Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 2067764e29dc2fd6d4d4b48940f41744eaff6d14 / . / docs / LangRef.html
blob: fa7fefafc4731658bc2b4d05edef6bfd385c9fc2 [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>
Anton Korobeynikovc6c98af2007-04-29 18:02:48 +0000[diff] [blame]27 <li><a href="#aliasstructure">Aliases</a>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]28 <li><a href="#paramattrs">Parameter Attributes</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]29 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]30 <li><a href="#datalayout">Data Layout</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]31 </ol>
32 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]33 <li><a href="#typesystem">Type System</a>
34 <ol>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]35 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]36 <ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]37 <li><a href="#t_classifications">Type Classifications</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]38 </ol>
39 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]40 <li><a href="#t_derived">Derived Types</a>
41 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]42 <li><a href="#t_array">Array Type</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]43 <li><a href="#t_function">Function Type</a></li>
44 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]45 <li><a href="#t_struct">Structure Type</a></li>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]46 <li><a href="#t_pstruct">Packed Structure Type</a></li>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]47 <li><a href="#t_vector">Vector Type</a></li>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]48 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]49 </ol>
50 </li>
51 </ol>
52 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]53 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]54 <ol>
55 <li><a href="#simpleconstants">Simple Constants</a>
56 <li><a href="#aggregateconstants">Aggregate Constants</a>
57 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
58 <li><a href="#undefvalues">Undefined Values</a>
59 <li><a href="#constantexprs">Constant Expressions</a>
60 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]61 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]62 <li><a href="#othervalues">Other Values</a>
63 <ol>
64 <li><a href="#inlineasm">Inline Assembler Expressions</a>
65 </ol>
66 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]67 <li><a href="#instref">Instruction Reference</a>
68 <ol>
69 <li><a href="#terminators">Terminator Instructions</a>
70 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]71 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
72 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]73 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
74 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]75 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]76 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]77 </ol>
78 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]79 <li><a href="#binaryops">Binary Operations</a>
80 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]81 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
82 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
83 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]84 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
85 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
86 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]87 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
88 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
89 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]90 </ol>
91 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]92 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
93 <ol>
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]94 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
95 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
96 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]97 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]98 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]99 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]100 </ol>
101 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]102 <li><a href="#vectorops">Vector Operations</a>
103 <ol>
104 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
105 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
106 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]107 </ol>
108 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]109 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]110 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]111 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
112 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
113 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]114 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
115 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
116 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]117 </ol>
118 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]119 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]120 <ol>
121 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
122 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
123 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
124 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
125 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]126 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
127 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
128 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
129 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]130 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
131 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]132 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]133 </ol>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]134 <li><a href="#otherops">Other Operations</a>
135 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]136 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
137 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]138 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]139 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]140 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]141 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]142 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]143 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]144 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]145 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]146 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]147 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]148 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
149 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]150 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
151 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
152 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]153 </ol>
154 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]155 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
156 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]157 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
158 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
159 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]160 </ol>
161 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]162 <li><a href="#int_codegen">Code Generator Intrinsics</a>
163 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]164 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
165 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
166 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
167 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
168 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
169 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
170 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000[diff] [blame]171 </ol>
172 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]173 <li><a href="#int_libc">Standard C Library Intrinsics</a>
174 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]175 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
176 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
177 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
178 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
179 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]180 </ol>
181 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]182 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]183 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]184 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]185 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
186 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
187 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]188 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
189 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]190 </ol>
191 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]192 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]193 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]194 <li><a href="#int_atomics">Atomic Operations and Synchronization Intrinsics</a>
195 <ol>
196 <li><a href="#int_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_ls">'<tt>llvm.atomic.ls.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a></li>
199 <li><a href="#int_lss">'<tt>llvm.atomic.lss.*</tt>' Intrinsic</a></li>
200 <li><a href="#int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a></li>
201 </ol>
202 </li>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]203 <li><a href="#int_general">General intrinsics</a>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]204 <ol>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]205 <li><a href="#int_var_annotation">
206 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
207 </ol>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]208 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]209 </ol>
210 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]211</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]212
213<div class="doc_author">
214 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
215 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]216</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]217
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]218<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]219<div class="doc_section"> <a name="abstract">Abstract </a></div>
220<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]221
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]222<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]223<p>This document is a reference manual for the LLVM assembly language.
