Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 2fd21e6e4e0913c0713f15be427ca634eb41f59e / . / docs / LangRef.html
blob: be92ecc5d1682c390f59427bfb1823797b6f3765 [file] [log] [blame]
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencer3921c742004-08-26 20:44:00 +0000[diff] [blame]6 <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 Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]12
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]13<body>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]14
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]15<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]16<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a></li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]24 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]25 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]26 <li><a href="#functionstructure">Functions</a></li>
27 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]28 </ol>
29 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]30 <li><a href="#typesystem">Type System</a>
31 <ol>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]32 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]33 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]34 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]35 </ol>
36 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]37 <li><a href="#t_derived">Derived Types</a>
38 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]39 <li><a href="#t_array">Array Type</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]40 <li><a href="#t_function">Function Type</a></li>
41 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]42 <li><a href="#t_struct">Structure Type</a></li>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]43 <li><a href="#t_packed">Packed Type</a></li>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]44 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]45 </ol>
46 </li>
47 </ol>
48 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]49 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]50 <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 Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]57 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]58 <li><a href="#othervalues">Other Values</a>
59 <ol>
60 <li><a href="#inlineasm">Inline Assembler Expressions</a>
61 </ol>
62 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]63 <li><a href="#instref">Instruction Reference</a>
64 <ol>
65 <li><a href="#terminators">Terminator Instructions</a>
66 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]67 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
68 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]69 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
70 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]71 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]72 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]73 </ol>
74 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]75 <li><a href="#binaryops">Binary Operations</a>
76 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]77 <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>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]80 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
81 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
82 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]83 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
84 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
85 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]86 <li><a href="#i_setcc">'<tt>set<i>cc</i></tt>' Instructions</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]87 </ol>
88 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]89 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
90 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]91 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]92 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]93 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
94 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
95 <li><a href="#i_shr">'<tt>shr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]96 </ol>
97 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]98 <li><a href="#vectorops">Vector Operations</a>
99 <ol>
100 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
101 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
102 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]103 <li><a href="#i_vsetint">'<tt>vsetint</tt>' Instruction</a></li>
104 <li><a href="#i_vsetfp">'<tt>vsetfp</tt>' Instruction</a></li>
105 <li><a href="#i_vselect">'<tt>vselect</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]106 </ol>
107 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]108 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]109 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]110 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
111 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
112 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]113 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
114 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
115 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]116 </ol>
117 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame^]118 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]119 <ol>
120 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
121 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
122 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
123 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
124 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
125 <li><a href="#i_fp2uint">'<tt>fp2uint .. to</tt>' Instruction</a></li>
126 <li><a href="#i_fp2sint">'<tt>fp2sint .. to</tt>' Instruction</a></li>
127 <li><a href="#i_uint2fp">'<tt>uint2fp .. to</tt>' Instruction</a></li>
128 <li><a href="#i_sint2fp">'<tt>sint2fp .. to</tt>' Instruction</a></li>
129 <li><a href="#i_bitconvert">'<tt>bitconvert .. to</tt>' Instruction</a></li>
130 </ol>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]131 <li><a href="#otherops">Other Operations</a>
132 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]133 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]134 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]135 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]136 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]137 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]138 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]139 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]140 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]141 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]142 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]143 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
144 <ol>
145 <li><a href="#i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
146 <li><a href="#i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
147 <li><a href="#i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
148 </ol>
149 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]150 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
151 <ol>
152 <li><a href="#i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
153 <li><a href="#i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
154 <li><a href="#i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
155 </ol>
156 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]157 <li><a href="#int_codegen">Code Generator Intrinsics</a>
158 <ol>
159 <li><a href="#i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
160 <li><a href="#i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]161 <li><a href="#i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
162 <li><a href="#i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]163 <li><a href="#i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]164 <li><a href="#i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]165 <li><a href="#i_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000[diff] [blame]166 </ol>
167 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]168 <li><a href="#int_libc">Standard C Library Intrinsics</a>
169 <ol>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]170 <li><a href="#i_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
171 <li><a href="#i_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
172 <li><a href="#i_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]173 <li><a href="#i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a></li>
174 <li><a href="#i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]175 <li><a href="#i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]176 </ol>
177 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]178 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]179 <ol>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]180 <li><a href="#i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]181 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
182 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
183 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]184 </ol>
185 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]186 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]187 </ol>
188 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]189</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]190
191<div class="doc_author">
192 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
193 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]194</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]195
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]196<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]197<div class="doc_section"> <a name="abstract">Abstract </a></div>
198<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]199
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]200<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]201<p>This document is a reference manual for the LLVM assembly language.
202LLVM is an SSA based representation that provides type safety,
203low-level operations, flexibility, and the capability of representing
204'all' high-level languages cleanly. It is the common code
205representation used throughout all phases of the LLVM compilation
206strategy.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]207</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]208
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]209<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]210<div class="doc_section"> <a name="introduction">Introduction</a> </div>
211<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]212
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]213<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]214
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]215<p>The LLVM code representation is designed to be used in three
216different forms: as an in-memory compiler IR, as an on-disk bytecode
217representation (suitable for fast loading by a Just-In-Time compiler),
218and as a human readable assembly language representation. This allows
219LLVM to provide a powerful intermediate representation for efficient
220compiler transformations and analysis, while providing a natural means
221to debug and visualize the transformations. The three different forms
222of LLVM are all equivalent. This document describes the human readable
223representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]224
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]225<p>The LLVM representation aims to be light-weight and low-level
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]226while being expressive, typed, and extensible at the same time. It
227aims to be a "universal IR" of sorts, by being at a low enough level
228that high-level ideas may be cleanly mapped to it (similar to how
229microprocessors are "universal IR's", allowing many source languages to
230be mapped to them). By providing type information, LLVM can be used as
231the target of optimizations: for example, through pointer analysis, it
232can be proven that a C automatic variable is never accessed outside of
233the current function... allowing it to be promoted to a simple SSA
234value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]235
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]236</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]237
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]238<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]239<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]240
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]241<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]242
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]243<p>It is important to note that this document describes 'well formed'
244LLVM assembly language. There is a difference between what the parser
245accepts and what is considered 'well formed'. For example, the
246following instruction is syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]247
248<pre>
249 %x = <a href="#i_add">add</a> int 1, %x
250</pre>
251
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]252<p>...because the definition of <tt>%x</tt> does not dominate all of
253its uses. The LLVM infrastructure provides a verification pass that may
254be used to verify that an LLVM module is well formed. This pass is
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]255automatically run by the parser after parsing input assembly and by
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]256the optimizer before it outputs bytecode. The violations pointed out
257by the verifier pass indicate bugs in transformation passes or input to
258the parser.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]259
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]260<!-- Describe the typesetting conventions here. --> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]261
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]262<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]263<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]264<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]265
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]266<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]267
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]268<p>LLVM uses three different forms of identifiers, for different
269purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]270
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]271<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]272 <li>Named values are represented as a string of characters with a '%' prefix.
273 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
274 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
275 Identifiers which require other characters in their names can be surrounded
276 with quotes. In this way, anything except a <tt>"</tt> character can be used
277 in a name.</li>
278
279 <li>Unnamed values are represented as an unsigned numeric value with a '%'
280 prefix. For example, %12, %2, %44.</li>
281
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]282 <li>Constants, which are described in a <a href="#constants">section about
283 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]284</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]285
286<p>LLVM requires that values start with a '%' sign for two reasons: Compilers
287don't need to worry about name clashes with reserved words, and the set of
288reserved words may be expanded in the future without penalty. Additionally,
289unnamed identifiers allow a compiler to quickly come up with a temporary
290variable without having to avoid symbol table conflicts.</p>
291
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]292<p>Reserved words in LLVM are very similar to reserved words in other
293languages. There are keywords for different opcodes ('<tt><a
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]294href="#i_add">add</a></tt>', '<tt><a href="#i_cast">cast</a></tt>', '<tt><a
295href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
296href="#t_void">void</a></tt>', '<tt><a href="#t_uint">uint</a></tt>', etc...),
297and others. These reserved words cannot conflict with variable names, because
298none of them start with a '%' character.</p>
299
300<p>Here is an example of LLVM code to multiply the integer variable
301'<tt>%X</tt>' by 8:</p>
302
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]303<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]304
305<pre>
306 %result = <a href="#i_mul">mul</a> uint %X, 8
307</pre>
308
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]309<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]310
311<pre>
312 %result = <a href="#i_shl">shl</a> uint %X, ubyte 3
313</pre>
314
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]315<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]316
317<pre>
318 <a href="#i_add">add</a> uint %X, %X <i>; yields {uint}:%0</i>
319 <a href="#i_add">add</a> uint %0, %0 <i>; yields {uint}:%1</i>
320 %result = <a href="#i_add">add</a> uint %1, %1
321</pre>
322
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]323<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
324important lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]325
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]326<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]327
328 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
329 line.</li>
330
331 <li>Unnamed temporaries are created when the result of a computation is not
332 assigned to a named value.</li>
333
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]334 <li>Unnamed temporaries are numbered sequentially</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]335
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]336</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]337
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]338<p>...and it also shows a convention that we follow in this document. When
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]339demonstrating instructions, we will follow an instruction with a comment that
340defines the type and name of value produced. Comments are shown in italic
341text.</p>
342
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]343</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]344
345<!-- *********************************************************************** -->
346<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
347<!-- *********************************************************************** -->
348
349<!-- ======================================================================= -->
350<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
351</div>
352
353<div class="doc_text">
354
355<p>LLVM programs are composed of "Module"s, each of which is a
356translation unit of the input programs. Each module consists of
357functions, global variables, and symbol table entries. Modules may be
358combined together with the LLVM linker, which merges function (and
359global variable) definitions, resolves forward declarations, and merges
360symbol table entries. Here is an example of the "hello world" module:</p>
361
362<pre><i>; Declare the string constant as a global constant...</i>
363<a href="#identifiers">%.LC0</a> = <a href="#linkage_internal">internal</a> <a
364 href="#globalvars">constant</a> <a href="#t_array">[13 x sbyte]</a> c"hello world\0A\00" <i>; [13 x sbyte]*</i>
365
366<i>; External declaration of the puts function</i>
367<a href="#functionstructure">declare</a> int %puts(sbyte*) <i>; int(sbyte*)* </i>
368
Chris Lattner81c01f02006-06-13 03:05:47 +0000[diff] [blame]369<i>; Global variable / Function body section separator</i>
370implementation
371
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]372<i>; Definition of main function</i>
373int %main() { <i>; int()* </i>
374 <i>; Convert [13x sbyte]* to sbyte *...</i>
375 %cast210 = <a
376 href="#i_getelementptr">getelementptr</a> [13 x sbyte]* %.LC0, long 0, long 0 <i>; sbyte*</i>
377
378 <i>; Call puts function to write out the string to stdout...</i>
379 <a
380 href="#i_call">call</a> int %puts(sbyte* %cast210) <i>; int</i>
381 <a
382 href="#i_ret">ret</a> int 0<br>}<br></pre>
383
384<p>This example is made up of a <a href="#globalvars">global variable</a>
385named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
386function, and a <a href="#functionstructure">function definition</a>
387for "<tt>main</tt>".</p>
388
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]389<p>In general, a module is made up of a list of global values,
390where both functions and global variables are global values. Global values are
391represented by a pointer to a memory location (in this case, a pointer to an
392array of char, and a pointer to a function), and have one of the following <a
393href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]394
Chris Lattner81c01f02006-06-13 03:05:47 +0000[diff] [blame]395<p>Due to a limitation in the current LLVM assembly parser (it is limited by
396one-token lookahead), modules are split into two pieces by the "implementation"
397keyword. Global variable prototypes and definitions must occur before the
398keyword, and function definitions must occur after it. Function prototypes may
399occur either before or after it. In the future, the implementation keyword may
400become a noop, if the parser gets smarter.</p>
401
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]402</div>
403
404<!-- ======================================================================= -->
405<div class="doc_subsection">
406 <a name="linkage">Linkage Types</a>
407</div>
408
409<div class="doc_text">
410
411<p>
412All Global Variables and Functions have one of the following types of linkage:
413</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]414
415<dl>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]416
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]417 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]418
419 <dd>Global values with internal linkage are only directly accessible by
420 objects in the current module. In particular, linking code into a module with
421 an internal global value may cause the internal to be renamed as necessary to
422 avoid collisions. Because the symbol is internal to the module, all
423 references can be updated. This corresponds to the notion of the
424 '<tt>static</tt>' keyword in C, or the idea of "anonymous namespaces" in C++.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]425 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]426
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]427 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]428
429 <dd>"<tt>linkonce</tt>" linkage is similar to <tt>internal</tt> linkage, with
430 the twist that linking together two modules defining the same
431 <tt>linkonce</tt> globals will cause one of the globals to be discarded. This
432 is typically used to implement inline functions. Unreferenced
433 <tt>linkonce</tt> globals are allowed to be discarded.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]434 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]435
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]436 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]437
438 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
439 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
440 used to implement constructs in C such as "<tt>int X;</tt>" at global scope.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]441 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]442
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]443 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]444
445 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
446 pointer to array type. When two global variables with appending linkage are
447 linked together, the two global arrays are appended together. This is the
448 LLVM, typesafe, equivalent of having the system linker append together
449 "sections" with identical names when .o files are linked.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]450 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]451
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]452 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]453
454 <dd>If none of the above identifiers are used, the global is externally
455 visible, meaning that it participates in linkage and can be used to resolve
456 external symbol references.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]457 </dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]458
459 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
460
461 <dd>"<tt>extern_weak</tt>" TBD
462 </dd>
463
464 <p>
465 The next two types of linkage are targeted for Microsoft Windows platform
466 only. They are designed to support importing (exporting) symbols from (to)
467 DLLs.
468 </p>
469
470 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
471
472 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
473 or variable via a global pointer to a pointer that is set up by the DLL
474 exporting the symbol. On Microsoft Windows targets, the pointer name is
475 formed by combining <code>_imp__</code> and the function or variable name.
476 </dd>
477
478 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
479
480 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
481 pointer to a pointer in a DLL, so that it can be referenced with the
482 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
483 name is formed by combining <code>_imp__</code> and the function or variable
484 name.
485 </dd>
486
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]487</dl>
488
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]489<p><a name="linkage_external">For example, since the "<tt>.LC0</tt>"
490variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
491variable and was linked with this one, one of the two would be renamed,
492preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
493external (i.e., lacking any linkage declarations), they are accessible
494outside of the current module. It is illegal for a function <i>declaration</i>
495to have any linkage type other than "externally visible".</a></p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]496
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]497</div>
498
499<!-- ======================================================================= -->
500<div class="doc_subsection">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]501 <a name="callingconv">Calling Conventions</a>
502</div>
503
504<div class="doc_text">
505
506<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
507and <a href="#i_invoke">invokes</a> can all have an optional calling convention
508specified for the call. The calling convention of any pair of dynamic
509caller/callee must match, or the behavior of the program is undefined. The
510following calling conventions are supported by LLVM, and more may be added in
511the future:</p>
512
513<dl>
514 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
515
516 <dd>This calling convention (the default if no other calling convention is
517 specified) matches the target C calling conventions. This calling convention
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]518 supports varargs function calls and tolerates some mismatch in the declared
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]519 prototype and implemented declaration of the function (as does normal C).
520 </dd>
521
Chris Lattner5710ce92006-05-19 21:15:36 +0000[diff] [blame]522 <dt><b>"<tt>csretcc</tt>" - The C struct return calling convention</b>:</dt>
523
524 <dd>This calling convention matches the target C calling conventions, except
525 that functions with this convention are required to take a pointer as their
526 first argument, and the return type of the function must be void. This is
527 used for C functions that return aggregates by-value. In this case, the
528 function has been transformed to take a pointer to the struct as the first
529 argument to the function. For targets where the ABI specifies specific
530 behavior for structure-return calls, the calling convention can be used to
531 distinguish between struct return functions and other functions that take a
532 pointer to a struct as the first argument.
533 </dd>
534
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]535 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
536
537 <dd>This calling convention attempts to make calls as fast as possible
538 (e.g. by passing things in registers). This calling convention allows the
539 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner8cdc5bc2005-05-06 23:08:23 +0000[diff] [blame]540 without having to conform to an externally specified ABI. Implementations of
541 this convention should allow arbitrary tail call optimization to be supported.
542 This calling convention does not support varargs and requires the prototype of
543 all callees to exactly match the prototype of the function definition.
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]544 </dd>
545
546 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
547
548 <dd>This calling convention attempts to make code in the caller as efficient
549 as possible under the assumption that the call is not commonly executed. As
550 such, these calls often preserve all registers so that the call does not break
551 any live ranges in the caller side. This calling convention does not support
552 varargs and requires the prototype of all callees to exactly match the
553 prototype of the function definition.
554 </dd>
555
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]556 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]557
558 <dd>Any calling convention may be specified by number, allowing
559 target-specific calling conventions to be used. Target specific calling
560 conventions start at 64.
