004-02-12 17:01:32 +00002890this. Intrinsic functions must always be external functions: you cannot define
2891the body of intrinsic functions. Intrinsic functions may only be used in call
2892or invoke instructions: it is illegal to take the address of an intrinsic
2893function. Additionally, because intrinsic functions are part of the LLVM
2894language, it is required that they all be documented here if any are added.</p>
2895
2896
John Criswell88190562005-05-16 16:17:45 +00002897<p>To learn how to add an intrinsic function, please see the <a
Chris Lattner90391c12005-05-11 03:35:57 +00002898href="ExtendingLLVM.html">Extending LLVM Guide</a>.
Chris Lattnerfee11462004-02-12 17:01:32 +00002899</p>
2900
Misha Brukman76307852003-11-08 01:05:38 +00002901</div>
Chris Lattner941515c2004-01-06 05:31:32 +00002902
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002903<!-- ======================================================================= -->
Chris Lattner941515c2004-01-06 05:31:32 +00002904<div class="doc_subsection">
2905 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
2906</div>
2907
Misha Brukman76307852003-11-08 01:05:38 +00002908<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +00002909
Misha Brukman76307852003-11-08 01:05:38 +00002910<p>Variable argument support is defined in LLVM with the <a
Chris Lattner33337472006-01-13 23:26:01 +00002911 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
Chris Lattner48b383b02003-11-25 01:02:51 +00002912intrinsic functions. These functions are related to the similarly
2913named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00002914
Chris Lattner48b383b02003-11-25 01:02:51 +00002915<p>All of these functions operate on arguments that use a
2916target-specific value type "<tt>va_list</tt>". The LLVM assembly
2917language reference manual does not define what this type is, so all
2918transformations should be prepared to handle intrinsics with any type
2919used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00002920
Chris Lattner30b868d2006-05-15 17:26:46 +00002921<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Chris Lattner48b383b02003-11-25 01:02:51 +00002922instruction and the variable argument handling intrinsic functions are
2923used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00002924
Chris Lattnerfee11462004-02-12 17:01:32 +00002925<pre>
2926int %test(int %X, ...) {
2927 ; Initialize variable argument processing
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002928 %ap = alloca sbyte*
2929 call void %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
Chris Lattnerfee11462004-02-12 17:01:32 +00002930
2931 ; Read a single integer argument
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002932 %tmp = va_arg sbyte** %ap, int
Chris Lattnerfee11462004-02-12 17:01:32 +00002933
2934 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002935 %aq = alloca sbyte*
Andrew Lenharth5305ea52005-06-22 20:38:11 +00002936 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte** %ap)
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002937 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
Chris Lattnerfee11462004-02-12 17:01:32 +00002938
2939 ; Stop processing of arguments.
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002940 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
Chris Lattnerfee11462004-02-12 17:01:32 +00002941 ret int %tmp
2942}
2943</pre>
Misha Brukman76307852003-11-08 01:05:38 +00002944</div>
Chris Lattner941515c2004-01-06 05:31:32 +00002945
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002946<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +00002947<div class="doc_subsubsection">
2948 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
2949</div>
2950
2951
Misha Brukman76307852003-11-08 01:05:38 +00002952<div class="doc_text">
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002953<h5>Syntax:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002954<pre> declare void %llvm.va_start(&lt;va_list&gt;* &lt;arglist&gt;)<br></pre>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002955<h5>Overview:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002956<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
2957<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
2958href="#i_va_arg">va_arg</a></tt>.</p>
2959
2960<h5>Arguments:</h5>
2961
2962<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
2963
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002964<h5>Semantics:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002965
2966<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
2967macro available in C. In a target-dependent way, it initializes the
2968<tt>va_list</tt> element the argument points to, so that the next call to
2969<tt>va_arg</tt> will produce the first variable argument passed to the function.