224LLVM is an SSA based representation that provides type safety,
225low-level operations, flexibility, and the capability of representing
226'all' high-level languages cleanly. It is the common code
227representation used throughout all phases of the LLVM compilation
228strategy.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]229</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]230
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]231<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]232<div class="doc_section"> <a name="introduction">Introduction</a> </div>
233<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]234
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]235<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]236
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]237<p>The LLVM code representation is designed to be used in three
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]238different forms: as an in-memory compiler IR, as an on-disk bitcode
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]239representation (suitable for fast loading by a Just-In-Time compiler),
240and as a human readable assembly language representation. This allows
241LLVM to provide a powerful intermediate representation for efficient
242compiler transformations and analysis, while providing a natural means
243to debug and visualize the transformations. The three different forms
244of LLVM are all equivalent. This document describes the human readable
245representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]246
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]247<p>The LLVM representation aims to be light-weight and low-level
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]248while being expressive, typed, and extensible at the same time. It
249aims to be a "universal IR" of sorts, by being at a low enough level
250that high-level ideas may be cleanly mapped to it (similar to how
251microprocessors are "universal IR's", allowing many source languages to
252be mapped to them). By providing type information, LLVM can be used as
253the target of optimizations: for example, through pointer analysis, it
254can be proven that a C automatic variable is never accessed outside of
255the current function... allowing it to be promoted to a simple SSA
256value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]257
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]258</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]259
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]260<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]261<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]262
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]263<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]264
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]265<p>It is important to note that this document describes 'well formed'
266LLVM assembly language. There is a difference between what the parser
267accepts and what is considered 'well formed'. For example, the
268following instruction is syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]269
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]270<div class="doc_code">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]271<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]272%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]273</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]274</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]275
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]276<p>...because the definition of <tt>%x</tt> does not dominate all of
277its uses. The LLVM infrastructure provides a verification pass that may
278be used to verify that an LLVM module is well formed. This pass is
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]279automatically run by the parser after parsing input assembly and by
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]280the optimizer before it outputs bitcode. The violations pointed out
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]281by the verifier pass indicate bugs in transformation passes or input to
282the parser.</p>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]283</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]284
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]285<!-- Describe the typesetting conventions here. -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]286
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]287<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]288<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]289<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]290
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]291<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]292
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]293<p>LLVM uses three different forms of identifiers, for different
294purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]295
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]296<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]297 <li>Named values are represented as a string of characters with a '%' prefix.
298 For example, %foo, %DivisionByZero, %a.really.long.identifier. The actual
299 regular expression used is '<tt>%[a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
300 Identifiers which require other characters in their names can be surrounded
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]301 with quotes. In this way, anything except a <tt>&quot;</tt> character can be used
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]302 in a name.</li>
303
304 <li>Unnamed values are represented as an unsigned numeric value with a '%'
305 prefix. For example, %12, %2, %44.</li>
306
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]307 <li>Constants, which are described in a <a href="#constants">section about
308 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]309</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]310
311<p>LLVM requires that values start with a '%' sign for two reasons: Compilers
312don't need to worry about name clashes with reserved words, and the set of
313reserved words may be expanded in the future without penalty. Additionally,
314unnamed identifiers allow a compiler to quickly come up with a temporary
315variable without having to avoid symbol table conflicts.</p>
316
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]317<p>Reserved words in LLVM are very similar to reserved words in other
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]318languages. There are keywords for different opcodes
319('<tt><a href="#i_add">add</a></tt>',
320 '<tt><a href="#i_bitcast">bitcast</a></tt>',
321 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]322href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]323and others. These reserved words cannot conflict with variable names, because
324none of them start with a '%' character.</p>
325
326<p>Here is an example of LLVM code to multiply the integer variable
327'<tt>%X</tt>' by 8:</p>
328
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]329<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]330
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]331<div class="doc_code">
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]332<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]333%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]334</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]335</div>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]336
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]337<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]338
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]339<div class="doc_code">
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]340<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]341%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]342</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]343</div>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]344