561 </dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]562</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]563
564<p>More calling conventions can be added/defined on an as-needed basis, to
565support pascal conventions or any other well-known target-independent
566convention.</p>
567
568</div>
569
570<!-- ======================================================================= -->
571<div class="doc_subsection">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]572 <a name="globalvars">Global Variables</a>
573</div>
574
575<div class="doc_text">
576
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]577<p>Global variables define regions of memory allocated at compilation time
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]578instead of run-time. Global variables may optionally be initialized, may have
579an explicit section to be placed in, and may
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]580have an optional explicit alignment specified. A
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]581variable may be defined as a global "constant," which indicates that the
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]582contents of the variable will <b>never</b> be modified (enabling better
583optimization, allowing the global data to be placed in the read-only section of
584an executable, etc). Note that variables that need runtime initialization
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]585cannot be marked "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]586
587<p>
588LLVM explicitly allows <em>declarations</em> of global variables to be marked
589constant, even if the final definition of the global is not. This capability
590can be used to enable slightly better optimization of the program, but requires
591the language definition to guarantee that optimizations based on the
592'constantness' are valid for the translation units that do not include the
593definition.
594</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]595
596<p>As SSA values, global variables define pointer values that are in
597scope (i.e. they dominate) all basic blocks in the program. Global
598variables always define a pointer to their "content" type because they
599describe a region of memory, and all memory objects in LLVM are
600accessed through pointers.</p>
601
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]602<p>LLVM allows an explicit section to be specified for globals. If the target
603supports it, it will emit globals to the section specified.</p>
604
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]605<p>An explicit alignment may be specified for a global. If not present, or if
606the alignment is set to zero, the alignment of the global is set by the target
607to whatever it feels convenient. If an explicit alignment is specified, the
608global is forced to have at least that much alignment. All alignments must be
609a power of 2.</p>
610
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]611</div>
612
613
614<!-- ======================================================================= -->
615<div class="doc_subsection">
616 <a name="functionstructure">Functions</a>
617</div>
618
619<div class="doc_text">
620
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]621<p>LLVM function definitions consist of an optional <a href="#linkage">linkage
622type</a>, an optional <a href="#callingconv">calling convention</a>, a return
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]623type, a function name, a (possibly empty) argument list, an optional section,
624an optional alignment, an opening curly brace,
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]625a list of basic blocks, and a closing curly brace. LLVM function declarations
626are defined with the "<tt>declare</tt>" keyword, an optional <a
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]627href="#callingconv">calling convention</a>, a return type, a function name,
628a possibly empty list of arguments, and an optional alignment.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]629
630<p>A function definition contains a list of basic blocks, forming the CFG for
631the function. Each basic block may optionally start with a label (giving the
632basic block a symbol table entry), contains a list of instructions, and ends
633with a <a href="#terminators">terminator</a> instruction (such as a branch or
634function return).</p>
635
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]636<p>The first basic block in a program is special in two ways: it is immediately
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]637executed on entrance to the function, and it is not allowed to have predecessor
638basic blocks (i.e. there can not be any branches to the entry block of a
639function). Because the block can have no predecessors, it also cannot have any
640<a href="#i_phi">PHI nodes</a>.</p>
641
642<p>LLVM functions are identified by their name and type signature. Hence, two
643functions with the same name but different parameter lists or return values are
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]644considered different functions, and LLVM will resolve references to each
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]645appropriately.</p>
646
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]647<p>LLVM allows an explicit section to be specified for functions. If the target
648supports it, it will emit functions to the section specified.</p>
649
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]650<p>An explicit alignment may be specified for a function. If not present, or if
651the alignment is set to zero, the alignment of the function is set by the target
652to whatever it feels convenient. If an explicit alignment is specified, the
653function is forced to have at least that much alignment. All alignments must be
654a power of 2.</p>
655
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]656</div>
657
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]658<!-- ======================================================================= -->
659<div class="doc_subsection">
Chris Lattner1eeeb0c2006-04-08 04:40:53 +0000[diff] [blame]660 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]661</div>
662
663<div class="doc_text">
664<p>
665Modules may contain "module-level inline asm" blocks, which corresponds to the
666GCC "file scope inline asm" blocks. These blocks are internally concatenated by
667LLVM and treated as a single unit, but may be separated in the .ll file if
668desired. The syntax is very simple:
669</p>
670
671<div class="doc_code"><pre>
Chris Lattner52599e12006-01-24 00:37:20 +0000[diff] [blame]672 module asm "inline asm code goes here"
673 module asm "more can go here"
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]674</pre></div>
675
676<p>The strings can contain any character by escaping non-printable characters.
677 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
678 for the number.
679</p>
680
681<p>
682 The inline asm code is simply printed to the machine code .s file when
683 assembly code is generated.
684</p>
685</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]686
687
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]688<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]689<div class="doc_section"> <a name="typesystem">Type System</a> </div>
690<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]691
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]692<div class="doc_text">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]693
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]694<p>The LLVM type system is one of the most important features of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]695intermediate representation. Being typed enables a number of
696optimizations to be performed on the IR directly, without having to do
697extra analyses on the side before the transformation. A strong type
698system makes it easier to read the generated code and enables novel
699analyses and transformations that are not feasible to perform on normal
700three address code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]701
702</div>
703
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]704<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]705<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]706<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +0000[diff] [blame]707<p>The primitive types are the fundamental building blocks of the LLVM
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]708system. The current set of primitive types is as follows:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]709
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]710<table class="layout">
711 <tr class="layout">
712 <td class="left">
713 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]714 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]715 <tr><th>Type</th><th>Description</th></tr>
716 <tr><td><tt>void</tt></td><td>No value</td></tr>
Misha Brukmancfa87bc2005-04-22 18:02:52 +0000[diff] [blame]717 <tr><td><tt>ubyte</tt></td><td>Unsigned 8-bit value</td></tr>
718 <tr><td><tt>ushort</tt></td><td>Unsigned 16-bit value</td></tr>
719 <tr><td><tt>uint</tt></td><td>Unsigned 32-bit value</td></tr>
720 <tr><td><tt>ulong</tt></td><td>Unsigned 64-bit value</td></tr>
721 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]722 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]723 </tbody>
724 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]725 </td>
726 <td class="right">
727 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]728 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]729 <tr><th>Type</th><th>Description</th></tr>
730 <tr><td><tt>bool</tt></td><td>True or False value</td></tr>
Misha Brukmancfa87bc2005-04-22 18:02:52 +0000[diff] [blame]731 <tr><td><tt>sbyte</tt></td><td>Signed 8-bit value</td></tr>
732 <tr><td><tt>short</tt></td><td>Signed 16-bit value</td></tr>
733 <tr><td><tt>int</tt></td><td>Signed 32-bit value</td></tr>
734 <tr><td><tt>long</tt></td><td>Signed 64-bit value</td></tr>
735 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]736 </tbody>
737 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]738 </td>
739 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]740</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]741</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]742
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]743<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]744<div class="doc_subsubsection"> <a name="t_classifications">Type
745Classifications</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]746<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]747<p>These different primitive types fall into a few useful
748classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]749
750<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]751 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]752 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]753 <tr>
754 <td><a name="t_signed">signed</a></td>
755 <td><tt>sbyte, short, int, long, float, double</tt></td>
756 </tr>
757 <tr>
758 <td><a name="t_unsigned">unsigned</a></td>
759 <td><tt>ubyte, ushort, uint, ulong</tt></td>
760 </tr>
761 <tr>
762 <td><a name="t_integer">integer</a></td>
763 <td><tt>ubyte, sbyte, ushort, short, uint, int, ulong, long</tt></td>
764 </tr>
765 <tr>
766 <td><a name="t_integral">integral</a></td>
Misha Brukmanc24b7582004-08-12 20:16:08 +0000[diff] [blame]767 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long</tt>
768 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]769 </tr>
770 <tr>
771 <td><a name="t_floating">floating point</a></td>
772 <td><tt>float, double</tt></td>
773 </tr>
774 <tr>
775 <td><a name="t_firstclass">first class</a></td>
Misha Brukmanc24b7582004-08-12 20:16:08 +0000[diff] [blame]776 <td><tt>bool, ubyte, sbyte, ushort, short, uint, int, ulong, long,<br>
777 float, double, <a href="#t_pointer">pointer</a>,
778 <a href="#t_packed">packed</a></tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]779 </tr>
780 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]781</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]782
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]783<p>The <a href="#t_firstclass">first class</a> types are perhaps the
784most important. Values of these types are the only ones which can be
785produced by instructions, passed as arguments, or used as operands to
786instructions. This means that all structures and arrays must be
787manipulated either by pointer or by component.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]788</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]789
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]790<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]791<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]792
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]793<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]794
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]795<p>The real power in LLVM comes from the derived types in the system.
796This is what allows a programmer to represent arrays, functions,
797pointers, and other useful types. Note that these derived types may be
798recursive: For example, it is possible to have a two dimensional array.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]799
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]800</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]801
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]802<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]803<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]804
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]805<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]806
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]807<h5>Overview:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]808
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]809<p>The array type is a very simple derived type that arranges elements
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]810sequentially in memory. The array type requires a size (number of
811elements) and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]812
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]813<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]814
815<pre>
816 [&lt;# elements&gt; x &lt;elementtype&gt;]
817</pre>
818
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]819<p>The number of elements is a constant integer value; elementtype may
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]820be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]821
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]822<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]823<table class="layout">
824 <tr class="layout">
825 <td class="left">
826 <tt>[40 x int ]</tt><br/>
827 <tt>[41 x int ]</tt><br/>
828 <tt>[40 x uint]</tt><br/>
829 </td>
830 <td class="left">
831 Array of 40 integer values.<br/>
832 Array of 41 integer values.<br/>
833 Array of 40 unsigned integer values.<br/>
834 </td>
835 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]836</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]837<p>Here are some examples of multidimensional arrays:</p>
838<table class="layout">
839 <tr class="layout">
840 <td class="left">
841 <tt>[3 x [4 x int]]</tt><br/>
842 <tt>[12 x [10 x float]]</tt><br/>
843 <tt>[2 x [3 x [4 x uint]]]</tt><br/>
844 </td>
845 <td class="left">
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]846 3x4 array of integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]847 12x10 array of single precision floating point values.<br/>
848 2x3x4 array of unsigned integer values.<br/>
849 </td>
850 </tr>
851</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]852
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]853<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
854length array. Normally, accesses past the end of an array are undefined in
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]855LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
856As a special case, however, zero length arrays are recognized to be variable
857length. This allows implementation of 'pascal style arrays' with the LLVM
858type "{ int, [0 x float]}", for example.</p>
859
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]860</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]861
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]862<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]863<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]864<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]865<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]866<p>The function type can be thought of as a function signature. It
867consists of a return type and a list of formal parameter types.
John Criswell009900b2003-11-25 21:45:46 +0000[diff] [blame]868Function types are usually used to build virtual function tables
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]869(which are structures of pointers to functions), for indirect function
870calls, and when defining a function.</p>
John Criswell009900b2003-11-25 21:45:46 +0000[diff] [blame]871<p>
872The return type of a function type cannot be an aggregate type.
873</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]874<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]875<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]876<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Misha Brukmanc24b7582004-08-12 20:16:08 +0000[diff] [blame]877specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]878which indicates that the function takes a variable number of arguments.
879Variable argument functions can access their arguments with the <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]880 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]881<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]882<table class="layout">
883 <tr class="layout">
884 <td class="left">
885 <tt>int (int)</tt> <br/>
886 <tt>float (int, int *) *</tt><br/>
887 <tt>int (sbyte *, ...)</tt><br/>
888 </td>
889 <td class="left">
890 function taking an <tt>int</tt>, returning an <tt>int</tt><br/>
891 <a href="#t_pointer">Pointer</a> to a function that takes an
Misha Brukmanc24b7582004-08-12 20:16:08 +0000[diff] [blame]892 <tt>int</tt> and a <a href="#t_pointer">pointer</a> to <tt>int</tt>,
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]893 returning <tt>float</tt>.<br/>
894 A vararg function that takes at least one <a href="#t_pointer">pointer</a>
895 to <tt>sbyte</tt> (signed char in C), which returns an integer. This is
896 the signature for <tt>printf</tt> in LLVM.<br/>
897 </td>
898 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]899</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]900
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]901</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]902<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]903<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]904<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]905<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]906<p>The structure type is used to represent a collection of data members
907together in memory. The packing of the field types is defined to match
908the ABI of the underlying processor. The elements of a structure may
909be any type that has a size.</p>
910<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
911and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
912field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
913instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]914<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]915<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]916<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]917<table class="layout">
918 <tr class="layout">
919 <td class="left">
920 <tt>{ int, int, int }</tt><br/>
921 <tt>{ float, int (int) * }</tt><br/>
922 </td>
923 <td class="left">
924 a triple of three <tt>int</tt> values<br/>
925 A pair, where the first element is a <tt>float</tt> and the second element
926 is a <a href="#t_pointer">pointer</a> to a <a href="#t_function">function</a>
927 that takes an <tt>int</tt>, returning an <tt>int</tt>.<br/>
928 </td>
929 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]930</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]931</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]932
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]933<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]934<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]935<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]936<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]937<p>As in many languages, the pointer type represents a pointer or
938reference to another object, which must live in memory.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]939<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]940<pre> &lt;type&gt; *<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]941<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]942<table class="layout">
943 <tr class="layout">
944 <td class="left">
945 <tt>[4x int]*</tt><br/>
946 <tt>int (int *) *</tt><br/>
947 </td>
948 <td class="left">
949 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
950 four <tt>int</tt> values<br/>
951 A <a href="#t_pointer">pointer</a> to a <a
Chris Lattnera977c482005-02-19 02:22:14 +0000[diff] [blame]952 href="#t_function">function</a> that takes an <tt>int*</tt>, returning an
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]953 <tt>int</tt>.<br/>
954 </td>
955 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]956</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]957</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]958
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]959<!-- _______________________________________________________________________ -->
960<div class="doc_subsubsection"> <a name="t_packed">Packed Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]961<div class="doc_text">
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]962
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]963<h5>Overview:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]964
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]965<p>A packed type is a simple derived type that represents a vector
966of elements. Packed types are used when multiple primitive data
967are operated in parallel using a single instruction (SIMD).
968A packed type requires a size (number of
Chris Lattnerb8d172f2005-11-10 01:44:22 +0000[diff] [blame]969elements) and an underlying primitive data type. Vectors must have a power
970of two length (1, 2, 4, 8, 16 ...). Packed types are
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]971considered <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]972
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]973<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]974
975<pre>
976 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
977</pre>
978
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]979<p>The number of elements is a constant integer value; elementtype may
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]980be any integral or floating point type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]981
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]982<h5>Examples:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]983
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]984<table class="layout">
985 <tr class="layout">
986 <td class="left">
987 <tt>&lt;4 x int&gt;</tt><br/>
988 <tt>&lt;8 x float&gt;</tt><br/>
989 <tt>&lt;2 x uint&gt;</tt><br/>
990 </td>
991 <td class="left">
992 Packed vector of 4 integer values.<br/>
993 Packed vector of 8 floating-point values.<br/>
994 Packed vector of 2 unsigned integer values.<br/>
995 </td>
996 </tr>
997</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]998</div>
999
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1000<!-- _______________________________________________________________________ -->
1001<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1002<div class="doc_text">
1003
1004<h5>Overview:</h5>
1005
1006<p>Opaque types are used to represent unknown types in the system. This
1007corresponds (for example) to the C notion of a foward declared structure type.
1008In LLVM, opaque types can eventually be resolved to any type (not just a
1009structure type).</p>
1010
1011<h5>Syntax:</h5>
1012
1013<pre>
1014 opaque
1015</pre>
1016
1017<h5>Examples:</h5>
1018
1019<table class="layout">
1020 <tr class="layout">
1021 <td class="left">
1022 <tt>opaque</tt>
1023 </td>
1024 <td class="left">
1025 An opaque type.<br/>
1026 </td>
1027 </tr>
1028</table>
1029</div>
1030
1031
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1032<!-- *********************************************************************** -->
1033<div class="doc_section"> <a name="constants">Constants</a> </div>
1034<!-- *********************************************************************** -->
1035
1036<div class="doc_text">
1037
1038<p>LLVM has several different basic types of constants. This section describes
1039them all and their syntax.</p>
1040
1041</div>
1042
1043<!-- ======================================================================= -->
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1044<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1045
1046<div class="doc_text">
1047
1048<dl>
1049 <dt><b>Boolean constants</b></dt>
1050
1051 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1052 constants of the <tt><a href="#t_primitive">bool</a></tt> type.
1053 </dd>
1054
1055 <dt><b>Integer constants</b></dt>
1056
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1057 <dd>Standard integers (such as '4') are constants of the <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1058 href="#t_integer">integer</a> type. Negative numbers may be used with signed
1059 integer types.
1060 </dd>
1061
1062 <dt><b>Floating point constants</b></dt>
1063
1064 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1065 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1066 notation (see below). Floating point constants must have a <a
1067 href="#t_floating">floating point</a> type. </dd>
1068
1069 <dt><b>Null pointer constants</b></dt>
1070
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1071 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1072 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1073
1074</dl>
1075
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1076<p>The one non-intuitive notation for constants is the optional hexadecimal form
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1077of floating point constants. For example, the form '<tt>double
10780x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
10794.5e+15</tt>'. The only time hexadecimal floating point constants are required
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1080(and the only time that they are generated by the disassembler) is when a
1081floating point constant must be emitted but it cannot be represented as a
1082decimal floating point number. For example, NaN's, infinities, and other
1083special values are represented in their IEEE hexadecimal format so that
1084assembly and disassembly do not cause any bits to change in the constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1085
1086</div>
1087
1088<!-- ======================================================================= -->
1089<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1090</div>
1091
1092<div class="doc_text">
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]1093<p>Aggregate constants arise from aggregation of simple constants
1094and smaller aggregate constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1095
1096<dl>
1097 <dt><b>Structure constants</b></dt>
1098
1099 <dd>Structure constants are represented with notation similar to structure
1100 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]1101 (<tt>{}</tt>)). For example: "<tt>{ int 4, float 17.0, int* %G }</tt>",
1102 where "<tt>%G</tt>" is declared as "<tt>%G = external global int</tt>". Structure constants
1103 must have <a href="#t_struct">structure type</a>, and the number and
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1104 types of elements must match those specified by the type.