2970Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
2971last argument of the function, the compiler can figure that out.</p>
2972
Misha Brukman76307852003-11-08 01:05:38 +00002973</div>
Chris Lattner941515c2004-01-06 05:31:32 +00002974
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002975<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +00002976<div class="doc_subsubsection">
2977 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
2978</div>
2979
Misha Brukman76307852003-11-08 01:05:38 +00002980<div class="doc_text">
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002981<h5>Syntax:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00002982<pre> declare void %llvm.va_end(&lt;va_list*&gt; &lt;arglist&gt;)<br></pre>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002983<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00002984<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>&lt;arglist&gt;</tt>
2985which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
2986or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002987<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00002988<p>The argument is a <tt>va_list</tt> to destroy.</p>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002989<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00002990<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Chris Lattner48b383b02003-11-25 01:02:51 +00002991macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
2992Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
2993 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
2994with calls to <tt>llvm.va_end</tt>.</p>
Misha Brukman76307852003-11-08 01:05:38 +00002995</div>
Chris Lattner941515c2004-01-06 05:31:32 +00002996
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00002997<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +00002998<div class="doc_subsubsection">
2999 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3000</div>
3001
Misha Brukman76307852003-11-08 01:05:38 +00003002<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +00003003
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00003004<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003005
3006<pre>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00003007 declare void %llvm.va_copy(&lt;va_list&gt;* &lt;destarglist&gt;,
Andrew Lenharth5305ea52005-06-22 20:38:11 +00003008 &lt;va_list&gt;* &lt;srcarglist&gt;)
Chris Lattner757528b0b2004-05-23 21:06:01 +00003009</pre>
3010
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00003011<h5>Overview:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003012
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00003013<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
3014the source argument list to the destination argument list.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003015
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00003016<h5>Arguments:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003017
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00003018<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Andrew Lenharth5305ea52005-06-22 20:38:11 +00003019The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00003020
Chris Lattner757528b0b2004-05-23 21:06:01 +00003021
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00003022<h5>Semantics:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003023
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00003024<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
3025available in C. In a target-dependent way, it copies the source
3026<tt>va_list</tt> element into the destination list. This intrinsic is necessary
3027because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
Chris Lattner757528b0b2004-05-23 21:06:01 +00003028arbitrarily complex and require memory allocation, for example.</p>
3029
Misha Brukman76307852003-11-08 01:05:38 +00003030</div>
Chris Lattner941515c2004-01-06 05:31:32 +00003031
Chris Lattnerfee11462004-02-12 17:01:32 +00003032<!-- ======================================================================= -->
3033<div class="doc_subsection">
Chris Lattner757528b0b2004-05-23 21:06:01 +00003034 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3035</div>
3036
3037<div class="doc_text">
3038
3039<p>
3040LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3041Collection</a> requires the implementation and generation of these intrinsics.
3042These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
3043stack</a>, as well as garbage collector implementations that require <a
3044href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
3045Front-ends for type-safe garbage collected languages should generate these
3046intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3047href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3048</p>
3049</div>
3050
3051<!-- _______________________________________________________________________ -->
3052<div class="doc_subsubsection">
3053 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3054</div>
3055
3056<div class="doc_text">
3057
3058<h5>Syntax:</h5>
3059
3060<pre>
Reid Spencer7821d062005-04-26 20:50:44 +00003061 declare void %llvm.gcroot(&lt;ty&gt;** %ptrloc, &lt;ty2&gt;* %metadata)
Chris Lattner757528b0b2004-05-23 21:06:01 +00003062</pre>
3063
3064<h5>Overview:</h5>
3065
John Criswelldfe6a862004-12-10 15:51:16 +00003066<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Chris Lattner757528b0b2004-05-23 21:06:01 +00003067the code generator, and allows some metadata to be associated with it.</p>
3068
3069<h5>Arguments:</h5>
3070
3071<p>The first argument specifies the address of a stack object that contains the
3072root pointer. The second pointer (which must be either a constant or a global
3073value address) contains the meta-data to be associated with the root.</p>
3074
3075<h5>Semantics:</h5>
3076
3077<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3078location. At compile-time, the code generator generates information to allow
3079the runtime to find the pointer at GC safe points.
3080</p>
3081
3082</div>
3083
3084
3085<!-- _______________________________________________________________________ -->
3086<div class="doc_subsubsection">
3087 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3088</div>
3089
3090<div class="doc_text">
3091
3092<h5>Syntax:</h5>
3093
3094<pre>
Chris Lattnerf9228072006-03-14 20:02:51 +00003095 declare sbyte* %llvm.gcread(sbyte* %ObjPtr, sbyte** %Ptr)
Chris Lattner757528b0b2004-05-23 21:06:01 +00003096</pre>
3097
3098<h5>Overview:</h5>
3099
3100<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3101locations, allowing garbage collector implementations that require read
3102barriers.</p>
3103
3104<h5>Arguments:</h5>
3105
Chris Lattnerf9228072006-03-14 20:02:51 +00003106<p>The second argument is the address to read from, which should be an address
3107allocated from the garbage collector. The first object is a pointer to the
3108start of the referenced object, if needed by the language runtime (otherwise
3109null).</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003110
3111<h5>Semantics:</h5>
3112
3113<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3114instruction, but may be replaced with substantially more complex code by the
3115garbage collector runtime, as needed.</p>
3116
3117</div>
3118
3119
3120<!-- _______________________________________________________________________ -->
3121<div class="doc_subsubsection">
3122 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3123</div>
3124
3125<div class="doc_text">
3126
3127<h5>Syntax:</h5>
3128
3129<pre>
Chris Lattnerf9228072006-03-14 20:02:51 +00003130 declare void %llvm.gcwrite(sbyte* %P1, sbyte* %Obj, sbyte** %P2)
Chris Lattner757528b0b2004-05-23 21:06:01 +00003131</pre>
3132
3133<h5>Overview:</h5>
3134
3135<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3136locations, allowing garbage collector implementations that require write
3137barriers (such as generational or reference counting collectors).</p>
3138
3139<h5>Arguments:</h5>
3140
Chris Lattnerf9228072006-03-14 20:02:51 +00003141<p>The first argument is the reference to store, the second is the start of the
3142object to store it to, and the third is the address of the field of Obj to
3143store to. If the runtime does not require a pointer to the object, Obj may be
3144null.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00003145
3146<h5>Semantics:</h5>
3147
3148<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3149instruction, but may be replaced with substantially more complex code by the
3150garbage collector runtime, as needed.</p>
3151
3152</div>
3153
3154
3155
3156<!-- ======================================================================= -->
3157<div class="doc_subsection">
Chris Lattner3649c3a2004-02-14 04:08:35 +00003158 <a name="int_codegen">Code Generator Intrinsics</a>
3159</div>
3160
3161<div class="doc_text">
3162<p>
3163These intrinsics are provided by LLVM to expose special features that may only
3164be implemented with code generator support.