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]345<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]346
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]347<div class="doc_code">
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]348<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]349<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
350<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
351%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]352</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]353</div>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]354
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]355<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
356important lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]357
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]358<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]359
360 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
361 line.</li>
362
363 <li>Unnamed temporaries are created when the result of a computation is not
364 assigned to a named value.</li>
365
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]366 <li>Unnamed temporaries are numbered sequentially</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]367
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]368</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]369
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]370<p>...and it also shows a convention that we follow in this document. When
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]371demonstrating instructions, we will follow an instruction with a comment that
372defines the type and name of value produced. Comments are shown in italic
373text.</p>
374
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]375</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]376
377<!-- *********************************************************************** -->
378<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
379<!-- *********************************************************************** -->
380
381<!-- ======================================================================= -->
382<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
383</div>
384
385<div class="doc_text">
386
387<p>LLVM programs are composed of "Module"s, each of which is a
388translation unit of the input programs. Each module consists of
389functions, global variables, and symbol table entries. Modules may be
390combined together with the LLVM linker, which merges function (and
391global variable) definitions, resolves forward declarations, and merges
392symbol table entries. Here is an example of the "hello world" module:</p>
393
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]394<div class="doc_code">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]395<pre><i>; Declare the string constant as a global constant...</i>
Chris Lattnera89e5f12007-06-12 17:00:26 +0000[diff] [blame]396<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
397 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]398
399<i>; External declaration of the puts function</i>
Chris Lattnera89e5f12007-06-12 17:00:26 +0000[diff] [blame]400<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]401
402<i>; Definition of main function</i>
Chris Lattnera89e5f12007-06-12 17:00:26 +0000[diff] [blame]403define i32 @main() { <i>; i32()* </i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]404 <i>; Convert [13x i8 ]* to i8 *...</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]405 %cast210 = <a
Chris Lattner6c0955b2007-06-12 17:01:15 +0000[diff] [blame]406 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]407
408 <i>; Call puts function to write out the string to stdout...</i>
409 <a
Chris Lattnera89e5f12007-06-12 17:00:26 +0000[diff] [blame]410 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]411 <a
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]412 href="#i_ret">ret</a> i32 0<br>}<br>
413</pre>
414</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]415
416<p>This example is made up of a <a href="#globalvars">global variable</a>
417named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
418function, and a <a href="#functionstructure">function definition</a>
419for "<tt>main</tt>".</p>
420
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]421<p>In general, a module is made up of a list of global values,
422where both functions and global variables are global values. Global values are
423represented by a pointer to a memory location (in this case, a pointer to an
424array of char, and a pointer to a function), and have one of the following <a
425href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]426
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]427</div>
428
429<!-- ======================================================================= -->
430<div class="doc_subsection">
431 <a name="linkage">Linkage Types</a>
432</div>
433
434<div class="doc_text">
435
436<p>
437All Global Variables and Functions have one of the following types of linkage:
438</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]439
440<dl>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]441
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]442 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]443
444 <dd>Global values with internal linkage are only directly accessible by
445 objects in the current module. In particular, linking code into a module with
446 an internal global value may cause the internal to be renamed as necessary to
447 avoid collisions. Because the symbol is internal to the module, all
448 references can be updated. This corresponds to the notion of the
Chris Lattner4887bd82007-01-14 06:51:48 +0000[diff] [blame]449 '<tt>static</tt>' keyword in C.
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_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]453
Chris Lattner4887bd82007-01-14 06:51:48 +0000[diff] [blame]454 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
455 the same name when linkage occurs. This is typically used to implement
456 inline functions, templates, or other code which must be generated in each
457 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
458 allowed to be discarded.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]459 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]460
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]461 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]462
463 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
464 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
Chris Lattner4887bd82007-01-14 06:51:48 +0000[diff] [blame]465 used for globals that may be emitted in multiple translation units, but that
466 are not guaranteed to be emitted into every translation unit that uses them.
467 One example of this are common globals in C, such as "<tt>int X;</tt>" at
468 global scope.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]469 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]470
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]471 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]472
473 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
474 pointer to array type. When two global variables with appending linkage are
475 linked together, the two global arrays are appended together. This is the
476 LLVM, typesafe, equivalent of having the system linker append together
477 "sections" with identical names when .o files are linked.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]478 </dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]479
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]480 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
481 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
482 until linked, if not linked, the symbol becomes null instead of being an
483 undefined reference.