1105 </dd>
1106
1107 <dt><b>Array constants</b></dt>
1108
1109 <dd>Array constants are represented with notation similar to array type
1110 definitions (a comma separated list of elements, surrounded by square brackets
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1111 (<tt>[]</tt>)). For example: "<tt>[ int 42, int 11, int 74 ]</tt>". Array
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1112 constants must have <a href="#t_array">array type</a>, and the number and
1113 types of elements must match those specified by the type.
1114 </dd>
1115
1116 <dt><b>Packed constants</b></dt>
1117
1118 <dd>Packed constants are represented with notation similar to packed type
1119 definitions (a comma separated list of elements, surrounded by
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1120 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; int 42,
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1121 int 11, int 74, int 100 &gt;</tt>". Packed constants must have <a
1122 href="#t_packed">packed type</a>, and the number and types of elements must
1123 match those specified by the type.
1124 </dd>
1125
1126 <dt><b>Zero initialization</b></dt>
1127
1128 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1129 value to zero of <em>any</em> type, including scalar and aggregate types.
1130 This is often used to avoid having to print large zero initializers (e.g. for
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1131 large arrays) and is always exactly equivalent to using explicit zero
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1132 initializers.
1133 </dd>
1134</dl>
1135
1136</div>
1137
1138<!-- ======================================================================= -->
1139<div class="doc_subsection">
1140 <a name="globalconstants">Global Variable and Function Addresses</a>
1141</div>
1142
1143<div class="doc_text">
1144
1145<p>The addresses of <a href="#globalvars">global variables</a> and <a
1146href="#functionstructure">functions</a> are always implicitly valid (link-time)
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1147constants. These constants are explicitly referenced when the <a
1148href="#identifiers">identifier for the global</a> is used and always have <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1149href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1150file:</p>
1151
1152<pre>
1153 %X = global int 17
1154 %Y = global int 42
1155 %Z = global [2 x int*] [ int* %X, int* %Y ]
1156</pre>
1157
1158</div>
1159
1160<!-- ======================================================================= -->
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1161<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1162<div class="doc_text">
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1163 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1164 no specific value. Undefined values may be of any type and be used anywhere
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1165 a constant is permitted.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1166
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1167 <p>Undefined values indicate to the compiler that the program is well defined
1168 no matter what value is used, giving the compiler more freedom to optimize.
1169 </p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1170</div>
1171
1172<!-- ======================================================================= -->
1173<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1174</div>
1175
1176<div class="doc_text">
1177
1178<p>Constant expressions are used to allow expressions involving other constants
1179to be used as constants. Constant expressions may be of any <a
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1180href="#t_firstclass">first class</a> type and may involve any LLVM operation
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1181that does not have side effects (e.g. load and call are not supported). The
1182following is the syntax for constant expressions:</p>
1183
1184<dl>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1185 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1186 <dd>Truncate a constant to another type. The bit size of CST must be larger
1187 than the bit size of TYPE. Both types must be integral.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1188
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1189 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1190 <dd>Zero extend a constant to another type. The bit size of CST must be
1191 smaller or equal to the bit size of TYPE. Both types must be integral.</dd>
1192
1193 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1194 <dd>Sign extend a constant to another type. The bit size of CST must be
1195 smaller or equal to the bit size of TYPE. Both types must be integral.</dd>
1196
1197 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1198 <dd>Truncate a floating point constant to another floating point type. The
1199 size of CST must be larger than the size of TYPE. Both types must be
1200 floating point.</dd>
1201
1202 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1203 <dd>Floating point extend a constant to another type. The size of CST must be
1204 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1205
1206 <dt><b><tt>fp2uint ( CST to TYPE )</tt></b></dt>
1207 <dd>Convert a floating point constant to the corresponding unsigned integer
1208 constant. TYPE must be an integer type. CST must be floating point. If the
1209 value won't fit in the integer type, the results are undefined.</dd>
1210
1211 <dt><b><tt>fp2sint ( CST to TYPE )</tt></b></dt>
1212 <dd>Convert a floating point constant to the corresponding signed integer
1213 constant. TYPE must be an integer type. CST must be floating point. If the
1214 value won't fit in the integer type, the results are undefined.</dd>
1215
1216 <dt><b><tt>uint2fp ( CST to TYPE )</tt></b></dt>
1217 <dd>Convert an unsigned integer constant to the corresponding floating point
1218 constant. TYPE must be floating point. CST must be of integer type. If the
1219 value won't fit in the floating point type, the results are undefined.</dd>
1220
1221 <dt><b><tt>sint2fp ( CST to TYPE )</tt></b></dt>
1222 <dd>Convert a signed integer constant to the corresponding floating point
1223 constant. TYPE must be floating point. CST must be of integer type. If the
1224 value won't fit in the floating point type, the results are undefined.</dd>
1225
1226 <dt><b><tt>bitconvert ( CST to TYPE )</tt></b></dt>
1227 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1228 identical (same number of bits). The conversion is done as if the CST value
1229 was stored to memory and read back as TYPE. In other words, no bits change
1230 with this operator, just the type. This can be used for conversion of pointer
1231 and packed types to any other type, as long as they have the same bit width.
1232 </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1233
1234 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1235
1236 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1237 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1238 instruction, the index list may have zero or more indexes, which are required
1239 to make sense for the type of "CSTPTR".</dd>
1240
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]1241 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1242
1243 <dd>Perform the <a href="#i_select">select operation</a> on
1244 constants.
1245
1246 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1247
1248 <dd>Perform the <a href="#i_extractelement">extractelement
1249 operation</a> on constants.
1250
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]1251 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1252
1253 <dd>Perform the <a href="#i_insertelement">insertelement
1254 operation</a> on constants.
1255
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]1256
1257 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1258
1259 <dd>Perform the <a href="#i_shufflevector">shufflevector
1260 operation</a> on constants.
1261
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1262 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1263
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1264 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1265 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1266 binary</a> operations. The constraints on operands are the same as those for
1267 the corresponding instruction (e.g. no bitwise operations on floating point
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1268 values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1269</dl>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1270</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000[diff] [blame]1271
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1272<!-- *********************************************************************** -->
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1273<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1274<!-- *********************************************************************** -->
1275
1276<!-- ======================================================================= -->
1277<div class="doc_subsection">
1278<a name="inlineasm">Inline Assembler Expressions</a>
1279</div>
1280
1281<div class="doc_text">
1282
1283<p>
1284LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1285Module-Level Inline Assembly</a>) through the use of a special value. This
1286value represents the inline assembler as a string (containing the instructions
1287to emit), a list of operand constraints (stored as a string), and a flag that
1288indicates whether or not the inline asm expression has side effects. An example
1289inline assembler expression is:
1290</p>
1291
1292<pre>
1293 int(int) asm "bswap $0", "=r,r"
1294</pre>
1295
1296<p>
1297Inline assembler expressions may <b>only</b> be used as the callee operand of
1298a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1299</p>
1300
1301<pre>
1302 %X = call int asm "<a href="#i_bswap">bswap</a> $0", "=r,r"(int %Y)
1303</pre>
1304
1305<p>
1306Inline asms with side effects not visible in the constraint list must be marked
1307as having side effects. This is done through the use of the
1308'<tt>sideeffect</tt>' keyword, like so:
1309</p>
1310
1311<pre>
1312 call void asm sideeffect "eieio", ""()
1313</pre>
1314
1315<p>TODO: The format of the asm and constraints string still need to be
1316documented here. Constraints on what can be done (e.g. duplication, moving, etc
1317need to be documented).
1318</p>
1319
1320</div>
1321
1322<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1323<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1324<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1325
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1326<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1327
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1328<p>The LLVM instruction set consists of several different
1329classifications of instructions: <a href="#terminators">terminator
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1330instructions</a>, <a href="#binaryops">binary instructions</a>,
1331<a href="#bitwiseops">bitwise binary instructions</a>, <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1332 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1333instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1334
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1335</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1336
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1337<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1338<div class="doc_subsection"> <a name="terminators">Terminator
1339Instructions</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1340
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1341<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1342
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1343<p>As mentioned <a href="#functionstructure">previously</a>, every
1344basic block in a program ends with a "Terminator" instruction, which
1345indicates which block should be executed after the current block is
1346finished. These terminator instructions typically yield a '<tt>void</tt>'
1347value: they produce control flow, not values (the one exception being
1348the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1349<p>There are six different terminator instructions: the '<a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1350 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1351instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1352the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1353 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1354 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1355
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1356</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1357
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1358<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1359<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1360Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1361<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1362<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1363<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1364 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1365</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1366<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1367<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1368value) from a function back to the caller.</p>
John Criswell4457dc92004-04-09 16:48:45 +0000[diff] [blame]1369<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1370returns a value and then causes control flow, and one that just causes
1371control flow to occur.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1372<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1373<p>The '<tt>ret</tt>' instruction may return any '<a
1374 href="#t_firstclass">first class</a>' type. Notice that a function is
1375not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1376instruction inside of the function that returns a value that does not
1377match the return type of the function.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1378<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1379<p>When the '<tt>ret</tt>' instruction is executed, control flow
1380returns back to the calling function's context. If the caller is a "<a
John Criswellfa081872004-06-25 15:16:57 +0000[diff] [blame]1381 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1382the instruction after the call. If the caller was an "<a
1383 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1384at the beginning of the "normal" destination block. If the instruction
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1385returns a value, that value shall set the call or invoke instruction's
1386return value.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1387<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1388<pre> ret int 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1389 ret void <i>; Return from a void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1390</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1391</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1392<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1393<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1394<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1395<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1396<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 Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1397</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1398<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1399<p>The '<tt>br</tt>' instruction is used to cause control flow to
1400transfer to a different basic block in the current function. There are
1401two forms of this instruction, corresponding to a conditional branch
1402and an unconditional branch.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1403<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1404<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
1405single '<tt>bool</tt>' value and two '<tt>label</tt>' values. The
1406unconditional form of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>'
1407value as a target.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1408<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1409<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>bool</tt>'
1410argument is evaluated. If the value is <tt>true</tt>, control flows
1411to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1412control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1413<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1414<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
1415 href="#i_ret">ret</a> int 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> int 0<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1416</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1417<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1418<div class="doc_subsubsection">
1419 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1420</div>
1421
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1422<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1423<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1424
1425<pre>
1426 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1427</pre>
1428
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1429<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1430
1431<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1432several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1433instruction, allowing a branch to occur to one of many possible
1434destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1435
1436
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1437<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1438
1439<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1440comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1441an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1442table is not allowed to contain duplicate constant entries.</p>
1443
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1444<h5>Semantics:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1445
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1446<p>The <tt>switch</tt> instruction specifies a table of values and
1447destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswell84114752004-06-25 16:05:06 +0000[diff] [blame]1448table is searched for the given value. If the value is found, control flow is
1449transfered to the corresponding destination; otherwise, control flow is
1450transfered to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1451
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1452<h5>Implementation:</h5>
1453
1454<p>Depending on properties of the target machine and the particular
1455<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswell84114752004-06-25 16:05:06 +0000[diff] [blame]1456ways. For example, it could be generated as a series of chained conditional
1457branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1458
1459<h5>Example:</h5>
1460
1461<pre>
1462 <i>; Emulate a conditional br instruction</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1463 %Val = <a href="#i_zext">zext</a> bool %value to int
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1464 switch int %Val, label %truedest [int 0, label %falsedest ]
1465
1466 <i>; Emulate an unconditional br instruction</i>
1467 switch uint 0, label %dest [ ]
1468
1469 <i>; Implement a jump table:</i>
1470 switch uint %val, label %otherwise [ uint 0, label %onzero
1471 uint 1, label %onone
1472 uint 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1473</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1474</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1475
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1476<!-- _______________________________________________________________________ -->
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1477<div class="doc_subsubsection">
1478 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1479</div>
1480
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1481<div class="doc_text">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1482
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1483<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1484
1485<pre>
1486 &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 Lattner76b8a332006-05-14 18:23:06 +0000[diff] [blame]1487 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1488</pre>
1489
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]1490<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1491
1492<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1493function, with the possibility of control flow transfer to either the
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1494'<tt>normal</tt>' label or the
1495'<tt>exception</tt>' label. If the callee function returns with the
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1496"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1497"normal" label. If the callee (or any indirect callees) returns with the "<a
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1498href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1499continued at the dynamically nearest "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1500
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1501<h5>Arguments:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1502
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1503<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1504
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1505<ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1506 <li>
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1507 The optional "cconv" marker indicates which <a href="callingconv">calling
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1508 convention</a> the call should use. If none is specified, the call defaults
1509 to using C calling conventions.
1510 </li>
1511 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1512 function value being invoked. In most cases, this is a direct function
1513 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1514 an arbitrary pointer to function value.
1515 </li>
1516
1517 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1518 function to be invoked. </li>
1519
1520 <li>'<tt>function args</tt>': argument list whose types match the function
1521 signature argument types. If the function signature indicates the function
1522 accepts a variable number of arguments, the extra arguments can be
1523 specified. </li>
1524
1525 <li>'<tt>normal label</tt>': the label reached when the called function
1526 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1527
1528 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1529 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1530
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1531</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1532
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1533<h5>Semantics:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1534
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1535<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1536href="#i_call">call</a></tt>' instruction in most regards. The primary
1537difference is that it establishes an association with a label, which is used by
1538the runtime library to unwind the stack.</p>
1539
1540<p>This instruction is used in languages with destructors to ensure that proper
1541cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1542exception. Additionally, this is important for implementation of
1543'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1544
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1545<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1546<pre>
1547 %retval = invoke int %Test(int 15) to label %Continue
Chris Lattner76b8a332006-05-14 18:23:06 +0000[diff] [blame]1548 unwind label %TestCleanup <i>; {int}:retval set</i>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1549 %retval = invoke <a href="#callingconv">coldcc</a> int %Test(int 15) to label %Continue
Chris Lattner76b8a332006-05-14 18:23:06 +0000[diff] [blame]1550 unwind label %TestCleanup <i>; {int}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1551</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1552</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1553
1554
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1555<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1556
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1557<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1558Instruction</a> </div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1559
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1560<div class="doc_text">
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1561
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1562<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1563<pre>
1564 unwind
1565</pre>
1566
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1567<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1568
1569<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1570at the first callee in the dynamic call stack which used an <a
1571href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1572primarily used to implement exception handling.</p>
1573
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1574<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1575
1576<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1577immediately halt. The dynamic call stack is then searched for the first <a
1578href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1579execution continues at the "exceptional" destination block specified by the
1580<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1581dynamic call chain, undefined behavior results.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1582</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1583
1584<!-- _______________________________________________________________________ -->
1585
1586<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1587Instruction</a> </div>
1588
1589<div class="doc_text">
1590
1591<h5>Syntax:</h5>
1592<pre>
1593 unreachable
1594</pre>
1595
1596<h5>Overview:</h5>
1597
1598<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1599instruction is used to inform the optimizer that a particular portion of the
1600code is not reachable. This can be used to indicate that the code after a
1601no-return function cannot be reached, and other facts.</p>
1602
1603<h5>Semantics:</h5>
1604
1605<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1606</div>
1607
1608
1609
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1610<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1611<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1612<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1613<p>Binary operators are used to do most of the computation in a
1614program. They require two operands, execute an operation on them, and
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1615produce a single value. The operands might represent
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1616multiple data, as is the case with the <a href="#t_packed">packed</a> data type.
1617The result value of a binary operator is not
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1618necessarily the same type as its operands.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1619<p>There are several different binary operators:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1620</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1621<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1622<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1623Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1624<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1625<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1626<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1627</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1628<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1629<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1630<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1631<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1632 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
1633 This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1634Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1635<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1636<p>The value produced is the integer or floating point sum of the two
1637operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1638<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1639<pre> &lt;result&gt; = add int 4, %var <i>; yields {int}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1640</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1641</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1642<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1643<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1644Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1645<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1646<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1647<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1648</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1649<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1650<p>The '<tt>sub</tt>' instruction returns the difference of its two
1651operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1652<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1653instruction present in most other intermediate representations.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1654<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1655<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1656 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1657values.
1658This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1659Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1660<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1661<p>The value produced is the integer or floating point difference of
1662the two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1663<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1664<pre> &lt;result&gt; = sub int 4, %var <i>; yields {int}:result = 4 - %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1665 &lt;result&gt; = sub int 0, %val <i>; yields {int}:result = -%var</i>
1666</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1667</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1668<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1669<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1670Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1671<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1672<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1673<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1674</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1675<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1676<p>The '<tt>mul</tt>' instruction returns the product of its two
1677operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1678<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1679<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1680 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1681values.
1682This instruction can also take <a href="#t_packed">packed</a> versions of the values.