3165</p>
3166
3167</div>
3168
3169<!-- _______________________________________________________________________ -->
3170<div class="doc_subsubsection">
3171 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3172</div>
3173
3174<div class="doc_text">
3175
3176<h5>Syntax:</h5>
3177<pre>
Chris Lattnerb3d430e2006-01-13 01:20:27 +00003178 declare sbyte *%llvm.returnaddress(uint &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00003179</pre>
3180
3181<h5>Overview:</h5>
3182
3183<p>
Chris Lattnerc1fb4262006-10-15 20:05:59 +00003184The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
3185target-specific value indicating the return address of the current function
3186or one of its callers.
Chris Lattner3649c3a2004-02-14 04:08:35 +00003187</p>
3188
3189<h5>Arguments:</h5>
3190
3191<p>
3192The argument to this intrinsic indicates which function to return the address
3193for. Zero indicates the calling function, one indicates its caller, etc. The
3194argument is <b>required</b> to be a constant integer value.
3195</p>
3196
3197<h5>Semantics:</h5>
3198
3199<p>
3200The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3201the return address of the specified call frame, or zero if it cannot be
3202identified. The value returned by this intrinsic is likely to be incorrect or 0
3203for arguments other than zero, so it should only be used for debugging purposes.
3204</p>
3205
3206<p>
3207Note that calling this intrinsic does not prevent function inlining or other
Chris Lattner2e6eb5f2005-03-07 20:30:51 +00003208aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner3649c3a2004-02-14 04:08:35 +00003209source-language caller.
3210</p>
3211</div>
3212
3213
3214<!-- _______________________________________________________________________ -->
3215<div class="doc_subsubsection">
3216 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3217</div>
3218
3219<div class="doc_text">
3220
3221<h5>Syntax:</h5>
3222<pre>
Chris Lattnerb3d430e2006-01-13 01:20:27 +00003223 declare sbyte *%llvm.frameaddress(uint &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00003224</pre>
3225
3226<h5>Overview:</h5>
3227
3228<p>
Chris Lattnerc1fb4262006-10-15 20:05:59 +00003229The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
3230target-specific frame pointer value for the specified stack frame.
Chris Lattner3649c3a2004-02-14 04:08:35 +00003231</p>
3232
3233<h5>Arguments:</h5>
3234
3235<p>
3236The argument to this intrinsic indicates which function to return the frame
3237pointer for. Zero indicates the calling function, one indicates its caller,
3238etc. The argument is <b>required</b> to be a constant integer value.
3239</p>
3240
3241<h5>Semantics:</h5>
3242
3243<p>
3244The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3245the frame address of the specified call frame, or zero if it cannot be
3246identified. The value returned by this intrinsic is likely to be incorrect or 0
3247for arguments other than zero, so it should only be used for debugging purposes.
3248</p>
3249
3250<p>
3251Note that calling this intrinsic does not prevent function inlining or other
Chris Lattner2e6eb5f2005-03-07 20:30:51 +00003252aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner3649c3a2004-02-14 04:08:35 +00003253source-language caller.
3254</p>
3255</div>
3256
Chris Lattnerc8a2c222005-02-28 19:24:19 +00003257<!-- _______________________________________________________________________ -->
3258<div class="doc_subsubsection">
Chris Lattner2f0f0012006-01-13 02:03:13 +00003259 <a name="i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
3260</div>
3261
3262<div class="doc_text">
3263
3264<h5>Syntax:</h5>
3265<pre>
3266 declare sbyte *%llvm.stacksave()
3267</pre>
3268
3269<h5>Overview:</h5>
3270
3271<p>
3272The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
3273the function stack, for use with <a href="#i_stackrestore">
3274<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
3275features like scoped automatic variable sized arrays in C99.