484 </dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]485
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]486 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]487
488 <dd>If none of the above identifiers are used, the global is externally
489 visible, meaning that it participates in linkage and can be used to resolve
490 external symbol references.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]491 </dd>
Reid Spencerc8910842007-04-11 23:49:50 +0000[diff] [blame]492</dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]493
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]494 <p>
495 The next two types of linkage are targeted for Microsoft Windows platform
496 only. They are designed to support importing (exporting) symbols from (to)
497 DLLs.
498 </p>
499
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]500 <dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]501 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
502
503 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
504 or variable via a global pointer to a pointer that is set up by the DLL
505 exporting the symbol. On Microsoft Windows targets, the pointer name is
506 formed by combining <code>_imp__</code> and the function or variable name.
507 </dd>
508
509 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
510
511 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
512 pointer to a pointer in a DLL, so that it can be referenced with the
513 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
514 name is formed by combining <code>_imp__</code> and the function or variable
515 name.
516 </dd>
517
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]518</dl>
519
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]520<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]521variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
522variable and was linked with this one, one of the two would be renamed,
523preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
524external (i.e., lacking any linkage declarations), they are accessible
Reid Spencerac8d2762007-01-05 00:59:10 +0000[diff] [blame]525outside of the current module.</p>
526<p>It is illegal for a function <i>declaration</i>
527to have any linkage type other than "externally visible", <tt>dllimport</tt>,
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]528or <tt>extern_weak</tt>.</p>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]529<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
530linkages.
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]531</div>
532
533<!-- ======================================================================= -->
534<div class="doc_subsection">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]535 <a name="callingconv">Calling Conventions</a>
536</div>
537
538<div class="doc_text">
539
540<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
541and <a href="#i_invoke">invokes</a> can all have an optional calling convention
542specified for the call. The calling convention of any pair of dynamic
543caller/callee must match, or the behavior of the program is undefined. The
544following calling conventions are supported by LLVM, and more may be added in
545the future:</p>
546
547<dl>
548 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
549
550 <dd>This calling convention (the default if no other calling convention is
551 specified) matches the target C calling conventions. This calling convention
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]552 supports varargs function calls and tolerates some mismatch in the declared
Reid Spencerc28d2bc2006-12-31 21:30:18 +0000[diff] [blame]553 prototype and implemented declaration of the function (as does normal C).
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]554 </dd>
555
556 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
557
558 <dd>This calling convention attempts to make calls as fast as possible
559 (e.g. by passing things in registers). This calling convention allows the
560 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner8cdc5bc2005-05-06 23:08:23 +0000[diff] [blame]561 without having to conform to an externally specified ABI. Implementations of
562 this convention should allow arbitrary tail call optimization to be supported.
563 This calling convention does not support varargs and requires the prototype of
564 all callees to exactly match the prototype of the function definition.
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]565 </dd>
566
567 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
568
569 <dd>This calling convention attempts to make code in the caller as efficient
570 as possible under the assumption that the call is not commonly executed. As
571 such, these calls often preserve all registers so that the call does not break
572 any live ranges in the caller side. This calling convention does not support
573 varargs and requires the prototype of all callees to exactly match the
574 prototype of the function definition.
575 </dd>
576
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]577 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]578
579 <dd>Any calling convention may be specified by number, allowing
580 target-specific calling conventions to be used. Target specific calling
581 conventions start at 64.
582 </dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]583</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]584
585<p>More calling conventions can be added/defined on an as-needed basis, to
586support pascal conventions or any other well-known target-independent
587convention.</p>
588
589</div>
590
591<!-- ======================================================================= -->
592<div class="doc_subsection">
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]593 <a name="visibility">Visibility Styles</a>
594</div>
595
596<div class="doc_text">
597
598<p>
599All Global Variables and Functions have one of the following visibility styles:
600</p>
601
602<dl>
603 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
604
605 <dd>On ELF, default visibility means that the declaration is visible to other
606 modules and, in shared libraries, means that the declared entity may be
607 overridden. On Darwin, default visibility means that the declaration is
608 visible to other modules. Default visibility corresponds to "external
609 linkage" in the language.