1683Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1684<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1685<p>The value produced is the integer or floating point product of the
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1686two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1687<p>There is no signed vs unsigned multiplication. The appropriate
1688action is taken based on the type of the operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1689<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1690<pre> &lt;result&gt; = mul int 4, %var <i>; yields {int}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1691</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1692</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1693<!-- _______________________________________________________________________ -->
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1694<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
1695</a></div>
1696<div class="doc_text">
1697<h5>Syntax:</h5>
1698<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1699</pre>
1700<h5>Overview:</h5>
1701<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
1702operands.</p>
1703<h5>Arguments:</h5>
1704<p>The two arguments to the '<tt>udiv</tt>' instruction must be
1705<a href="#t_integer">integer</a> values. Both arguments must have identical
1706types. This instruction can also take <a href="#t_packed">packed</a> versions
1707of the values in which case the elements must be integers.</p>
1708<h5>Semantics:</h5>
1709<p>The value produced is the unsigned integer quotient of the two operands. This
1710instruction always performs an unsigned division operation, regardless of
1711whether the arguments are unsigned or not.</p>
1712<h5>Example:</h5>
1713<pre> &lt;result&gt; = udiv uint 4, %var <i>; yields {uint}:result = 4 / %var</i>
1714</pre>
1715</div>
1716<!-- _______________________________________________________________________ -->
1717<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
1718</a> </div>
1719<div class="doc_text">
1720<h5>Syntax:</h5>
1721<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1722</pre>
1723<h5>Overview:</h5>
1724<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
1725operands.</p>
1726<h5>Arguments:</h5>
1727<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
1728<a href="#t_integer">integer</a> values. Both arguments must have identical
1729types. This instruction can also take <a href="#t_packed">packed</a> versions
1730of the values in which case the elements must be integers.</p>
1731<h5>Semantics:</h5>
1732<p>The value produced is the signed integer quotient of the two operands. This
1733instruction always performs a signed division operation, regardless of whether
1734the arguments are signed or not.</p>
1735<h5>Example:</h5>
1736<pre> &lt;result&gt; = sdiv int 4, %var <i>; yields {int}:result = 4 / %var</i>
1737</pre>
1738</div>
1739<!-- _______________________________________________________________________ -->
1740<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1741Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1742<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1743<h5>Syntax:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1744<pre> &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1745</pre>
1746<h5>Overview:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1747<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1748operands.</p>
1749<h5>Arguments:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1750<p>The two arguments to the '<tt>div</tt>' instruction must be
1751<a href="#t_floating">floating point</a> values. Both arguments must have
1752identical types. This instruction can also take <a href="#t_packed">packed</a>
1753versions of the values in which case the elements must be floating point.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1754<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1755<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1756<h5>Example:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]1757<pre> &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1758</pre>
1759</div>
1760<!-- _______________________________________________________________________ -->
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1761<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
1762</div>
1763<div class="doc_text">
1764<h5>Syntax:</h5>
1765<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1766</pre>
1767<h5>Overview:</h5>
1768<p>The '<tt>urem</tt>' instruction returns the remainder from the
1769unsigned division of its two arguments.</p>
1770<h5>Arguments:</h5>
1771<p>The two arguments to the '<tt>urem</tt>' instruction must be
1772<a href="#t_integer">integer</a> values. Both arguments must have identical
1773types.</p>
1774<h5>Semantics:</h5>
1775<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
1776This instruction always performs an unsigned division to get the remainder,
1777regardless of whether the arguments are unsigned or not.</p>
1778<h5>Example:</h5>
1779<pre> &lt;result&gt; = urem uint 4, %var <i>; yields {uint}:result = 4 % %var</i>
1780</pre>
1781
1782</div>
1783<!-- _______________________________________________________________________ -->
1784<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1785Instruction</a> </div>
1786<div class="doc_text">
1787<h5>Syntax:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1788<pre> &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1789</pre>
1790<h5>Overview:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1791<p>The '<tt>srem</tt>' instruction returns the remainder from the
1792signed division of its two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1793<h5>Arguments:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1794<p>The two arguments to the '<tt>srem</tt>' instruction must be
1795<a href="#t_integer">integer</a> values. Both arguments must have identical
1796types.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1797<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1798<p>This instruction returns the <i>remainder</i> of a division (where the result
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1799has the same sign as the divisor), not the <i>modulus</i> (where the
1800result has the same sign as the dividend) of a value. For more
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1801information about the difference, see <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1802 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
1803Math Forum</a>.</p>
1804<h5>Example:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]1805<pre> &lt;result&gt; = srem int 4, %var <i>; yields {int}:result = 4 % %var</i>
1806</pre>
1807
1808</div>
1809<!-- _______________________________________________________________________ -->
1810<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
1811Instruction</a> </div>
1812<div class="doc_text">
1813<h5>Syntax:</h5>
1814<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
1815</pre>
1816<h5>Overview:</h5>
1817<p>The '<tt>frem</tt>' instruction returns the remainder from the
1818division of its two operands.</p>
1819<h5>Arguments:</h5>
1820<p>The two arguments to the '<tt>frem</tt>' instruction must be
1821<a href="#t_floating">floating point</a> values. Both arguments must have
1822identical types.</p>
1823<h5>Semantics:</h5>
1824<p>This instruction returns the <i>remainder</i> of a division.</p>
1825<h5>Example:</h5>
1826<pre> &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1827</pre>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]1828
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1829</div>
1830<!-- _______________________________________________________________________ -->
1831<div class="doc_subsubsection"> <a name="i_setcc">'<tt>set<i>cc</i></tt>'
1832Instructions</a> </div>
1833<div class="doc_text">
1834<h5>Syntax:</h5>
1835<pre> &lt;result&gt; = seteq &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1836 &lt;result&gt; = setne &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1837 &lt;result&gt; = setlt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1838 &lt;result&gt; = setgt &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1839 &lt;result&gt; = setle &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1840 &lt;result&gt; = setge &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {bool}:result</i>
1841</pre>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1842<h5>Overview:</h5>
1843<p>The '<tt>set<i>cc</i></tt>' family of instructions returns a boolean
1844value based on a comparison of their two operands.</p>
1845<h5>Arguments:</h5>
1846<p>The two arguments to the '<tt>set<i>cc</i></tt>' instructions must
1847be of <a href="#t_firstclass">first class</a> type (it is not possible
1848to compare '<tt>label</tt>'s, '<tt>array</tt>'s, '<tt>structure</tt>'
1849or '<tt>void</tt>' values, etc...). Both arguments must have identical
1850types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1851<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1852<p>The '<tt>seteq</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1853value if both operands are equal.<br>
1854The '<tt>setne</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1855value if both operands are unequal.<br>
1856The '<tt>setlt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1857value if the first operand is less than the second operand.<br>
1858The '<tt>setgt</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1859value if the first operand is greater than the second operand.<br>
1860The '<tt>setle</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1861value if the first operand is less than or equal to the second operand.<br>
1862The '<tt>setge</tt>' instruction yields a <tt>true</tt> '<tt>bool</tt>'
1863value if the first operand is greater than or equal to the second
1864operand.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1865<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1866<pre> &lt;result&gt; = seteq int 4, 5 <i>; yields {bool}:result = false</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1867 &lt;result&gt; = setne float 4, 5 <i>; yields {bool}:result = true</i>
1868 &lt;result&gt; = setlt uint 4, 5 <i>; yields {bool}:result = true</i>
1869 &lt;result&gt; = setgt sbyte 4, 5 <i>; yields {bool}:result = false</i>
1870 &lt;result&gt; = setle sbyte 4, 5 <i>; yields {bool}:result = true</i>
1871 &lt;result&gt; = setge sbyte 4, 5 <i>; yields {bool}:result = false</i>
1872</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1873</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]1874
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1875<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1876<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
1877Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1878<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1879<p>Bitwise binary operators are used to do various forms of
1880bit-twiddling in a program. They are generally very efficient
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1881instructions and can commonly be strength reduced from other
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1882instructions. They require two operands, execute an operation on them,
1883and produce a single value. The resulting value of the bitwise binary
1884operators is always the same type as its first operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1885</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1886<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1887<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
1888Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1889<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1890<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1891<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1892</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1893<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1894<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
1895its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1896<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1897<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1898 href="#t_integral">integral</a> values. Both arguments must have
1899identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1900<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1901<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1902<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1903<div style="align: center">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1904<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1905 <tbody>
1906 <tr>
1907 <td>In0</td>
1908 <td>In1</td>
1909 <td>Out</td>
1910 </tr>
1911 <tr>
1912 <td>0</td>
1913 <td>0</td>
1914 <td>0</td>
1915 </tr>
1916 <tr>
1917 <td>0</td>
1918 <td>1</td>
1919 <td>0</td>
1920 </tr>
1921 <tr>
1922 <td>1</td>
1923 <td>0</td>
1924 <td>0</td>
1925 </tr>
1926 <tr>
1927 <td>1</td>
1928 <td>1</td>
1929 <td>1</td>
1930 </tr>
1931 </tbody>
1932</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1933</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1934<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1935<pre> &lt;result&gt; = and int 4, %var <i>; yields {int}:result = 4 &amp; %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1936 &lt;result&gt; = and int 15, 40 <i>; yields {int}:result = 8</i>
1937 &lt;result&gt; = and int 4, 8 <i>; yields {int}:result = 0</i>
1938</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1939</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1940<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1941<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1942<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1943<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1944<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1945</pre>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1946<h5>Overview:</h5>
1947<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
1948or of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1949<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1950<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1951 href="#t_integral">integral</a> values. Both arguments must have
1952identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1953<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1954<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1955<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1956<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1957<table border="1" cellspacing="0" cellpadding="4">
1958 <tbody>
1959 <tr>
1960 <td>In0</td>
1961 <td>In1</td>
1962 <td>Out</td>
1963 </tr>
1964 <tr>
1965 <td>0</td>
1966 <td>0</td>
1967 <td>0</td>
1968 </tr>
1969 <tr>
1970 <td>0</td>
1971 <td>1</td>
1972 <td>1</td>
1973 </tr>
1974 <tr>
1975 <td>1</td>
1976 <td>0</td>
1977 <td>1</td>
1978 </tr>
1979 <tr>
1980 <td>1</td>
1981 <td>1</td>
1982 <td>1</td>
1983 </tr>
1984 </tbody>
1985</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1986</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1987<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1988<pre> &lt;result&gt; = or int 4, %var <i>; yields {int}:result = 4 | %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1989 &lt;result&gt; = or int 15, 40 <i>; yields {int}:result = 47</i>
1990 &lt;result&gt; = or int 4, 8 <i>; yields {int}:result = 12</i>
1991</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1992</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1993<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1994<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
1995Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1996<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1997<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1998<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1999</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2000<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2001<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2002or of its two operands. The <tt>xor</tt> is used to implement the
2003"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2004<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2005<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2006 href="#t_integral">integral</a> values. Both arguments must have
2007identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2008<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2009<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2010<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2011<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2012<table border="1" cellspacing="0" cellpadding="4">
2013 <tbody>
2014 <tr>
2015 <td>In0</td>
2016 <td>In1</td>
2017 <td>Out</td>
2018 </tr>
2019 <tr>
2020 <td>0</td>
2021 <td>0</td>
2022 <td>0</td>
2023 </tr>
2024 <tr>
2025 <td>0</td>
2026 <td>1</td>
2027 <td>1</td>
2028 </tr>
2029 <tr>
2030 <td>1</td>
2031 <td>0</td>
2032 <td>1</td>
2033 </tr>
2034 <tr>
2035 <td>1</td>
2036 <td>1</td>
2037 <td>0</td>
2038 </tr>
2039 </tbody>
2040</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2041</div>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2042<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2043<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2044<pre> &lt;result&gt; = xor int 4, %var <i>; yields {int}:result = 4 ^ %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2045 &lt;result&gt; = xor int 15, 40 <i>; yields {int}:result = 39</i>
2046 &lt;result&gt; = xor int 4, 8 <i>; yields {int}:result = 12</i>
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]2047 &lt;result&gt; = xor int %V, -1 <i>; yields {int}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2048</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2049</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2050<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2051<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2052Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2053<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2054<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2055<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, ubyte &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2056</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2057<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2058<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2059the left a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2060<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2061<p>The first argument to the '<tt>shl</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2062 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
2063type.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2064<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2065<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2066<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2067<pre> &lt;result&gt; = shl int 4, ubyte %var <i>; yields {int}:result = 4 &lt;&lt; %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2068 &lt;result&gt; = shl int 4, ubyte 2 <i>; yields {int}:result = 16</i>
2069 &lt;result&gt; = shl int 1, ubyte 10 <i>; yields {int}:result = 1024</i>
2070</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2071</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2072<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2073<div class="doc_subsubsection"> <a name="i_shr">'<tt>shr</tt>'
2074Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2075<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2076<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2077<pre> &lt;result&gt; = shr &lt;ty&gt; &lt;var1&gt;, ubyte &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2078</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2079<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2080<p>The '<tt>shr</tt>' instruction returns the first operand shifted to
2081the right a specified number of bits.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2082<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2083<p>The first argument to the '<tt>shr</tt>' instruction must be an <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2084 href="#t_integer">integer</a> type. The second argument must be an '<tt>ubyte</tt>'
2085type.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2086<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2087<p>If the first argument is a <a href="#t_signed">signed</a> type, the
2088most significant bit is duplicated in the newly free'd bit positions.
2089If the first argument is unsigned, zero bits shall fill the empty
2090positions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2091<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2092<pre> &lt;result&gt; = shr int 4, ubyte %var <i>; yields {int}:result = 4 &gt;&gt; %var</i>
Chris Lattner8c6bb902003-06-18 21:30:51 +0000[diff] [blame]2093 &lt;result&gt; = shr uint 4, ubyte 1 <i>; yields {uint}:result = 2</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2094 &lt;result&gt; = shr int 4, ubyte 2 <i>; yields {int}:result = 1</i>
Chris Lattner8c6bb902003-06-18 21:30:51 +0000[diff] [blame]2095 &lt;result&gt; = shr sbyte 4, ubyte 3 <i>; yields {sbyte}:result = 0</i>
2096 &lt;result&gt; = shr sbyte -2, ubyte 1 <i>; yields {sbyte}:result = -1</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2097</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2098</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2099
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2100<!-- ======================================================================= -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2101<div class="doc_subsection">
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2102 <a name="vectorops">Vector Operations</a>
2103</div>
2104
2105<div class="doc_text">
2106
2107<p>LLVM supports several instructions to represent vector operations in a
2108target-independent manner. This instructions cover the element-access and
2109vector-specific operations needed to process vectors effectively. While LLVM
2110does directly support these vector operations, many sophisticated algorithms
2111will want to use target-specific intrinsics to take full advantage of a specific
2112target.</p>
2113
2114</div>
2115
2116<!-- _______________________________________________________________________ -->
2117<div class="doc_subsubsection">
2118 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2119</div>
2120
2121<div class="doc_text">
2122
2123<h5>Syntax:</h5>
2124
2125<pre>
2126 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, uint &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
2127</pre>
2128
2129<h5>Overview:</h5>
2130
2131<p>
2132The '<tt>extractelement</tt>' instruction extracts a single scalar
2133element from a packed vector at a specified index.
2134</p>
2135
2136
2137<h5>Arguments:</h5>
2138
2139<p>
2140The first operand of an '<tt>extractelement</tt>' instruction is a
2141value of <a href="#t_packed">packed</a> type. The second operand is
2142an index indicating the position from which to extract the element.
2143The index may be a variable.</p>
2144
2145<h5>Semantics:</h5>
2146
2147<p>
2148The result is a scalar of the same type as the element type of
2149<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2150<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2151results are undefined.
2152</p>
2153
2154<h5>Example:</h5>
2155
2156<pre>
2157 %result = extractelement &lt;4 x int&gt; %vec, uint 0 <i>; yields int</i>
2158</pre>
2159</div>
2160
2161
2162<!-- _______________________________________________________________________ -->
2163<div class="doc_subsubsection">
2164 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2165</div>
2166
2167<div class="doc_text">
2168
2169<h5>Syntax:</h5>
2170
2171<pre>
2172 &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>
2173</pre>
2174
2175<h5>Overview:</h5>
2176
2177<p>
2178The '<tt>insertelement</tt>' instruction inserts a scalar
2179element into a packed vector at a specified index.
2180</p>
2181
2182
2183<h5>Arguments:</h5>
2184
2185<p>
2186The first operand of an '<tt>insertelement</tt>' instruction is a
2187value of <a href="#t_packed">packed</a> type. The second operand is a
2188scalar value whose type must equal the element type of the first
2189operand. The third operand is an index indicating the position at
2190which to insert the value. The index may be a variable.</p>
2191
2192<h5>Semantics:</h5>
2193
2194<p>
2195The result is a packed vector of the same type as <tt>val</tt>. Its
2196element values are those of <tt>val</tt> except at position
2197<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2198exceeds the length of <tt>val</tt>, the results are undefined.
2199</p>
2200
2201<h5>Example:</h5>
2202
2203<pre>
2204 %result = insertelement &lt;4 x int&gt; %vec, int 1, uint 0 <i>; yields &lt;4 x int&gt;</i>
2205</pre>
2206</div>
2207
2208<!-- _______________________________________________________________________ -->
2209<div class="doc_subsubsection">
2210 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2211</div>
2212
2213<div class="doc_text">
2214
2215<h5>Syntax:</h5>
2216
2217<pre>
2218 &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>
2219</pre>
2220
2221<h5>Overview:</h5>
2222
2223<p>
2224The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2225from two input vectors, returning a vector of the same type.
2226</p>
2227
2228<h5>Arguments:</h5>
2229
2230<p>
2231The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2232with types that match each other and types that match the result of the
2233instruction. The third argument is a shuffle mask, which has the same number
2234of elements as the other vector type, but whose element type is always 'uint'.