3276</p>
3277
3278<h5>Semantics:</h5>
3279
3280<p>
3281This intrinsic returns a opaque pointer value that can be passed to <a
3282href="#i_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
3283<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
3284<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
3285state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
3286practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
3287that were allocated after the <tt>llvm.stacksave</tt> was executed.
3288</p>
3289
3290</div>
3291
3292<!-- _______________________________________________________________________ -->
3293<div class="doc_subsubsection">
3294 <a name="i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
3295</div>
3296
3297<div class="doc_text">
3298
3299<h5>Syntax:</h5>
3300<pre>
3301 declare void %llvm.stackrestore(sbyte* %ptr)
3302</pre>
3303
3304<h5>Overview:</h5>
3305
3306<p>
3307The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
3308the function stack to the state it was in when the corresponding <a
3309href="#llvm.stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
3310useful for implementing language features like scoped automatic variable sized
3311arrays in C99.
3312</p>
3313
3314<h5>Semantics:</h5>
3315
3316<p>
3317See the description for <a href="#i_stacksave"><tt>llvm.stacksave</tt></a>.
3318</p>
3319
3320</div>
3321
3322
3323<!-- _______________________________________________________________________ -->
3324<div class="doc_subsubsection">
Chris Lattnerc8a2c222005-02-28 19:24:19 +00003325 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
3326</div>
3327
3328<div class="doc_text">
3329
3330<h5>Syntax:</h5>
3331<pre>
Reid Spencer7821d062005-04-26 20:50:44 +00003332 declare void %llvm.prefetch(sbyte * &lt;address&gt;,
3333 uint &lt;rw&gt;, uint &lt;locality&gt;)
Chris Lattnerc8a2c222005-02-28 19:24:19 +00003334</pre>
3335
3336<h5>Overview:</h5>
3337
3338
3339<p>
3340The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
John Criswell88190562005-05-16 16:17:45 +00003341a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
3342no
3343effect on the behavior of the program but can change its performance
Chris Lattnerff851072005-02-28 19:47:14 +00003344characteristics.
Chris Lattnerc8a2c222005-02-28 19:24:19 +00003345</p>
3346
3347<h5>Arguments:</h5>
3348
3349<p>
3350<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
3351determining if the fetch should be for a read (0) or write (1), and
3352<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Chris Lattnerd3e641c2005-03-07 20:31:38 +00003353locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
Chris Lattnerc8a2c222005-02-28 19:24:19 +00003354<tt>locality</tt> arguments must be constant integers.
3355</p>
3356
3357<h5>Semantics:</h5>
3358
3359<p>
3360This intrinsic does not modify the behavior of the program. In particular,
3361prefetches cannot trap and do not produce a value. On targets that support this
3362intrinsic, the prefetch can provide hints to the processor cache for better
3363performance.
3364</p>
3365
3366</div>
3367
Andrew Lenharthb4427912005-03-28 20:05:49 +00003368<!-- _______________________________________________________________________ -->
3369<div class="doc_subsubsection">
3370 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
3371</div>
3372
3373<div class="doc_text">
3374
3375<h5>Syntax:</h5>
3376<pre>
Reid Spencer7821d062005-04-26 20:50:44 +00003377 declare void %llvm.pcmarker( uint &lt;id&gt; )
Andrew Lenharthb4427912005-03-28 20:05:49 +00003378</pre>
3379
3380<h5>Overview:</h5>
3381
3382
3383<p>
John Criswell88190562005-05-16 16:17:45 +00003384The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
3385(PC) in a region of
Andrew Lenharthb4427912005-03-28 20:05:49 +00003386code to simulators and other tools. The method is target specific, but it is
3387expected that the marker will use exported symbols to transmit the PC of the marker.
Jeff Cohendc6bfea2005-11-11 02:15:27 +00003388The marker makes no guarantees that it will remain with any specific instruction
Chris Lattnere64d41d2005-11-15 06:07:55 +00003389after optimizations. It is possible that the presence of a marker will inhibit
Chris Lattnerb40261e2006-03-24 07:16:10 +00003390optimizations. The intended use is to be inserted after optimizations to allow
John Criswell88190562005-05-16 16:17:45 +00003391correlations of simulation runs.
Andrew Lenharthb4427912005-03-28 20:05:49 +00003392</p>
3393
3394<h5>Arguments:</h5>
3395
3396<p>
3397<tt>id</tt> is a numerical id identifying the marker.
3398</p>
3399
3400<h5>Semantics:</h5>
3401
3402<p>
3403This intrinsic does not modify the behavior of the program. Backends that do not
3404support this intrinisic may ignore it.