610 </dd>
611
612 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
613
614 <dd>Two declarations of an object with hidden visibility refer to the same
615 object if they are in the same shared object. Usually, hidden visibility
616 indicates that the symbol will not be placed into the dynamic symbol table,
617 so no other module (executable or shared library) can reference it
618 directly.
619 </dd>
620
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]621 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
622
623 <dd>On ELF, protected visibility indicates that the symbol will be placed in
624 the dynamic symbol table, but that references within the defining module will
625 bind to the local symbol. That is, the symbol cannot be overridden by another
626 module.
627 </dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]628</dl>
629
630</div>
631
632<!-- ======================================================================= -->
633<div class="doc_subsection">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]634 <a name="globalvars">Global Variables</a>
635</div>
636
637<div class="doc_text">
638
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]639<p>Global variables define regions of memory allocated at compilation time
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]640instead of run-time. Global variables may optionally be initialized, may have
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]641an explicit section to be placed in, and may have an optional explicit alignment
642specified. A variable may be defined as "thread_local", which means that it
643will not be shared by threads (each thread will have a separated copy of the
644variable). A variable may be defined as a global "constant," which indicates
645that the contents of the variable will <b>never</b> be modified (enabling better
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]646optimization, allowing the global data to be placed in the read-only section of
647an executable, etc). Note that variables that need runtime initialization
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]648cannot be marked "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]649
650<p>
651LLVM explicitly allows <em>declarations</em> of global variables to be marked
652constant, even if the final definition of the global is not. This capability
653can be used to enable slightly better optimization of the program, but requires
654the language definition to guarantee that optimizations based on the
655'constantness' are valid for the translation units that do not include the
656definition.
657</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]658
659<p>As SSA values, global variables define pointer values that are in
660scope (i.e. they dominate) all basic blocks in the program. Global
661variables always define a pointer to their "content" type because they
662describe a region of memory, and all memory objects in LLVM are
663accessed through pointers.</p>
664
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]665<p>LLVM allows an explicit section to be specified for globals. If the target
666supports it, it will emit globals to the section specified.</p>
667
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]668<p>An explicit alignment may be specified for a global. If not present, or if
669the alignment is set to zero, the alignment of the global is set by the target
670to whatever it feels convenient. If an explicit alignment is specified, the
671global is forced to have at least that much alignment. All alignments must be
672a power of 2.</p>
673
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]674<p>For example, the following defines a global with an initializer, section,
675 and alignment:</p>
676
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]677<div class="doc_code">
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]678<pre>
Chris Lattner3e63a9d2007-07-13 20:01:46 +0000[diff] [blame]679@G = constant float 1.0, section "foo", align 4
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]680</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]681</div>
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]682
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]683</div>
684
685
686<!-- ======================================================================= -->
687<div class="doc_subsection">
688 <a name="functionstructure">Functions</a>
689</div>
690
691<div class="doc_text">
692
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]693<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
694an optional <a href="#linkage">linkage type</a>, an optional
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]695<a href="#visibility">visibility style</a>, an optional
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]696<a href="#callingconv">calling convention</a>, a return type, an optional
697<a href="#paramattrs">parameter attribute</a> for the return type, a function
698name, a (possibly empty) argument list (each with optional
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]699<a href="#paramattrs">parameter attributes</a>), an optional section, an
700optional alignment, an opening curly brace, a list of basic blocks, and a
701closing curly brace.