2235</p>
2236
2237<p>
2238The shuffle mask operand is required to be a constant vector with either
2239constant integer or undef values.
2240</p>
2241
2242<h5>Semantics:</h5>
2243
2244<p>
2245The elements of the two input vectors are numbered from left to right across
2246both of the vectors. The shuffle mask operand specifies, for each element of
2247the result vector, which element of the two input registers the result element
2248gets. The element selector may be undef (meaning "don't care") and the second
2249operand may be undef if performing a shuffle from only one vector.
2250</p>
2251
2252<h5>Example:</h5>
2253
2254<pre>
2255 %result = shufflevector &lt;4 x int&gt; %v1, &lt;4 x int&gt; %v2,
2256 &lt;4 x uint&gt; &lt;uint 0, uint 4, uint 1, uint 5&gt; <i>; yields &lt;4 x int&gt;</i>
2257 %result = shufflevector &lt;4 x int&gt; %v1, &lt;4 x int&gt; undef,
2258 &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.
2259</pre>
2260</div>
2261
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]2262
2263<!-- _______________________________________________________________________ -->
2264<div class="doc_subsubsection"> <a name="i_vsetint">'<tt>vsetint</tt>'
2265Instruction</a> </div>
2266<div class="doc_text">
2267<h5>Syntax:</h5>
2268<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>
2269</pre>
2270
2271<h5>Overview:</h5>
2272
2273<p>The '<tt>vsetint</tt>' instruction takes two integer vectors and
2274returns a vector of boolean values representing, at each position, the
2275result of the comparison between the values at that position in the
2276two operands.</p>
2277
2278<h5>Arguments:</h5>
2279
2280<p>The arguments to a '<tt>vsetint</tt>' instruction are a comparison
2281operation and two value arguments. The value arguments must be of <a
2282href="#t_integral">integral</a> <a href="#t_packed">packed</a> type,
2283and they must have identical types. The operation argument must be
2284one of <tt>eq</tt>, <tt>ne</tt>, <tt>slt</tt>, <tt>sgt</tt>,
2285<tt>sle</tt>, <tt>sge</tt>, <tt>ult</tt>, <tt>ugt</tt>, <tt>ule</tt>,
2286<tt>uge</tt>, <tt>true</tt>, and <tt>false</tt>. The result is a
2287packed <tt>bool</tt> value with the same length as each operand.</p>
2288
2289<h5>Semantics:</h5>
2290
2291<p>The following table shows the semantics of '<tt>vsetint</tt>'. For
2292each position of the result, the comparison is done on the
2293corresponding positions of the two value arguments. Note that the
2294signedness of the comparison depends on the comparison opcode and
2295<i>not</i> on the signedness of the value operands. E.g., <tt>vsetint
2296slt <4 x unsigned> %x, %y</tt> does an elementwise <i>signed</i>
2297comparison of <tt>%x</tt> and <tt>%y</tt>.</p>
2298
2299<table border="1" cellspacing="0" cellpadding="4">
2300 <tbody>
2301 <tr><th>Operation</th><th>Result is true iff</th><th>Comparison is</th></tr>
2302 <tr><td><tt>eq</tt></td><td>var1 == var2</td><td>--</td></tr>
2303 <tr><td><tt>ne</tt></td><td>var1 != var2</td><td>--</td></tr>
2304 <tr><td><tt>slt</tt></td><td>var1 &lt; var2</td><td>signed</td></tr>
2305 <tr><td><tt>sgt</tt></td><td>var1 &gt; var2</td><td>signed</td></tr>
2306 <tr><td><tt>sle</tt></td><td>var1 &lt;= var2</td><td>signed</td></tr>
2307 <tr><td><tt>sge</tt></td><td>var1 &gt;= var2</td><td>signed</td></tr>
2308 <tr><td><tt>ult</tt></td><td>var1 &lt; var2</td><td>unsigned</td></tr>
2309 <tr><td><tt>ugt</tt></td><td>var1 &gt; var2</td><td>unsigned</td></tr>
2310 <tr><td><tt>ule</tt></td><td>var1 &lt;= var2</td><td>unsigned</td></tr>
2311 <tr><td><tt>uge</tt></td><td>var1 &gt;= var2</td><td>unsigned</td></tr>
2312 <tr><td><tt>true</tt></td><td>always</td><td>--</td></tr>
2313 <tr><td><tt>false</tt></td><td>never</td><td>--</td></tr>
2314 </tbody>
2315</table>
2316
2317<h5>Example:</h5>
2318<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>
2319 &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>
2320 &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>
2321 &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>
2322 &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>
2323 &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>
2324</pre>
2325</div>
2326
2327<!-- _______________________________________________________________________ -->
2328<div class="doc_subsubsection"> <a name="i_vsetfp">'<tt>vsetfp</tt>'
2329Instruction</a> </div>
2330<div class="doc_text">
2331<h5>Syntax:</h5>
2332<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>
2333</pre>
2334
2335<h5>Overview:</h5>
2336
2337<p>The '<tt>vsetfp</tt>' instruction takes two floating point vector
2338arguments and returns a vector of boolean values representing, at each
2339position, the result of the comparison between the values at that
2340position in the two operands.</p>
2341
2342<h5>Arguments:</h5>
2343
2344<p>The arguments to a '<tt>vsetfp</tt>' instruction are a comparison
2345operation and two value arguments. The value arguments must be of <a
2346href="t_floating">floating point</a> <a href="#t_packed">packed</a>
2347type, and they must have identical types. The operation argument must
2348be one of <tt>eq</tt>, <tt>ne</tt>, <tt>lt</tt>, <tt>gt</tt>,
2349<tt>le</tt>, <tt>ge</tt>, <tt>oeq</tt>, <tt>one</tt>, <tt>olt</tt>,
2350<tt>ogt</tt>, <tt>ole</tt>, <tt>oge</tt>, <tt>ueq</tt>, <tt>une</tt>,
2351<tt>ult</tt>, <tt>ugt</tt>, <tt>ule</tt>, <tt>uge</tt>, <tt>o</tt>,
2352<tt>u</tt>, <tt>true</tt>, and <tt>false</tt>. The result is a packed
2353<tt>bool</tt> value with the same length as each operand.</p>
2354
2355<h5>Semantics:</h5>
2356
2357<p>The following table shows the semantics of '<tt>vsetfp</tt>' for
2358floating point types. If either operand is a floating point Not a
2359Number (NaN) value, the operation is unordered, and the value in the
2360first column below is produced at that position. Otherwise, the
2361operation is ordered, and the value in the second column is
2362produced.</p>
2363
2364<table border="1" cellspacing="0" cellpadding="4">
2365 <tbody>
2366 <tr><th>Operation</th><th>If unordered<th>Otherwise true iff</th></tr>
2367 <tr><td><tt>eq</tt></td><td>undefined</td><td>var1 == var2</td></tr>
2368 <tr><td><tt>ne</tt></td><td>undefined</td><td>var1 != var2</td></tr>
2369 <tr><td><tt>lt</tt></td><td>undefined</td><td>var1 &lt; var2</td></tr>
2370 <tr><td><tt>gt</tt></td><td>undefined</td><td>var1 &gt; var2</td></tr>
2371 <tr><td><tt>le</tt></td><td>undefined</td><td>var1 &lt;= var2</td></tr>
2372 <tr><td><tt>ge</tt></td><td>undefined</td><td>var1 &gt;= var2</td></tr>
2373 <tr><td><tt>oeq</tt></td><td>false</td><td>var1 == var2</td></tr>
2374 <tr><td><tt>one</tt></td><td>false</td><td>var1 != var2</td></tr>
2375 <tr><td><tt>olt</tt></td><td>false</td><td>var1 &lt; var2</td></tr>
2376 <tr><td><tt>ogt</tt></td><td>false</td><td>var1 &gt; var2</td></tr>
2377 <tr><td><tt>ole</tt></td><td>false</td><td>var1 &lt;= var2</td></tr>
2378 <tr><td><tt>oge</tt></td><td>false</td><td>var1 &gt;= var2</td></tr>
2379 <tr><td><tt>ueq</tt></td><td>true</td><td>var1 == var2</td></tr>
2380 <tr><td><tt>une</tt></td><td>true</td><td>var1 != var2</td></tr>
2381 <tr><td><tt>ult</tt></td><td>true</td><td>var1 &lt; var2</td></tr>
2382 <tr><td><tt>ugt</tt></td><td>true</td><td>var1 &gt; var2</td></tr>
2383 <tr><td><tt>ule</tt></td><td>true</td><td>var1 &lt;= var2</td></tr>
2384 <tr><td><tt>uge</tt></td><td>true</td><td>var1 &gt;= var2</td></tr>
2385 <tr><td><tt>o</tt></td><td>false</td><td>always</td></tr>
2386 <tr><td><tt>u</tt></td><td>true</td><td>never</td></tr>
2387 <tr><td><tt>true</tt></td><td>true</td><td>always</td></tr>
2388 <tr><td><tt>false</tt></td><td>false</td><td>never</td></tr>
2389 </tbody>
2390</table>
2391
2392<h5>Example:</h5>
2393<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>
2394 &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>
2395 &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>
2396 &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>
2397 &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>
2398 &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>
2399</pre>
2400</div>
2401
2402<!-- _______________________________________________________________________ -->
2403<div class="doc_subsubsection">
2404 <a name="i_vselect">'<tt>vselect</tt>' Instruction</a>
2405</div>
2406
2407<div class="doc_text">
2408
2409<h5>Syntax:</h5>
2410
2411<pre>
2412 &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>
2413</pre>
2414
2415<h5>Overview:</h5>
2416
2417<p>
2418The '<tt>vselect</tt>' instruction chooses one value at each position
2419of a vector based on a condition.
2420</p>
2421
2422
2423<h5>Arguments:</h5>
2424
2425<p>
2426The '<tt>vselect</tt>' instruction requires a <a
2427href="#t_packed">packed</a> <tt>bool</tt> value indicating the
2428condition at each vector position, and two values of the same packed
2429type. All three operands must have the same length. The type of the
2430result is the same as the type of the two value operands.</p>
2431
2432<h5>Semantics:</h5>
2433
2434<p>
2435At each position where the <tt>bool</tt> vector is true, that position
2436of the result gets its value from the first value argument; otherwise,
2437it gets its value from the second value argument.
2438</p>
2439
2440<h5>Example:</h5>
2441
2442<pre>
2443 %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;,
2444 &lt;2 x ubyte&gt; &lt;ubyte 42, ubyte 42&gt; <i>; yields &lt;2 x ubyte&gt;:17, 42</i>
2445</pre>
2446</div>
2447
2448
2449
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2450<!-- ======================================================================= -->
2451<div class="doc_subsection">
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]2452 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2453</div>
2454
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2455<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2456
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2457<p>A key design point of an SSA-based representation is how it
2458represents memory. In LLVM, no memory locations are in SSA form, which
2459makes things very simple. This section describes how to read, write,
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2460allocate, and free memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2461
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2462</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2463
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2464<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2465<div class="doc_subsubsection">
2466 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2467</div>
2468
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2469<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2470
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2471<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2472
2473<pre>
2474 &lt;result&gt; = malloc &lt;type&gt;[, uint &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2475</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2476
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2477<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2478
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2479<p>The '<tt>malloc</tt>' instruction allocates memory from the system
2480heap and returns a pointer to it.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2481
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2482<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2483
2484<p>The '<tt>malloc</tt>' instruction allocates
2485<tt>sizeof(&lt;type&gt;)*NumElements</tt>
John Criswell6e4ca612004-02-24 16:13:56 +0000[diff] [blame]2486bytes of memory from the operating system and returns a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2487appropriate type to the program. If "NumElements" is specified, it is the
2488number of elements allocated. If an alignment is specified, the value result
2489of the allocation is guaranteed to be aligned to at least that boundary. If
2490not specified, or if zero, the target can choose to align the allocation on any
2491convenient boundary.</p>
2492
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2493<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2494
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2495<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2496
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2497<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2498a pointer is returned.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2499
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2500<h5>Example:</h5>
2501
2502<pre>
2503 %array = malloc [4 x ubyte ] <i>; yields {[%4 x ubyte]*}:array</i>
2504
2505 %size = <a href="#i_add">add</a> uint 2, 2 <i>; yields {uint}:size = uint 4</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2506 %array1 = malloc ubyte, uint 4 <i>; yields {ubyte*}:array1</i>
2507 %array2 = malloc [12 x ubyte], uint %size <i>; yields {[12 x ubyte]*}:array2</i>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2508 %array3 = malloc int, uint 4, align 1024 <i>; yields {int*}:array3</i>
2509 %array4 = malloc int, align 1024 <i>; yields {int*}:array4</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2510</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2511</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2512
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2513<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2514<div class="doc_subsubsection">
2515 <a name="i_free">'<tt>free</tt>' Instruction</a>
2516</div>
2517
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2518<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2519
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2520<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2521
2522<pre>
2523 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2524</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2525
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2526<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2527
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2528<p>The '<tt>free</tt>' instruction returns memory back to the unused
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2529memory heap to be reallocated in the future.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2530
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2531<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2532
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2533<p>'<tt>value</tt>' shall be a pointer value that points to a value
2534that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2535instruction.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2536
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2537<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2538
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2539<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2540after this instruction executes.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2541
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2542<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2543
2544<pre>
2545 %array = <a href="#i_malloc">malloc</a> [4 x ubyte] <i>; yields {[4 x ubyte]*}:array</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2546 free [4 x ubyte]* %array
2547</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2548</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2549
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2550<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2551<div class="doc_subsubsection">
2552 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2553</div>
2554
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2555<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2556
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2557<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2558
2559<pre>
2560 &lt;result&gt; = alloca &lt;type&gt;[, uint &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2561</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2562
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2563<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2564
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2565<p>The '<tt>alloca</tt>' instruction allocates memory on the current
2566stack frame of the procedure that is live until the current function
2567returns to its caller.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2568
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2569<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2570
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2571<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2572bytes of memory on the runtime stack, returning a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2573appropriate type to the program. If "NumElements" is specified, it is the
2574number of elements allocated. If an alignment is specified, the value result
2575of the allocation is guaranteed to be aligned to at least that boundary. If
2576not specified, or if zero, the target can choose to align the allocation on any
2577convenient boundary.</p>
2578
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2579<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2580
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2581<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2582
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2583<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2584memory is automatically released when the function returns. The '<tt>alloca</tt>'
2585instruction is commonly used to represent automatic variables that must
2586have an address available. When the function returns (either with the <tt><a
John Criswelldae2e932005-05-12 16:55:34 +0000[diff] [blame]2587 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2588instructions), the memory is reclaimed.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2589
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2590<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2591
2592<pre>
2593 %ptr = alloca int <i>; yields {int*}:ptr</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2594 %ptr = alloca int, uint 4 <i>; yields {int*}:ptr</i>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2595 %ptr = alloca int, uint 4, align 1024 <i>; yields {int*}:ptr</i>
2596 %ptr = alloca int, align 1024 <i>; yields {int*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2597</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2598</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2599
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2600<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2601<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2602Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2603<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2604<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2605<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 Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2606<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2607<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2608<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2609<p>The argument to the '<tt>load</tt>' instruction specifies the memory
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2610address from which to load. The pointer must point to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +0000[diff] [blame]2611 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2612marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2613the number or order of execution of this <tt>load</tt> with other
2614volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2615instructions. </p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2616<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2617<p>The location of memory pointed to is loaded.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2618<h5>Examples:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2619<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
2620 <a
2621 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2622 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
2623</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2624</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2625<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2626<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2627Instruction</a> </div>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2628<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2629<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; <i>; yields {void}</i>
Chris Lattnerf0651072003-09-08 18:27:49 +0000[diff] [blame]2630 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2631</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2632<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2633<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2634<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2635<p>There are two arguments to the '<tt>store</tt>' instruction: a value
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2636to store and an address in which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2637operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2638operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2639optimizer is not allowed to modify the number or order of execution of
2640this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2641 href="#i_store">store</a></tt> instructions.</p>
2642<h5>Semantics:</h5>
2643<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2644at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2645<h5>Example:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2646<pre> %ptr = <a href="#i_alloca">alloca</a> int <i>; yields {int*}:ptr</i>
2647 <a
2648 href="#i_store">store</a> int 3, int* %ptr <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2649 %val = load int* %ptr <i>; yields {int}:val = int 3</i>
2650</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2651<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2652<div class="doc_subsubsection">
2653 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2654</div>
2655
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2656<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2657<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2658<pre>
2659 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2660</pre>
2661
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2662<h5>Overview:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2663
2664<p>
2665The '<tt>getelementptr</tt>' instruction is used to get the address of a
2666subelement of an aggregate data structure.</p>
2667
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2668<h5>Arguments:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2669
2670<p>This instruction takes a list of integer constants that indicate what
2671elements of the aggregate object to index to. The actual types of the arguments
2672provided depend on the type of the first pointer argument. The
2673'<tt>getelementptr</tt>' instruction is used to index down through the type
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]2674levels of a structure or to a specific index in an array. When indexing into a
2675structure, only <tt>uint</tt>
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2676integer constants are allowed. When indexing into an array or pointer,
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2677<tt>int</tt> and <tt>long</tt> indexes are allowed of any sign.</p>
2678
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2679<p>For example, let's consider a C code fragment and how it gets
2680compiled to LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2681
2682<pre>
2683 struct RT {
2684 char A;
2685 int B[10][20];
2686 char C;
2687 };
2688 struct ST {
2689 int X;
2690 double Y;
2691 struct RT Z;
2692 };
2693
2694 int *foo(struct ST *s) {
2695 return &amp;s[1].Z.B[5][13];
2696 }
2697</pre>
2698
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2699<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2700
2701<pre>
2702 %RT = type { sbyte, [10 x [20 x int]], sbyte }
2703 %ST = type { int, double, %RT }
2704
Brian Gaeke7283e7c2004-07-02 21:08:14 +0000[diff] [blame]2705 implementation
2706
2707 int* %foo(%ST* %s) {
2708 entry:
2709 %reg = getelementptr %ST* %s, int 1, uint 2, uint 1, int 5, int 13
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2710 ret int* %reg
2711 }
2712</pre>
2713
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2714<h5>Semantics:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2715
2716<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2717on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
Chris Lattnere53e5082004-06-03 22:57:15 +0000[diff] [blame]2718and <a href="#t_array">array</a> types require <tt>uint</tt>, <tt>int</tt>,
2719<tt>ulong</tt>, or <tt>long</tt> values, and <a href="#t_struct">structure</a>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2720types require <tt>uint</tt> <b>constants</b>.</p>
2721
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2722<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2723type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ int, double, %RT
2724}</tt>' type, a structure. The second index indexes into the third element of
2725the structure, yielding a '<tt>%RT</tt>' = '<tt>{ sbyte, [10 x [20 x int]],
2726sbyte }</tt>' type, another structure. The third index indexes into the second
2727element of the structure, yielding a '<tt>[10 x [20 x int]]</tt>' type, an
2728array. The two dimensions of the array are subscripted into, yielding an
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]2729'<tt>int</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2730to this element, thus computing a value of '<tt>int*</tt>' type.</p>
2731
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2732<p>Note that it is perfectly legal to index partially through a
2733structure, returning a pointer to an inner element. Because of this,
2734the LLVM code for the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2735
2736<pre>
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]2737 int* %foo(%ST* %s) {
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2738 %t1 = getelementptr %ST* %s, int 1 <i>; yields %ST*:%t1</i>
2739 %t2 = getelementptr %ST* %t1, int 0, uint 2 <i>; yields %RT*:%t2</i>
2740 %t3 = getelementptr %RT* %t2, int 0, uint 1 <i>; yields [10 x [20 x int]]*:%t3</i>
2741 %t4 = getelementptr [10 x [20 x int]]* %t3, int 0, int 5 <i>; yields [20 x int]*:%t4</i>
2742 %t5 = getelementptr [20 x int]* %t4, int 0, int 13 <i>; yields int*:%t5</i>
2743 ret int* %t5
2744 }
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2745</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2746
2747<p>Note that it is undefined to access an array out of bounds: array and
2748pointer indexes must always be within the defined bounds of the array type.