3405</p>
3406
3407</div>
3408
Andrew Lenharth01aa5632005-11-11 16:47:30 +00003409<!-- _______________________________________________________________________ -->
3410<div class="doc_subsubsection">
3411 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
3412</div>
3413
3414<div class="doc_text">
3415
3416<h5>Syntax:</h5>
3417<pre>
3418 declare ulong %llvm.readcyclecounter( )
3419</pre>
3420
3421<h5>Overview:</h5>
3422
3423
3424<p>
3425The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
3426counter register (or similar low latency, high accuracy clocks) on those targets
3427that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
3428As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
3429should only be used for small timings.
3430</p>
3431
3432<h5>Semantics:</h5>
3433
3434<p>
3435When directly supported, reading the cycle counter should not modify any memory.
3436Implementations are allowed to either return a application specific value or a
3437system wide value. On backends without support, this is lowered to a constant 0.
3438</p>
3439
3440</div>
3441
Chris Lattner3649c3a2004-02-14 04:08:35 +00003442<!-- ======================================================================= -->
3443<div class="doc_subsection">
Chris Lattnerfee11462004-02-12 17:01:32 +00003444 <a name="int_libc">Standard C Library Intrinsics</a>
3445</div>
3446
3447<div class="doc_text">
3448<p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00003449LLVM provides intrinsics for a few important standard C library functions.
3450These intrinsics allow source-language front-ends to pass information about the
3451alignment of the pointer arguments to the code generator, providing opportunity
3452for more efficient code generation.
Chris Lattnerfee11462004-02-12 17:01:32 +00003453</p>
3454
3455</div>
3456
3457<!-- _______________________________________________________________________ -->
3458<div class="doc_subsubsection">
3459 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
3460</div>
3461
3462<div class="doc_text">
3463
3464<h5>Syntax:</h5>
3465<pre>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003466 declare void %llvm.memcpy.i32(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3467 uint &lt;len&gt;, uint &lt;align&gt;)
3468 declare void %llvm.memcpy.i64(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3469 ulong &lt;len&gt;, uint &lt;align&gt;)
Chris Lattnerfee11462004-02-12 17:01:32 +00003470</pre>
3471
3472<h5>Overview:</h5>
3473
3474<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003475The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattnerfee11462004-02-12 17:01:32 +00003476location to the destination location.
3477</p>
3478
3479<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003480Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
3481intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattnerfee11462004-02-12 17:01:32 +00003482</p>
3483
3484<h5>Arguments:</h5>
3485
3486<p>
3487The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner0c8b2592006-03-03 00:07:20 +00003488the source. The third argument is an integer argument
Chris Lattnerfee11462004-02-12 17:01:32 +00003489specifying the number of bytes to copy, and the fourth argument is the alignment
3490of the source and destination locations.
3491</p>
3492
Chris Lattner4c67c482004-02-12 21:18:15 +00003493<p>
3494If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattner5316e5d2006-03-04 00:02:10 +00003495the caller guarantees that both the source and destination pointers are aligned
3496to that boundary.
Chris Lattner4c67c482004-02-12 21:18:15 +00003497</p>
3498
Chris Lattnerfee11462004-02-12 17:01:32 +00003499<h5>Semantics:</h5>
3500
3501<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003502The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattnerfee11462004-02-12 17:01:32 +00003503location to the destination location, which are not allowed to overlap. It
3504copies "len" bytes of memory over. If the argument is known to be aligned to
3505some boundary, this can be specified as the fourth argument, otherwise it should
3506be set to 0 or 1.
3507</p>
3508</div>
3509
3510
Chris Lattnerf30152e2004-02-12 18:10:10 +00003511<!-- _______________________________________________________________________ -->
3512<div class="doc_subsubsection">
3513 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
3514</div>
3515
3516<div class="doc_text">
3517
3518<h5>Syntax:</h5>
3519<pre>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003520 declare void %llvm.memmove.i32(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3521 uint &lt;len&gt;, uint &lt;align&gt;)
3522 declare void %llvm.memmove.i64(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3523 ulong &lt;len&gt;, uint &lt;align&gt;)
Chris Lattnerf30152e2004-02-12 18:10:10 +00003524</pre>
3525
3526<h5>Overview:</h5>
3527
3528<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003529The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
3530location to the destination location. It is similar to the
3531'<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
Chris Lattnerf30152e2004-02-12 18:10:10 +00003532</p>
3533
3534<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003535Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
3536intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattnerf30152e2004-02-12 18:10:10 +00003537</p>
3538
3539<h5>Arguments:</h5>
3540
3541<p>
3542The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner0c8b2592006-03-03 00:07:20 +00003543the source. The third argument is an integer argument
Chris Lattnerf30152e2004-02-12 18:10:10 +00003544specifying the number of bytes to copy, and the fourth argument is the alignment
3545of the source and destination locations.