2749The one exception for this rules is zero length arrays. These arrays are
2750defined to be accessible as variable length arrays, which requires access
2751beyond the zero'th element.</p>
2752
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]2753<p>The getelementptr instruction is often confusing. For some more insight
2754into how it works, see <a href="GetElementPtr.html">the getelementptr
2755FAQ</a>.</p>
2756
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2757<h5>Example:</h5>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2758
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2759<pre>
2760 <i>; yields [12 x ubyte]*:aptr</i>
2761 %aptr = getelementptr {int, [12 x ubyte]}* %sptr, long 0, uint 1
2762</pre>
2763
2764</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2765<!-- ======================================================================= -->
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame^]2766<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2767</div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2768<div class="doc_text">
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame^]2769<p>The instructions in this category are the conversion instructions (casting)
2770which all take a single operand and a type. They perform various bit conversions
2771on the operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2772</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2773
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2774<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2775<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2776 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2777</div>
2778<div class="doc_text">
2779
2780<h5>Syntax:</h5>
2781<pre>
2782 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2783</pre>
2784
2785<h5>Overview:</h5>
2786<p>
2787The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2788</p>
2789
2790<h5>Arguments:</h5>
2791<p>
2792The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2793be an <a href="#t_integer">integer</a> type, and a type that specifies the size
2794and type of the result, which must be an <a href="#t_integral">integral</a>
2795type.</p>
2796
2797<h5>Semantics:</h5>
2798<p>
2799The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
2800and converts the reamining bits to <tt>ty2</tt>. The bit size of <tt>value</tt>
2801must be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
2802allowed. This implies that a <tt>trunc</tt> cannot be a <i>no-op cast</i>. It
2803will always truncate bits.</p>
2804
2805<p>When truncating to bool, the truncation is done as a comparison against
2806zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
2807If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
2808
2809<h5>Example:</h5>
2810<pre>
2811 %X = trunc int 257 to ubyte <i>; yields ubyte:1</i>
2812 %Y = trunc int 123 to bool <i>; yields bool:true</i>
2813</pre>
2814</div>
2815
2816<!-- _______________________________________________________________________ -->
2817<div class="doc_subsubsection">
2818 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2819</div>
2820<div class="doc_text">
2821
2822<h5>Syntax:</h5>
2823<pre>
2824 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2825</pre>
2826
2827<h5>Overview:</h5>
2828<p>The '<tt>zext</tt>' instruction zero extends its operand to type
2829<tt>ty2</tt>.</p>
2830
2831
2832<h5>Arguments:</h5>
2833<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
2834<a href="#t_integral">integral</a> type, and a type to cast it to, which must
2835also be of <a href="#t_integral">integral</a> type. The bit size of the
2836<tt>value</tt> must be smaller than or equal to the bit size of the
2837destination type, <tt>ty2</tt>.</p>
2838
2839<h5>Semantics:</h5>
2840<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
2841bits until it reaches the size of the destination type, <tt>ty2</tt>. When the
2842the operand and the type are the same size, no bit filling is done and the
2843cast is considered a <i>no-op cast</i> because no bits change (only the type
2844changes).</p>
2845
2846<p>When zero extending to bool, the extension is done as a comparison against
2847zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
2848If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
2849
2850<h5>Example:</h5>
2851<pre>
2852 %X = zext int 257 to ulong <i>; yields ulong:257</i>
2853 %Y = zext bool true to int <i>; yields int:1</i>
2854</pre>
2855</div>
2856
2857<!-- _______________________________________________________________________ -->
2858<div class="doc_subsubsection">
2859 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2860</div>
2861<div class="doc_text">
2862
2863<h5>Syntax:</h5>
2864<pre>
2865 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2866</pre>
2867
2868<h5>Overview:</h5>
2869<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
2870
2871<h5>Arguments:</h5>
2872<p>
2873The '<tt>sext</tt>' instruction takes a value to cast, which must be of
2874<a href="#t_integral">integral</a> type, and a type to cast it to, which must
2875also be of <a href="#t_integral">integral</a> type.</p>
2876
2877<h5>Semantics:</h5>
2878<p>
2879The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
2880bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
2881the type <tt>ty2</tt>. When the the operand and the type are the same size,
2882no bit filling is done and the cast is considered a <i>no-op cast</i> because
2883no bits change (only the type changes).</p>
2884
2885<p>When sign extending to bool, the extension is done as a comparison against
2886zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
2887If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
2888
2889<h5>Example:</h5>
2890
2891<pre>
2892 %X = sext sbyte -1 to ushort <i>; yields ushort:65535</i>
2893 %Y = sext bool true to int <i>; yields int:-1</i>
2894</pre>
2895</div>
2896
2897<!-- _______________________________________________________________________ -->
2898<div class="doc_subsubsection">
2899 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
2900</div>
2901<div class="doc_text">
2902
2903<h5>Syntax:</h5>
2904<pre>
2905 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2906</pre>
2907
2908<h5>Overview:</h5>
2909<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
2910floating point value.</p>
2911
2912<h5>Arguments:</h5>
2913<p>The '<tt>fpext</tt>' instruction takes a
2914<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
2915and a <a href="#t_floating">floating point</a> type to cast it to.</p>
2916
2917<h5>Semantics:</h5>
2918<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from one floating
2919point type to another. If the type of the <tt>value</tt> and <tt>ty2</tt> are
2920the same, the instruction is considered a <i>no-op cast</i> because no bits
2921change.</p>
2922
2923<h5>Example:</h5>
2924<pre>
2925 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
2926 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
2927</pre>
2928</div>
2929
2930<!-- _______________________________________________________________________ -->
2931<div class="doc_subsubsection">
2932 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2933</div>
2934
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2935<div class="doc_text">
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2936
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2937<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2938
2939<pre>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2940 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2941</pre>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2942
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2943<h5>Overview:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2944<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
2945<tt>ty2</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2946
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2947
2948<h5>Arguments:</h5>
2949<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
2950 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
2951cast it to. The size of <tt>value</tt> must be larger than the size of
2952<tt>ty2</a>. This implies that <tt>fptrunc</tt> cannot be used to make a
2953<i>no-op cast</i>.</p>
2954
2955<h5>Semantics:</h5>
2956<p> The '<tt>fptrunc</tt>' instruction converts a
2957<a href="#t_floating">floating point</a> value from a larger type to a smaller
2958type. If the value cannot fit within the destination type, <tt>ty2</tt>, then
2959the results are undefined.</p>
2960
2961<h5>Example:</h5>
2962<pre>
2963 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
2964 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
2965</pre>
2966</div>
2967
2968<!-- _______________________________________________________________________ -->
2969<div class="doc_subsubsection">
2970 <a name="i_fp2uint">'<tt>fp2uint .. to</tt>' Instruction</a>
2971</div>
2972<div class="doc_text">
2973
2974<h5>Syntax:</h5>
2975<pre>
2976 &lt;result&gt; = fp2uint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2977</pre>
2978
2979<h5>Overview:</h5>
2980<p>The '<tt>fp2uint</tt>' converts a floating point <tt>value</tt> to its
2981unsigned integer equivalent of type <tt>ty2</tt>.
2982</p>
2983
2984<h5>Arguments:</h5>
2985<p>The '<tt>fp2uint</tt>' instruction takes a value to cast, which must be a
2986<a href="#t_floating">floating point</a> value, and a type to cast it to, which
2987must be an <a href="#t_integral">integral</a> type.</p>
2988
2989<h5>Semantics:</h5>
2990<p> The '<tt>fp2uint</tt>' instruction converts its
2991<a href="#t_floating">floating point</a> operand into the nearest (rounding
2992towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
2993the results are undefined.</p>
2994
2995<p>When converting to bool, the conversion is done as a comparison against
2996zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
2997If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
2998
2999<h5>Example:</h5>
3000<pre>
3001 %X = fp2uint double 123.0 to int <i>; yields int:123</i>
3002 %Y = fp2uint float 1.0E+300 to bool <i>; yields bool:true</i>
3003 %X = fp2uint float 1.04E+17 to ubyte <i>; yields undefined:1</i>
3004</pre>
3005</div>
3006
3007<!-- _______________________________________________________________________ -->
3008<div class="doc_subsubsection">
3009 <a name="i_fp2sint">'<tt>fp2sint .. to</tt>' Instruction</a>
3010</div>
3011<div class="doc_text">
3012
3013<h5>Syntax:</h5>
3014<pre>
3015 &lt;result&gt; = fp2sint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3016</pre>
3017
3018<h5>Overview:</h5>
3019<p>The '<tt>fp2sint</tt>' instruction converts
3020<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3021</p>
3022
3023
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3024<h5>Arguments:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3025<p> The '<tt>fp2sint</tt>' instruction takes a value to cast, which must be a
3026<a href="#t_floating">floating point</a> value, and a type to cast it to, which
3027must also be an <a href="#t_integral">integral</a> type.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3028
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3029<h5>Semantics:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3030<p>The '<tt>fp2sint</tt>' instruction converts its
3031<a href="#t_floating">floating point</a> operand into the nearest (rounding
3032towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3033the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3034
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3035<p>When converting to bool, the conversion is done as a comparison against
3036zero. If the <tt>value</tt> was zero, the bool result will be <tt>false</tt>.
3037If the <tt>value</tt> was non-zero, the bool result will be <tt>true</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3038
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]3039<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3040<pre>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3041 %X = fp2sint double -123.0 to int <i>; yields int:-123</i>
3042 %Y = fp2sint float 1.0E-247 to bool <i>; yields bool:true</i>
3043 %X = fp2sint float 1.04E+17 to sbyte <i>; yields undefined:1</i>
3044</pre>
3045</div>
3046
3047<!-- _______________________________________________________________________ -->
3048<div class="doc_subsubsection">
3049 <a name="i_uint2fp">'<tt>uint2fp .. to</tt>' Instruction</a>
3050</div>
3051<div class="doc_text">
3052
3053<h5>Syntax:</h5>
3054<pre>
3055 &lt;result&gt; = uint2fp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3056</pre>
3057
3058<h5>Overview:</h5>
3059<p>The '<tt>uint2fp</tt>' instruction regards <tt>value</tt> as an unsigned
3060integer and converts that value to the <tt>ty2</tt> type.</p>
3061
3062
3063<h5>Arguments:</h5>
3064<p>The '<tt>uint2fp</tt>' instruction takes a value to cast, which must be an
3065<a href="#t_integral">integral</a> value, and a type to cast it to, which must
3066be a <a href="#t_floating">floating point</a> type.</p>
3067
3068<h5>Semantics:</h5>
3069<p>The '<tt>uint2fp</tt>' instruction interprets its operand as an unsigned
3070integer quantity and converts it to the corresponding floating point value. If
3071the value cannot fit in the floating point value, the results are undefined.</p>
3072
3073
3074<h5>Example:</h5>
3075<pre>
3076 %X = uint2fp int 257 to float <i>; yields float:257.0</i>
3077 %Y = uint2fp sbyte -1 to double <i>; yields double:255.0</i>
3078</pre>
3079</div>
3080
3081<!-- _______________________________________________________________________ -->
3082<div class="doc_subsubsection">
3083 <a name="i_sint2fp">'<tt>sint2fp .. to</tt>' Instruction</a>
3084</div>
3085<div class="doc_text">
3086
3087<h5>Syntax:</h5>
3088<pre>
3089 &lt;result&gt; = sint2fp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3090</pre>
3091
3092<h5>Overview:</h5>
3093<p>The '<tt>sint2fp</tt>' instruction regards <tt>value</tt> as a signed
3094integer and converts that value to the <tt>ty2</tt> type.</p>
3095
3096<h5>Arguments:</h5>
3097<p>The '<tt>sint2fp</tt>' instruction takes a value to cast, which must be an
3098<a href="#t_integral">integral</a> value, and a type to cast it to, which must be
3099a <a href="#t_floating">floating point</a> type.</p>
3100
3101<h5>Semantics:</h5>
3102<p>The '<tt>sint2fp</tt>' instruction interprets its operand as a signed
3103integer quantity and converts it to the corresponding floating point value. If
3104the value cannot fit in the floating point value, the results are undefined.</p>
3105
3106<h5>Example:</h5>
3107<pre>
3108 %X = sint2fp int 257 to float <i>; yields float:257.0</i>
3109 %Y = sint2fp sbyte -1 to double <i>; yields double:-1.0</i>
3110</pre>
3111</div>
3112
3113<!-- _______________________________________________________________________ -->
3114<div class="doc_subsubsection">
3115 <a name="i_bitconvert">'<tt>bitconvert .. to</tt>' Instruction</a>
3116</div>
3117<div class="doc_text">
3118
3119<h5>Syntax:</h5>
3120<pre>
3121 &lt;result&gt; = bitconvert &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3122</pre>
3123
3124<h5>Overview:</h5>
3125<p>The '<tt>bitconvert</tt>' instruction converts <tt>value</tt> to type
3126<tt>ty2</tt> without changing any bits.</p>
3127
3128<h5>Arguments:</h5>
3129<p>The '<tt>bitconvert</tt>' instruction takes a value to cast, which must be
3130a first class value, and a type to cast it to, which must also be a <a
3131 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
3132and the destination type, <tt>ty2</tt>, must be identical.</p>
3133
3134<h5>Semantics:</h5>
3135<p>The '<tt>bitconvert</tt>' instruction converts <tt>value</tt> to type
3136<tt>ty2</tt> as if the value had been stored to memory and read back as type
3137<tt>ty2</tt>. That is, no bits are changed during the conversion. The
3138<tt>bitconvert</tt> instruction may be used to construct <i>no-op casts</i> that
3139the <tt>zext, sext, and fpext</tt> instructions do not permit.</p>
3140
3141<h5>Example:</h5>
3142<pre>
3143 %X = bitconvert ubyte 255 to sbyte <i>; yields sbyte:-1</i>
3144 %Y = bitconvert uint* %x to uint <i>; yields uint:%x</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]3145</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3146</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3147
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame^]3148<!-- ======================================================================= -->
3149<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3150<div class="doc_text">
3151<p>The instructions in this category are the "miscellaneous"
3152instructions, which defy better classification.</p>
3153</div>
3154<!-- _______________________________________________________________________ -->
3155<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3156Instruction</a> </div>
3157<div class="doc_text">
3158<h5>Syntax:</h5>
3159<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3160<h5>Overview:</h5>
3161<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3162the SSA graph representing the function.</p>
3163<h5>Arguments:</h5>
3164<p>The type of the incoming values are specified with the first type
3165field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3166as arguments, with one pair for each predecessor basic block of the
3167current block. Only values of <a href="#t_firstclass">first class</a>
3168type may be used as the value arguments to the PHI node. Only labels
3169may be used as the label arguments.</p>
3170<p>There must be no non-phi instructions between the start of a basic
3171block and the PHI instructions: i.e. PHI instructions must be first in
3172a basic block.</p>
3173<h5>Semantics:</h5>
3174<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the
3175value specified by the parameter, depending on which basic block we
3176came from in the last <a href="#terminators">terminator</a> instruction.</p>
3177<h5>Example:</h5>
3178<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>
3179</div>
3180
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3181<!-- _______________________________________________________________________ -->
3182<div class="doc_subsubsection">
3183 <a name="i_select">'<tt>select</tt>' Instruction</a>
3184</div>
3185
3186<div class="doc_text">
3187
3188<h5>Syntax:</h5>
3189
3190<pre>
3191 &lt;result&gt; = select bool &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
3192</pre>
3193
3194<h5>Overview:</h5>
3195
3196<p>
3197The '<tt>select</tt>' instruction is used to choose one value based on a
3198condition, without branching.
3199</p>
3200
3201
3202<h5>Arguments:</h5>
3203
3204<p>
3205The '<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.
3206</p>
3207
3208<h5>Semantics:</h5>
3209
3210<p>
3211If the boolean condition evaluates to true, the instruction returns the first
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3212value argument; otherwise, it returns the second value argument.
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3213</p>
3214
3215<h5>Example:</h5>
3216
3217<pre>
3218 %X = select bool true, ubyte 17, ubyte 42 <i>; yields ubyte:17</i>
3219</pre>
3220</div>
3221
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]3222
3223<!-- _______________________________________________________________________ -->
3224<div class="doc_subsubsection">
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3225 <a name="i_call">'<tt>call</tt>' Instruction</a>
3226</div>
3227
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3228<div class="doc_text">
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3229
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3230<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3231<pre>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3232 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt;* &lt;fnptrval&gt;(&lt;param list&gt;)
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3233</pre>
3234
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3235<h5>Overview:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3236
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3237<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3238
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3239<h5>Arguments:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3240
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3241<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3242
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3243<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3244 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3245 <p>The optional "tail" marker indicates whether the callee function accesses
3246 any allocas or varargs in the caller. If the "tail" marker is present, the
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3247 function call is eligible for tail call optimization. Note that calls may
3248 be marked "tail" even if they do not occur before a <a
3249 href="#i_ret"><tt>ret</tt></a> instruction.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3250 </li>
3251 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3252 <p>The optional "cconv" marker indicates which <a href="callingconv">calling
3253 convention</a> the call should use. If none is specified, the call defaults
3254 to using C calling conventions.
3255 </li>
3256 <li>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3257 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
3258 being invoked. The argument types must match the types implied by this
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3259 signature. This type can be omitted if the function is not varargs and
3260 if the function type does not return a pointer to a function.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3261 </li>
3262 <li>
3263 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3264 be invoked. In most cases, this is a direct function invocation, but
3265 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3266 to function value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3267 </li>
3268 <li>
3269 <p>'<tt>function args</tt>': argument list whose types match the
Reid Spencera7e302a2005-05-01 22:22:57 +0000[diff] [blame]3270 function signature argument types. All arguments must be of
3271 <a href="#t_firstclass">first class</a> type. If the function signature
3272 indicates the function accepts a variable number of arguments, the extra
3273 arguments can be specified.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3274 </li>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3275</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3276
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3277<h5>Semantics:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3278
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3279<p>The '<tt>call</tt>' instruction is used to cause control flow to
3280transfer to a specified function, with its incoming arguments bound to
3281the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3282instruction in the called function, control flow continues with the
3283instruction after the function call, and the return value of the
3284function is bound to the result argument. This is a simpler case of
3285the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3286
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3287<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3288
3289<pre>
3290 %retval = call int %test(int %argc)
3291 call int(sbyte*, ...) *%printf(sbyte* %msg, int 12, sbyte 42);
3292 %X = tail call int %foo()
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3293 %Y = tail call <a href="#callingconv">fastcc</a> int %foo()
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3294</pre>
3295
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3296</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3297
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3298<!-- _______________________________________________________________________ -->
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3299<div class="doc_subsubsection">
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]3300 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3301</div>
3302
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3303<div class="doc_text">
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3304
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3305<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3306
3307<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3308 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3309</pre>
3310
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3311<h5>Overview:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3312
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3313<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3314the "variable argument" area of a function call. It is used to implement the
3315<tt>va_arg</tt> macro in C.</p>
3316
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3317<h5>Arguments:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3318
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3319<p>This instruction takes a <tt>va_list*</tt> value and the type of
3320the argument. It returns a value of the specified argument type and
Jeff Cohen25d4f7e2005-11-11 02:15:27 +0000[diff] [blame]3321increments the <tt>va_list</tt> to point to the next argument. Again, the
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3322actual type of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3323
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3324<h5>Semantics:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3325
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3326<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3327type from the specified <tt>va_list</tt> and causes the
3328<tt>va_list</tt> to point to the next argument. For more information,
3329see the variable argument handling <a href="#int_varargs">Intrinsic
3330Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3331
3332<p>It is legal for this instruction to be called in a function which does not
3333take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3334function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3335
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3336<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3337href="#intrinsics">intrinsic function</a> because it takes a type as an
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3338argument.</p>
3339
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3340<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3341
3342<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3343
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3344</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3345
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3346<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3347<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3348<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3349
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3350<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3351
3352<p>LLVM supports the notion of an "intrinsic function". These functions have
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3353well known names and semantics and are required to follow certain
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3354restrictions. Overall, these instructions represent an extension mechanism for
3355the LLVM language that does not require changing all of the transformations in
3356LLVM to add to the language (or the bytecode reader/writer, the parser,
3357etc...).</p>
3358
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3359<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
3360prefix is reserved in LLVM for intrinsic names; thus, functions may not be named
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3361this. Intrinsic functions must always be external functions: you cannot define
3362the body of intrinsic functions. Intrinsic functions may only be used in call
3363or invoke instructions: it is illegal to take the address of an intrinsic
3364function. Additionally, because intrinsic functions are part of the LLVM
3365language, it is required that they all be documented here if any are added.</p>
3366
3367
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3368<p>To learn how to add an intrinsic function, please see the <a
Chris Lattner590cff32005-05-11 03:35:57 +0000[diff] [blame]3369href="ExtendingLLVM.html">Extending LLVM Guide</a>.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3370</p>
3371
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3372</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3373
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3374<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3375<div class="doc_subsection">
3376 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3377</div>
3378
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3379<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3380
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3381<p>Variable argument support is defined in LLVM with the <a
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]3382 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3383intrinsic functions. These functions are related to the similarly
3384named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3385
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3386<p>All of these functions operate on arguments that use a
3387target-specific value type "<tt>va_list</tt>". The LLVM assembly
3388language reference manual does not define what this type is, so all
3389transformations should be prepared to handle intrinsics with any type
3390used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3391
Chris Lattner374ab302006-05-15 17:26:46 +0000[diff] [blame]3392<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3393instruction and the variable argument handling intrinsic functions are
3394used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3395
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3396<pre>
3397int %test(int %X, ...) {
3398 ; Initialize variable argument processing
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3399 %ap = alloca sbyte*
3400 call void %<a href="#i_va_start">llvm.va_start</a>(sbyte** %ap)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3401
3402 ; Read a single integer argument
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3403 %tmp = va_arg sbyte** %ap, int
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3404
3405 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3406 %aq = alloca sbyte*
Andrew Lenharthd0a4c622005-06-22 20:38:11 +0000[diff] [blame]3407 call void %<a href="#i_va_copy">llvm.va_copy</a>(sbyte** %aq, sbyte** %ap)
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3408 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %aq)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3409
3410 ; Stop processing of arguments.
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3411 call void %<a href="#i_va_end">llvm.va_end</a>(sbyte** %ap)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3412 ret int %tmp
3413}
3414</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3415</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3416
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3417<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3418<div class="doc_subsubsection">
3419 <a name="i_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
3420</div>
3421
3422
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3423<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3424<h5>Syntax:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3425<pre> declare void %llvm.va_start(&lt;va_list&gt;* &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3426<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3427<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3428<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
3429href="#i_va_arg">va_arg</a></tt>.</p>
3430
3431<h5>Arguments:</h5>
3432
3433<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3434
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3435<h5>Semantics:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3436
3437<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3438macro available in C. In a target-dependent way, it initializes the
3439<tt>va_list</tt> element the argument points to, so that the next call to
3440<tt>va_arg</tt> will produce the first variable argument passed to the function.
3441Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
3442last argument of the function, the compiler can figure that out.</p>
3443
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3444</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3445
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3446<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3447<div class="doc_subsubsection">
3448 <a name="i_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
3449</div>
3450
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3451<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3452<h5>Syntax:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3453<pre> declare void %llvm.va_end(&lt;va_list*&gt; &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3454<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3455<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>&lt;arglist&gt;</tt>
3456which has been initialized previously with <tt><a href="#i_va_start">llvm.va_start</a></tt>
3457or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3458<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3459<p>The argument is a <tt>va_list</tt> to destroy.</p>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3460<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3461<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3462macro available in C. In a target-dependent way, it destroys the <tt>va_list</tt>.
3463Calls to <a href="#i_va_start"><tt>llvm.va_start</tt></a> and <a
3464 href="#i_va_copy"><tt>llvm.va_copy</tt></a> must be matched exactly
3465with calls to <tt>llvm.va_end</tt>.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3466</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3467
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3468<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3469<div class="doc_subsubsection">
3470 <a name="i_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
3471</div>
3472
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3473<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3474
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3475<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3476
3477<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3478 declare void %llvm.va_copy(&lt;va_list&gt;* &lt;destarglist&gt;,
Andrew Lenharthd0a4c622005-06-22 20:38:11 +0000[diff] [blame]3479 &lt;va_list&gt;* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3480</pre>
3481
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3482<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3483
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3484<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position from
3485the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3486
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3487<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3488
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3489<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Andrew Lenharthd0a4c622005-06-22 20:38:11 +0000[diff] [blame]3490The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3491
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3492
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3493<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3494
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3495<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt> macro
3496available in C. In a target-dependent way, it copies the source
3497<tt>va_list</tt> element into the destination list. This intrinsic is necessary
3498because the <tt><a href="i_va_begin">llvm.va_begin</a></tt> intrinsic may be
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3499arbitrarily complex and require memory allocation, for example.</p>
3500
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3501</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3502
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3503<!-- ======================================================================= -->
3504<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3505 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3506</div>
3507
3508<div class="doc_text">
3509
3510<p>
3511LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3512Collection</a> requires the implementation and generation of these intrinsics.
3513These intrinsics allow identification of <a href="#i_gcroot">GC roots on the
3514stack</a>, as well as garbage collector implementations that require <a
3515href="#i_gcread">read</a> and <a href="#i_gcwrite">write</a> barriers.
3516Front-ends for type-safe garbage collected languages should generate these
3517intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3518href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3519</p>
3520</div>
3521
3522<!-- _______________________________________________________________________ -->
3523<div class="doc_subsubsection">
3524 <a name="i_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
3525</div>
3526
3527<div class="doc_text">
3528
3529<h5>Syntax:</h5>
3530
3531<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +0000[diff] [blame]3532 declare void %llvm.gcroot(&lt;ty&gt;** %ptrloc, &lt;ty2&gt;* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3533</pre>
3534
3535<h5>Overview:</h5>
3536
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]3537<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3538the code generator, and allows some metadata to be associated with it.</p>
3539
3540<h5>Arguments:</h5>
3541
3542<p>The first argument specifies the address of a stack object that contains the
3543root pointer. The second pointer (which must be either a constant or a global
3544value address) contains the meta-data to be associated with the root.</p>
3545
3546<h5>Semantics:</h5>
3547
3548<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3549location. At compile-time, the code generator generates information to allow
3550the runtime to find the pointer at GC safe points.
3551</p>
3552
3553</div>
3554
3555
3556<!-- _______________________________________________________________________ -->
3557<div class="doc_subsubsection">
3558 <a name="i_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
3559</div>
3560
3561<div class="doc_text">
3562
3563<h5>Syntax:</h5>
3564
3565<pre>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3566 declare sbyte* %llvm.gcread(sbyte* %ObjPtr, sbyte** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3567</pre>
3568
3569<h5>Overview:</h5>
3570
3571<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3572locations, allowing garbage collector implementations that require read
3573barriers.</p>
3574
3575<h5>Arguments:</h5>
3576
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3577<p>The second argument is the address to read from, which should be an address
3578allocated from the garbage collector. The first object is a pointer to the
3579start of the referenced object, if needed by the language runtime (otherwise
3580null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3581
3582<h5>Semantics:</h5>
3583
3584<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3585instruction, but may be replaced with substantially more complex code by the
3586garbage collector runtime, as needed.</p>
3587
3588</div>
3589
3590
3591<!-- _______________________________________________________________________ -->
3592<div class="doc_subsubsection">
3593 <a name="i_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
3594</div>
3595
3596<div class="doc_text">
3597
3598<h5>Syntax:</h5>
3599
3600<pre>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3601 declare void %llvm.gcwrite(sbyte* %P1, sbyte* %Obj, sbyte** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3602</pre>
3603
3604<h5>Overview:</h5>
3605
3606<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3607locations, allowing garbage collector implementations that require write
3608barriers (such as generational or reference counting collectors).</p>
3609
3610<h5>Arguments:</h5>
3611
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3612<p>The first argument is the reference to store, the second is the start of the
3613object to store it to, and the third is the address of the field of Obj to
3614store to. If the runtime does not require a pointer to the object, Obj may be
3615null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3616
3617<h5>Semantics:</h5>
3618
3619<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3620instruction, but may be replaced with substantially more complex code by the
3621garbage collector runtime, as needed.</p>
3622
3623</div>
3624
3625
3626
3627<!-- ======================================================================= -->
3628<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3629 <a name="int_codegen">Code Generator Intrinsics</a>
3630</div>
3631
3632<div class="doc_text">
3633<p>
3634These intrinsics are provided by LLVM to expose special features that may only
3635be implemented with code generator support.
3636</p>
3637
3638</div>
3639
3640<!-- _______________________________________________________________________ -->
3641<div class="doc_subsubsection">
3642 <a name="i_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
3643</div>
3644
3645<div class="doc_text">
3646
3647<h5>Syntax:</h5>
3648<pre>
Chris Lattnerfcf39d42006-01-13 01:20:27 +0000[diff] [blame]3649 declare sbyte *%llvm.returnaddress(uint &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3650</pre>
3651
3652<h5>Overview:</h5>
3653
3654<p>
Chris Lattner32b5d712006-10-15 20:05:59 +0000[diff] [blame]3655The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
3656target-specific value indicating the return address of the current function
3657or one of its callers.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3658</p>
3659
3660<h5>Arguments:</h5>
3661
3662<p>
3663The argument to this intrinsic indicates which function to return the address
3664for. Zero indicates the calling function, one indicates its caller, etc. The
3665argument is <b>required</b> to be a constant integer value.
3666</p>
3667
3668<h5>Semantics:</h5>
3669
3670<p>
3671The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
3672the return address of the specified call frame, or zero if it cannot be
3673identified. The value returned by this intrinsic is likely to be incorrect or 0
3674for arguments other than zero, so it should only be used for debugging purposes.
3675</p>
3676
3677<p>
3678Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +0000[diff] [blame]3679aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3680source-language caller.
3681</p>
3682</div>
3683
3684
3685<!-- _______________________________________________________________________ -->
3686<div class="doc_subsubsection">
3687 <a name="i_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
3688</div>
3689
3690<div class="doc_text">
3691
3692<h5>Syntax:</h5>
3693<pre>
Chris Lattnerfcf39d42006-01-13 01:20:27 +0000[diff] [blame]3694 declare sbyte *%llvm.frameaddress(uint &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3695</pre>
3696
3697<h5>Overview:</h5>
3698
3699<p>
Chris Lattner32b5d712006-10-15 20:05:59 +0000[diff] [blame]3700The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
3701target-specific frame pointer value for the specified stack frame.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3702</p>
3703
3704<h5>Arguments:</h5>
3705
3706<p>
3707The argument to this intrinsic indicates which function to return the frame
3708pointer for. Zero indicates the calling function, one indicates its caller,
3709etc. The argument is <b>required</b> to be a constant integer value.
3710</p>
3711
3712<h5>Semantics:</h5>
3713
3714<p>
3715The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
3716the frame address of the specified call frame, or zero if it cannot be
3717identified. The value returned by this intrinsic is likely to be incorrect or 0
3718for arguments other than zero, so it should only be used for debugging purposes.
3719</p>
3720
3721<p>
3722Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +0000[diff] [blame]3723aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3724source-language caller.
3725</p>
3726</div>
3727
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3728<!-- _______________________________________________________________________ -->
3729<div class="doc_subsubsection">
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]3730 <a name="i_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
3731</div>
3732
3733<div class="doc_text">
3734
3735<h5>Syntax:</h5>
3736<pre>
3737 declare sbyte *%llvm.stacksave()
3738</pre>
3739
3740<h5>Overview:</h5>
3741
3742<p>
3743The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
3744the function stack, for use with <a href="#i_stackrestore">
3745<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
3746features like scoped automatic variable sized arrays in C99.
3747</p>
3748
3749<h5>Semantics:</h5>
3750
3751<p>
3752This intrinsic returns a opaque pointer value that can be passed to <a
3753href="#i_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
3754<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
3755<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
3756state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
3757practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
3758that were allocated after the <tt>llvm.stacksave</tt> was executed.