3546</p>
3547
Chris Lattner4c67c482004-02-12 21:18:15 +00003548<p>
3549If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattner5316e5d2006-03-04 00:02:10 +00003550the caller guarantees that the source and destination pointers are aligned to
3551that boundary.
Chris Lattner4c67c482004-02-12 21:18:15 +00003552</p>
3553
Chris Lattnerf30152e2004-02-12 18:10:10 +00003554<h5>Semantics:</h5>
3555
3556<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003557The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
Chris Lattnerf30152e2004-02-12 18:10:10 +00003558location to the destination location, which may overlap. It
3559copies "len" bytes of memory over. If the argument is known to be aligned to
3560some boundary, this can be specified as the fourth argument, otherwise it should
3561be set to 0 or 1.
3562</p>
3563</div>
3564
Chris Lattner941515c2004-01-06 05:31:32 +00003565
Chris Lattner3649c3a2004-02-14 04:08:35 +00003566<!-- _______________________________________________________________________ -->
3567<div class="doc_subsubsection">
Chris Lattner0c8b2592006-03-03 00:07:20 +00003568 <a name="i_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +00003569</div>
3570
3571<div class="doc_text">
3572
3573<h5>Syntax:</h5>
3574<pre>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003575 declare void %llvm.memset.i32(sbyte* &lt;dest&gt;, ubyte &lt;val&gt;,
3576 uint &lt;len&gt;, uint &lt;align&gt;)
3577 declare void %llvm.memset.i64(sbyte* &lt;dest&gt;, ubyte &lt;val&gt;,
3578 ulong &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00003579</pre>
3580
3581<h5>Overview:</h5>
3582
3583<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003584The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
Chris Lattner3649c3a2004-02-14 04:08:35 +00003585byte value.
3586</p>
3587
3588<p>
3589Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
3590does not return a value, and takes an extra alignment argument.
3591</p>
3592
3593<h5>Arguments:</h5>
3594
3595<p>
3596The first argument is a pointer to the destination to fill, the second is the
Chris Lattner0c8b2592006-03-03 00:07:20 +00003597byte value to fill it with, the third argument is an integer
Chris Lattner3649c3a2004-02-14 04:08:35 +00003598argument specifying the number of bytes to fill, and the fourth argument is the
3599known alignment of destination location.
3600</p>
3601
3602<p>
3603If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattner5316e5d2006-03-04 00:02:10 +00003604the caller guarantees that the destination pointer is aligned to that boundary.
Chris Lattner3649c3a2004-02-14 04:08:35 +00003605</p>
3606
3607<h5>Semantics:</h5>
3608
3609<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +00003610The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
3611the
Chris Lattner3649c3a2004-02-14 04:08:35 +00003612destination location. If the argument is known to be aligned to some boundary,
3613this can be specified as the fourth argument, otherwise it should be set to 0 or
36141.
3615</p>
3616</div>
3617
3618
Chris Lattner3b4f4372004-06-11 02:28:03 +00003619<!-- _______________________________________________________________________ -->
3620<div class="doc_subsubsection">
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003621 <a name="i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a>
Alkis Evlogimenos0fa39232004-06-13 01:16:15 +00003622</div>
3623
3624<div class="doc_text">
3625
3626<h5>Syntax:</h5>
3627<pre>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003628 declare bool %llvm.isunordered.f32(float Val1, float Val2)
3629 declare bool %llvm.isunordered.f64(double Val1, double Val2)
Alkis Evlogimenos0fa39232004-06-13 01:16:15 +00003630</pre>
3631
3632<h5>Overview:</h5>
3633
3634<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003635The '<tt>llvm.isunordered</tt>' intrinsics return true if either or both of the
Alkis Evlogimenos0fa39232004-06-13 01:16:15 +00003636specified floating point values is a NAN.
3637</p>
3638
3639<h5>Arguments:</h5>
3640
3641<p>
3642The arguments are floating point numbers of the same type.
3643</p>
3644
3645<h5>Semantics:</h5>
3646
3647<p>
3648If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
3649false.
3650</p>
3651</div>
3652
3653
Chris Lattner8a8f2e52005-07-21 01:29:16 +00003654<!-- _______________________________________________________________________ -->
3655<div class="doc_subsubsection">
Chris Lattner069b5bd2006-01-16 22:38:59 +00003656 <a name="i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00003657</div>
3658
3659<div class="doc_text">
3660
3661<h5>Syntax:</h5>
3662<pre>
Chris Lattner33b73f92006-09-08 06:34:02 +00003663 declare float %llvm.sqrt.f32(float %Val)
3664 declare double %llvm.sqrt.f64(double %Val)
Chris Lattner8a8f2e52005-07-21 01:29:16 +00003665</pre>
3666
3667<h5>Overview:</h5>
3668
3669<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003670The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Chris Lattner8a8f2e52005-07-21 01:29:16 +00003671returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
3672<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
3673negative numbers (which allows for better optimization).
3674</p>
3675
3676<h5>Arguments:</h5>
3677
3678<p>
3679The argument and return value are floating point numbers of the same type.
3680</p>
3681
3682<h5>Semantics:</h5>
3683
3684<p>
3685This function returns the sqrt of the specified operand if it is a positive
3686floating point number.
3687</p>
3688</div>
3689
Chris Lattner33b73f92006-09-08 06:34:02 +00003690<!-- _______________________________________________________________________ -->
3691<div class="doc_subsubsection">
3692 <a name="i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
3693</div>
3694
3695<div class="doc_text">
3696
3697<h5>Syntax:</h5>
3698<pre>
3699 declare float %llvm.powi.f32(float %Val, int %power)
3700 declare double %llvm.powi.f64(double %Val, int %power)
3701</pre>
3702
3703<h5>Overview:</h5>
3704
3705<p>
3706The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
3707specified (positive or negative) power. The order of evaluation of
3708multiplications is not defined.
3709</p>
3710
3711<h5>Arguments:</h5>
3712
3713<p>
3714The second argument is an integer power, and the first is a value to raise to
3715that power.
3716</p>
3717
3718<h5>Semantics:</h5>
3719
3720<p>
3721This function returns the first value raised to the second power with an
3722unspecified sequence of rounding operations.</p>
3723</div>
3724
3725
Andrew Lenharth1d463522005-05-03 18:01:48 +00003726<!-- ======================================================================= -->
3727<div class="doc_subsection">
Nate Begeman0f223bb2006-01-13 23:26:38 +00003728 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +00003729</div>
3730
3731<div class="doc_text">
3732<p>
Nate Begeman0f223bb2006-01-13 23:26:38 +00003733LLVM provides intrinsics for a few important bit manipulation operations.
Andrew Lenharth1d463522005-05-03 18:01:48 +00003734These allow efficient code generation for some algorithms.
3735</p>
3736
3737</div>
3738
3739<!-- _______________________________________________________________________ -->
3740<div class="doc_subsubsection">
Nate Begeman0f223bb2006-01-13 23:26:38 +00003741 <a name="i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
3742</div>
3743
3744<div class="doc_text">
3745
3746<h5>Syntax:</h5>
3747<pre>
Chris Lattner069b5bd2006-01-16 22:38:59 +00003748 declare ushort %llvm.bswap.i16(ushort &lt;id&gt;)
3749 declare uint %llvm.bswap.i32(uint &lt;id&gt;)
3750 declare ulong %llvm.bswap.i64(ulong &lt;id&gt;)
Nate Begeman0f223bb2006-01-13 23:26:38 +00003751</pre>
3752
3753<h5>Overview:</h5>
3754
3755<p>
3756The '<tt>llvm.bwsap</tt>' family of intrinsics is used to byteswap a 16, 32 or
375764 bit quantity. These are useful for performing operations on data that is not
3758in the target's native byte order.
3759</p>
3760
3761<h5>Semantics:</h5>
3762
3763<p>
Chris Lattner069b5bd2006-01-16 22:38:59 +00003764The <tt>llvm.bswap.16</tt> intrinsic returns a ushort value that has the high and low
3765byte of the input ushort swapped. Similarly, the <tt>llvm.bswap.i32</tt> intrinsic
Nate Begeman0f223bb2006-01-13 23:26:38 +00003766returns a uint value that has the four bytes of the input uint swapped, so that
3767if the input bytes are numbered 0, 1, 2, 3 then the returned uint will have its
Chris Lattner069b5bd2006-01-16 22:38:59 +00003768bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i64</tt> intrinsic extends this concept
Nate Begeman0f223bb2006-01-13 23:26:38 +00003769to 64 bits.
3770</p>
3771
3772</div>
3773
3774<!-- _______________________________________________________________________ -->
3775<div class="doc_subsubsection">
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003776 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +00003777</div>
3778
3779<div class="doc_text">
3780
3781<h5>Syntax:</h5>
3782<pre>
Chris Lattner069b5bd2006-01-16 22:38:59 +00003783 declare ubyte %llvm.ctpop.i8 (ubyte &lt;src&gt;)
3784 declare ushort %llvm.ctpop.i16(ushort &lt;src&gt;)
3785 declare uint %llvm.ctpop.i32(uint &lt;src&gt;)
3786 declare ulong %llvm.ctpop.i64(ulong &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +00003787</pre>
3788
3789<h5>Overview:</h5>
3790
3791<p>
Chris Lattner069b5bd2006-01-16 22:38:59 +00003792The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
3793value.
Andrew Lenharth1d463522005-05-03 18:01:48 +00003794</p>
3795
3796<h5>Arguments:</h5>
3797
3798<p>
Chris Lattner573f64e2005-05-07 01:46:40 +00003799The only argument is the value to be counted. The argument may be of any
Chris Lattner069b5bd2006-01-16 22:38:59 +00003800unsigned integer type. The return type must match the argument type.