3759</p>
3760
3761</div>
3762
3763<!-- _______________________________________________________________________ -->
3764<div class="doc_subsubsection">
3765 <a name="i_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
3766</div>
3767
3768<div class="doc_text">
3769
3770<h5>Syntax:</h5>
3771<pre>
3772 declare void %llvm.stackrestore(sbyte* %ptr)
3773</pre>
3774
3775<h5>Overview:</h5>
3776
3777<p>
3778The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
3779the function stack to the state it was in when the corresponding <a
3780href="#llvm.stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
3781useful for implementing language features like scoped automatic variable sized
3782arrays in C99.
3783</p>
3784
3785<h5>Semantics:</h5>
3786
3787<p>
3788See the description for <a href="#i_stacksave"><tt>llvm.stacksave</tt></a>.
3789</p>
3790
3791</div>
3792
3793
3794<!-- _______________________________________________________________________ -->
3795<div class="doc_subsubsection">
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3796 <a name="i_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
3797</div>
3798
3799<div class="doc_text">
3800
3801<h5>Syntax:</h5>
3802<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +0000[diff] [blame]3803 declare void %llvm.prefetch(sbyte * &lt;address&gt;,
3804 uint &lt;rw&gt;, uint &lt;locality&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3805</pre>
3806
3807<h5>Overview:</h5>
3808
3809
3810<p>
3811The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3812a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
3813no
3814effect on the behavior of the program but can change its performance
Chris Lattner2a615362005-02-28 19:47:14 +0000[diff] [blame]3815characteristics.
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3816</p>
3817
3818<h5>Arguments:</h5>
3819
3820<p>
3821<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
3822determining if the fetch should be for a read (0) or write (1), and
3823<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Chris Lattneraeffb4a2005-03-07 20:31:38 +0000[diff] [blame]3824locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]3825<tt>locality</tt> arguments must be constant integers.
3826</p>
3827
3828<h5>Semantics:</h5>
3829
3830<p>
3831This intrinsic does not modify the behavior of the program. In particular,
3832prefetches cannot trap and do not produce a value. On targets that support this
3833intrinsic, the prefetch can provide hints to the processor cache for better
3834performance.
3835</p>
3836
3837</div>
3838
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]3839<!-- _______________________________________________________________________ -->
3840<div class="doc_subsubsection">
3841 <a name="i_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
3842</div>
3843
3844<div class="doc_text">
3845
3846<h5>Syntax:</h5>
3847<pre>
Reid Spencera8d451e2005-04-26 20:50:44 +0000[diff] [blame]3848 declare void %llvm.pcmarker( uint &lt;id&gt; )
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]3849</pre>
3850
3851<h5>Overview:</h5>
3852
3853
3854<p>
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3855The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
3856(PC) in a region of
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]3857code to simulators and other tools. The method is target specific, but it is
3858expected that the marker will use exported symbols to transmit the PC of the marker.
Jeff Cohen25d4f7e2005-11-11 02:15:27 +0000[diff] [blame]3859The marker makes no guarantees that it will remain with any specific instruction
Chris Lattnerd07c3f42005-11-15 06:07:55 +0000[diff] [blame]3860after optimizations. It is possible that the presence of a marker will inhibit
Chris Lattnerb3e7afd2006-03-24 07:16:10 +0000[diff] [blame]3861optimizations. The intended use is to be inserted after optimizations to allow
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3862correlations of simulation runs.
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]3863</p>
3864
3865<h5>Arguments:</h5>
3866
3867<p>
3868<tt>id</tt> is a numerical id identifying the marker.
3869</p>
3870
3871<h5>Semantics:</h5>
3872
3873<p>
3874This intrinsic does not modify the behavior of the program. Backends that do not
3875support this intrinisic may ignore it.
3876</p>
3877
3878</div>
3879
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]3880<!-- _______________________________________________________________________ -->
3881<div class="doc_subsubsection">
3882 <a name="i_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
3883</div>
3884
3885<div class="doc_text">
3886
3887<h5>Syntax:</h5>
3888<pre>
3889 declare ulong %llvm.readcyclecounter( )
3890</pre>
3891
3892<h5>Overview:</h5>
3893
3894
3895<p>
3896The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
3897counter register (or similar low latency, high accuracy clocks) on those targets
3898that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
3899As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
3900should only be used for small timings.
3901</p>
3902
3903<h5>Semantics:</h5>
3904
3905<p>
3906When directly supported, reading the cycle counter should not modify any memory.
3907Implementations are allowed to either return a application specific value or a
3908system wide value. On backends without support, this is lowered to a constant 0.
3909</p>
3910
3911</div>
3912
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3913<!-- ======================================================================= -->
3914<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3915 <a name="int_libc">Standard C Library Intrinsics</a>
3916</div>
3917
3918<div class="doc_text">
3919<p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3920LLVM provides intrinsics for a few important standard C library functions.
3921These intrinsics allow source-language front-ends to pass information about the
3922alignment of the pointer arguments to the code generator, providing opportunity
3923for more efficient code generation.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3924</p>
3925
3926</div>
3927
3928<!-- _______________________________________________________________________ -->
3929<div class="doc_subsubsection">
3930 <a name="i_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
3931</div>
3932
3933<div class="doc_text">
3934
3935<h5>Syntax:</h5>
3936<pre>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]3937 declare void %llvm.memcpy.i32(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3938 uint &lt;len&gt;, uint &lt;align&gt;)
3939 declare void %llvm.memcpy.i64(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3940 ulong &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3941</pre>
3942
3943<h5>Overview:</h5>
3944
3945<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]3946The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3947location to the destination location.
3948</p>
3949
3950<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]3951Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
3952intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3953</p>
3954
3955<h5>Arguments:</h5>
3956
3957<p>
3958The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]3959the source. The third argument is an integer argument
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3960specifying the number of bytes to copy, and the fourth argument is the alignment
3961of the source and destination locations.
3962</p>
3963
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]3964<p>
3965If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattnerf0afc2c2006-03-04 00:02:10 +0000[diff] [blame]3966the caller guarantees that both the source and destination pointers are aligned
3967to that boundary.
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]3968</p>
3969
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3970<h5>Semantics:</h5>
3971
3972<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]3973The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3974location to the destination location, which are not allowed to overlap. It
3975copies "len" bytes of memory over. If the argument is known to be aligned to
3976some boundary, this can be specified as the fourth argument, otherwise it should
3977be set to 0 or 1.
3978</p>
3979</div>
3980
3981
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]3982<!-- _______________________________________________________________________ -->
3983<div class="doc_subsubsection">
3984 <a name="i_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
3985</div>
3986
3987<div class="doc_text">
3988
3989<h5>Syntax:</h5>
3990<pre>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]3991 declare void %llvm.memmove.i32(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3992 uint &lt;len&gt;, uint &lt;align&gt;)
3993 declare void %llvm.memmove.i64(sbyte* &lt;dest&gt;, sbyte* &lt;src&gt;,
3994 ulong &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]3995</pre>
3996
3997<h5>Overview:</h5>
3998
3999<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4000The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4001location to the destination location. It is similar to the
4002'<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4003</p>
4004
4005<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4006Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4007intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4008</p>
4009
4010<h5>Arguments:</h5>
4011
4012<p>
4013The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4014the source. The third argument is an integer argument
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4015specifying the number of bytes to copy, and the fourth argument is the alignment
4016of the source and destination locations.
4017</p>
4018
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4019<p>
4020If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattnerf0afc2c2006-03-04 00:02:10 +0000[diff] [blame]4021the caller guarantees that the source and destination pointers are aligned to
4022that boundary.
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4023</p>
4024
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4025<h5>Semantics:</h5>
4026
4027<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4028The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4029location to the destination location, which may overlap. It
4030copies "len" bytes of memory over. If the argument is known to be aligned to
4031some boundary, this can be specified as the fourth argument, otherwise it should
4032be set to 0 or 1.
4033</p>
4034</div>
4035
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]4036
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4037<!-- _______________________________________________________________________ -->
4038<div class="doc_subsubsection">
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4039 <a name="i_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4040</div>
4041
4042<div class="doc_text">
4043
4044<h5>Syntax:</h5>
4045<pre>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4046 declare void %llvm.memset.i32(sbyte* &lt;dest&gt;, ubyte &lt;val&gt;,
4047 uint &lt;len&gt;, uint &lt;align&gt;)
4048 declare void %llvm.memset.i64(sbyte* &lt;dest&gt;, ubyte &lt;val&gt;,
4049 ulong &lt;len&gt;, uint &lt;align&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4050</pre>
4051
4052<h5>Overview:</h5>
4053
4054<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4055The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4056byte value.
4057</p>
4058
4059<p>
4060Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4061does not return a value, and takes an extra alignment argument.
4062</p>
4063
4064<h5>Arguments:</h5>
4065
4066<p>
4067The first argument is a pointer to the destination to fill, the second is the
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4068byte value to fill it with, the third argument is an integer
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4069argument specifying the number of bytes to fill, and the fourth argument is the
4070known alignment of destination location.
4071</p>
4072
4073<p>
4074If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattnerf0afc2c2006-03-04 00:02:10 +0000[diff] [blame]4075the caller guarantees that the destination pointer is aligned to that boundary.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4076</p>
4077
4078<h5>Semantics:</h5>
4079
4080<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4081The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4082the
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4083destination location. If the argument is known to be aligned to some boundary,
4084this can be specified as the fourth argument, otherwise it should be set to 0 or
40851.
4086</p>
4087</div>
4088
4089
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]4090<!-- _______________________________________________________________________ -->
4091<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4092 <a name="i_isunordered">'<tt>llvm.isunordered.*</tt>' Intrinsic</a>
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +0000[diff] [blame]4093</div>
4094
4095<div class="doc_text">
4096
4097<h5>Syntax:</h5>
4098<pre>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4099 declare bool %llvm.isunordered.f32(float Val1, float Val2)
4100 declare bool %llvm.isunordered.f64(double Val1, double Val2)
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +0000[diff] [blame]4101</pre>
4102
4103<h5>Overview:</h5>
4104
4105<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4106The '<tt>llvm.isunordered</tt>' intrinsics return true if either or both of the
Alkis Evlogimenos26bbe932004-06-13 01:16:15 +0000[diff] [blame]4107specified floating point values is a NAN.
4108</p>
4109
4110<h5>Arguments:</h5>
4111
4112<p>
4113The arguments are floating point numbers of the same type.
4114</p>
4115
4116<h5>Semantics:</h5>
4117
4118<p>
4119If either or both of the arguments is a SNAN or QNAN, it returns true, otherwise
4120false.
4121</p>
4122</div>
4123
4124
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4125<!-- _______________________________________________________________________ -->
4126<div class="doc_subsubsection">
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4127 <a name="i_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4128</div>
4129
4130<div class="doc_text">
4131
4132<h5>Syntax:</h5>
4133<pre>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4134 declare float %llvm.sqrt.f32(float %Val)
4135 declare double %llvm.sqrt.f64(double %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4136</pre>
4137
4138<h5>Overview:</h5>
4139
4140<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4141The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4142returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4143<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4144negative numbers (which allows for better optimization).
4145</p>
4146
4147<h5>Arguments:</h5>
4148
4149<p>
4150The argument and return value are floating point numbers of the same type.
4151</p>
4152
4153<h5>Semantics:</h5>
4154
4155<p>
4156This function returns the sqrt of the specified operand if it is a positive
4157floating point number.
4158</p>
4159</div>
4160
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4161<!-- _______________________________________________________________________ -->
4162<div class="doc_subsubsection">
4163 <a name="i_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
4164</div>
4165
4166<div class="doc_text">
4167
4168<h5>Syntax:</h5>
4169<pre>
4170 declare float %llvm.powi.f32(float %Val, int %power)
4171 declare double %llvm.powi.f64(double %Val, int %power)
4172</pre>
4173
4174<h5>Overview:</h5>
4175
4176<p>
4177The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4178specified (positive or negative) power. The order of evaluation of
4179multiplications is not defined.
4180</p>
4181
4182<h5>Arguments:</h5>
4183
4184<p>
4185The second argument is an integer power, and the first is a value to raise to
4186that power.
4187</p>
4188
4189<h5>Semantics:</h5>
4190
4191<p>
4192This function returns the first value raised to the second power with an
4193unspecified sequence of rounding operations.</p>
4194</div>
4195
4196
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4197<!-- ======================================================================= -->
4198<div class="doc_subsection">
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4199 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4200</div>
4201
4202<div class="doc_text">
4203<p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4204LLVM provides intrinsics for a few important bit manipulation operations.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4205These allow efficient code generation for some algorithms.
4206</p>
4207
4208</div>
4209
4210<!-- _______________________________________________________________________ -->
4211<div class="doc_subsubsection">
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4212 <a name="i_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
4213</div>
4214
4215<div class="doc_text">
4216
4217<h5>Syntax:</h5>
4218<pre>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4219 declare ushort %llvm.bswap.i16(ushort &lt;id&gt;)
4220 declare uint %llvm.bswap.i32(uint &lt;id&gt;)
4221 declare ulong %llvm.bswap.i64(ulong &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4222</pre>
4223
4224<h5>Overview:</h5>
4225
4226<p>
4227The '<tt>llvm.bwsap</tt>' family of intrinsics is used to byteswap a 16, 32 or
422864 bit quantity. These are useful for performing operations on data that is not
4229in the target's native byte order.
4230</p>
4231
4232<h5>Semantics:</h5>
4233
4234<p>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4235The <tt>llvm.bswap.16</tt> intrinsic returns a ushort value that has the high and low
4236byte of the input ushort swapped. Similarly, the <tt>llvm.bswap.i32</tt> intrinsic
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4237returns a uint value that has the four bytes of the input uint swapped, so that
4238if the input bytes are numbered 0, 1, 2, 3 then the returned uint will have its
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4239bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i64</tt> intrinsic extends this concept
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4240to 64 bits.
4241</p>
4242
4243</div>
4244
4245<!-- _______________________________________________________________________ -->
4246<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4247 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4248</div>
4249
4250<div class="doc_text">
4251
4252<h5>Syntax:</h5>
4253<pre>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4254 declare ubyte %llvm.ctpop.i8 (ubyte &lt;src&gt;)
4255 declare ushort %llvm.ctpop.i16(ushort &lt;src&gt;)
4256 declare uint %llvm.ctpop.i32(uint &lt;src&gt;)
4257 declare ulong %llvm.ctpop.i64(ulong &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4258</pre>
4259
4260<h5>Overview:</h5>
4261
4262<p>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4263The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4264value.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4265</p>
4266
4267<h5>Arguments:</h5>
4268
4269<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]4270The only argument is the value to be counted. The argument may be of any
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4271unsigned integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4272</p>
4273
4274<h5>Semantics:</h5>
4275
4276<p>
4277The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4278</p>
4279</div>
4280
4281<!-- _______________________________________________________________________ -->
4282<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]4283 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4284</div>
4285
4286<div class="doc_text">
4287
4288<h5>Syntax:</h5>
4289<pre>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4290 declare ubyte %llvm.ctlz.i8 (ubyte &lt;src&gt;)
4291 declare ushort %llvm.ctlz.i16(ushort &lt;src&gt;)
4292 declare uint %llvm.ctlz.i32(uint &lt;src&gt;)
4293 declare ulong %llvm.ctlz.i64(ulong &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4294</pre>
4295
4296<h5>Overview:</h5>
4297
4298<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4299The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4300leading zeros in a variable.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4301</p>
4302
4303<h5>Arguments:</h5>
4304
4305<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]4306The only argument is the value to be counted. The argument may be of any
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4307unsigned integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4308</p>
4309
4310<h5>Semantics:</h5>
4311
4312<p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4313The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4314in a variable. If the src == 0 then the result is the size in bits of the type
Chris Lattner99d3c272006-04-21 21:37:40 +0000[diff] [blame]4315of src. For example, <tt>llvm.ctlz(int 2) = 30</tt>.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4316</p>
4317</div>
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]4318
4319
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4320
4321<!-- _______________________________________________________________________ -->
4322<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]4323 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4324</div>
4325
4326<div class="doc_text">
4327
4328<h5>Syntax:</h5>
4329<pre>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4330 declare ubyte %llvm.cttz.i8 (ubyte &lt;src&gt;)
4331 declare ushort %llvm.cttz.i16(ushort &lt;src&gt;)
4332 declare uint %llvm.cttz.i32(uint &lt;src&gt;)
4333 declare ulong %llvm.cttz.i64(ulong &lt;src&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4334</pre>
4335
4336<h5>Overview:</h5>
4337
4338<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4339The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4340trailing zeros.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4341</p>
4342
4343<h5>Arguments:</h5>
4344
4345<p>
4346The only argument is the value to be counted. The argument may be of any
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4347unsigned integer type. The return type must match the argument type.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4348</p>
4349
4350<h5>Semantics:</h5>
4351
4352<p>
4353The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4354in a variable. If the src == 0 then the result is the size in bits of the type
4355of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4356</p>
4357</div>
4358
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]4359<!-- ======================================================================= -->
4360<div class="doc_subsection">
4361 <a name="int_debugger">Debugger Intrinsics</a>
4362</div>
4363
4364<div class="doc_text">
4365<p>
4366The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4367are described in the <a
4368href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4369Debugging</a> document.
4370</p>
4371</div>
4372
4373
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4374<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4375<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]4376<address>
4377 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
4378 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
4379 <a href="http://validator.w3.org/check/referer"><img
4380 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
4381
4382 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]4383 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]4384 Last modified: $Date$
4385</address>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4386</body>
4387</html>