Andrew Lenharth1d463522005-05-03 18:01:48 +00003801</p>
3802
3803<h5>Semantics:</h5>
3804
3805<p>
3806The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
3807</p>
3808</div>
3809
3810<!-- _______________________________________________________________________ -->
3811<div class="doc_subsubsection">
Chris Lattnerb748c672006-01-16 22:34:14 +00003812 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +00003813</div>
3814
3815<div class="doc_text">
3816
3817<h5>Syntax:</h5>
3818<pre>
Chris Lattner069b5bd2006-01-16 22:38:59 +00003819 declare ubyte %llvm.ctlz.i8 (ubyte &lt;src&gt;)
3820 declare ushort %llvm.ctlz.i16(ushort &lt;src&gt;)
3821 declare uint %llvm.ctlz.i32(uint &lt;src&gt;)
3822 declare ulong %llvm.ctlz.i64(ulong &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +00003823</pre>
3824
3825<h5>Overview:</h5>
3826
3827<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003828The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
3829leading zeros in a variable.
Andrew Lenharth1d463522005-05-03 18:01:48 +00003830</p>
3831
3832<h5>Arguments:</h5>
3833
3834<p>
Chris Lattner573f64e2005-05-07 01:46:40 +00003835The only argument is the value to be counted. The argument may be of any
Chris Lattner069b5bd2006-01-16 22:38:59 +00003836unsigned integer type. The return type must match the argument type.
Andrew Lenharth1d463522005-05-03 18:01:48 +00003837</p>
3838
3839<h5>Semantics:</h5>
3840
3841<p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00003842The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
3843in a variable. If the src == 0 then the result is the size in bits of the type
Chris Lattner905bd172006-04-21 21:37:40 +00003844of src. For example, <tt>llvm.ctlz(int 2) = 30</tt>.
Andrew Lenharth1d463522005-05-03 18:01:48 +00003845</p>
3846</div>
Chris Lattner3b4f4372004-06-11 02:28:03 +00003847
3848
Chris Lattnerefa20fa2005-05-15 19:39:26 +00003849
3850<!-- _______________________________________________________________________ -->
3851<div class="doc_subsubsection">
Chris Lattnerb748c672006-01-16 22:34:14 +00003852 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00003853</div>
3854
3855<div class="doc_text">
3856
3857<h5>Syntax:</h5>
3858<pre>
Chris Lattner069b5bd2006-01-16 22:38:59 +00003859 declare ubyte %llvm.cttz.i8 (ubyte &lt;src&gt;)
3860 declare ushort %llvm.cttz.i16(ushort &lt;src&gt;)
3861 declare uint %llvm.cttz.i32(uint &lt;src&gt;)
3862 declare ulong %llvm.cttz.i64(ulong &lt;src&gt;)
Chris Lattnerefa20fa2005-05-15 19:39:26 +00003863</pre>
3864
3865<h5>Overview:</h5>
3866
3867<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00003868The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
3869trailing zeros.
Chris Lattnerefa20fa2005-05-15 19:39:26 +00003870</p>
3871
3872<h5>Arguments:</h5>
3873
3874<p>
3875The only argument is the value to be counted. The argument may be of any
Chris Lattner069b5bd2006-01-16 22:38:59 +00003876unsigned integer type. The return type must match the argument type.
Chris Lattnerefa20fa2005-05-15 19:39:26 +00003877</p>
3878
3879<h5>Semantics:</h5>
3880
3881<p>
3882The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
3883in a variable. If the src == 0 then the result is the size in bits of the type
3884of src. For example, <tt>llvm.cttz(2) = 1</tt>.
3885</p>
3886</div>
3887
Chris Lattner941515c2004-01-06 05:31:32 +00003888<!-- ======================================================================= -->
3889<div class="doc_subsection">
3890 <a name="int_debugger">Debugger Intrinsics</a>
3891</div>
3892
3893<div class="doc_text">
3894<p>
3895The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
3896are described in the <a
3897href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
3898Debugging</a> document.
3899</p>
3900</div>
3901
3902
Chris Lattner2f7c9632001-06-06 20:29:01 +00003903<!-- *********************************************************************** -->
Chris Lattner2f7c9632001-06-06 20:29:01 +00003904<hr>
Misha Brukmanc501f552004-03-01 17:47:27 +00003905<address>
3906 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
3907 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
3908 <a href="http://validator.w3.org/check/referer"><img
3909 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
3910
3911 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencerca058542006-03-14 05:39:39 +00003912 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmanc501f552004-03-01 17:47:27 +00003913 Last modified: $Date$
3914</address>
Misha Brukman76307852003-11-08 01:05:38 +00003915</body>
3916</html>