702
703LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
704optional <a href="#linkage">linkage type</a>, an optional
705<a href="#visibility">visibility style</a>, an optional
706<a href="#callingconv">calling convention</a>, a return type, an optional
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]707<a href="#paramattrs">parameter attribute</a> for the return type, a function
708name, a possibly empty list of arguments, and an optional alignment.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]709
710<p>A function definition contains a list of basic blocks, forming the CFG for
711the function. Each basic block may optionally start with a label (giving the
712basic block a symbol table entry), contains a list of instructions, and ends
713with a <a href="#terminators">terminator</a> instruction (such as a branch or
714function return).</p>
715
Chris Lattner4a3c9012007-06-08 16:52:14 +0000[diff] [blame]716<p>The first basic block in a function is special in two ways: it is immediately
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]717executed on entrance to the function, and it is not allowed to have predecessor
718basic blocks (i.e. there can not be any branches to the entry block of a
719function). Because the block can have no predecessors, it also cannot have any
720<a href="#i_phi">PHI nodes</a>.</p>
721
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]722<p>LLVM allows an explicit section to be specified for functions. If the target
723supports it, it will emit functions to the section specified.</p>
724
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]725<p>An explicit alignment may be specified for a function. If not present, or if
726the alignment is set to zero, the alignment of the function is set by the target
727to whatever it feels convenient. If an explicit alignment is specified, the
728function is forced to have at least that much alignment. All alignments must be
729a power of 2.</p>
730
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]731</div>
732
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]733
734<!-- ======================================================================= -->
735<div class="doc_subsection">
736 <a name="aliasstructure">Aliases</a>
737</div>
738<div class="doc_text">
739 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikova80e1182007-04-28 13:45:00 +0000[diff] [blame]740 function or global variable or bitcast of global value). Aliases may have an
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]741 optional <a href="#linkage">linkage type</a>, and an
742 optional <a href="#visibility">visibility style</a>.</p>
743
744 <h5>Syntax:</h5>
745
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]746<div class="doc_code">
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]747<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]748@&lt;Name&gt; = [Linkage] [Visibility] alias &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]749</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]750</div>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]751
752</div>
753
754
755
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]756<!-- ======================================================================= -->
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]757<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
758<div class="doc_text">
759 <p>The return type and each parameter of a function type may have a set of
760 <i>parameter attributes</i> associated with them. Parameter attributes are
761 used to communicate additional information about the result or parameters of
762 a function. Parameter attributes are considered to be part of the function
763 type so two functions types that differ only by the parameter attributes
764 are different function types.</p>
765
Reid Spencer950e9f82007-01-15 18:27:39 +0000[diff] [blame]766 <p>Parameter attributes are simple keywords that follow the type specified. If
767 multiple parameter attributes are needed, they are space separated. For
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]768 example:</p>
769
770<div class="doc_code">
771<pre>
Reid Spencer9445e9a2007-07-19 23:13:04 +0000[diff] [blame]772%someFunc = i16 (i8 signext %someParam) zeroext
773%someFunc = i16 (i8 zeroext %someParam) zeroext
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]774</pre>
775</div>
776
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]777 <p>Note that the two function types above are unique because the parameter has
Reid Spencer9445e9a2007-07-19 23:13:04 +0000[diff] [blame]778 a different attribute (<tt>signext</tt> in the first one, <tt>zeroext</tt> in
779 the second). Also note that the attribute for the function result
780 (<tt>zeroext</tt>) comes immediately after the argument list.</p>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]781
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]782 <p>Currently, only the following parameter attributes are defined:</p>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]783 <dl>
Reid Spencer9445e9a2007-07-19 23:13:04 +0000[diff] [blame]784 <dt><tt>zeroext</tt></dt>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]785 <dd>This indicates that the parameter should be zero extended just before
786 a call to this function.</dd>
Reid Spencer9445e9a2007-07-19 23:13:04 +0000[diff] [blame]787 <dt><tt>signext</tt></dt>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]788 <dd>This indicates that the parameter should be sign extended just before
789 a call to this function.</dd>
Anton Korobeynikov34d6dea2007-01-28 14:30:45 +0000[diff] [blame]790 <dt><tt>inreg</tt></dt>
791 <dd>This indicates that the parameter should be placed in register (if
Anton Korobeynikov66a8c8c2007-01-28 15:27:21 +0000[diff] [blame]792 possible) during assembling function call. Support for this attribute is
793 target-specific</dd>
Anton Korobeynikov34d6dea2007-01-28 14:30:45 +0000[diff] [blame]794 <dt><tt>sret</tt></dt>
Anton Korobeynikov66a8c8c2007-01-28 15:27:21 +0000[diff] [blame]795 <dd>This indicates that the parameter specifies the address of a structure
Reid Spencer67606122007-03-22 02:02:11 +0000[diff] [blame]796 that is the return value of the function in the source program.</dd>
Zhou Shengfebca342007-06-05 05:28:26 +0000[diff] [blame]797 <dt><tt>noalias</tt></dt>
798 <dd>This indicates that the parameter not alias any other object or any
799 other "noalias" objects during the function call.
Reid Spencer2dc52012007-03-22 02:18:56 +0000[diff] [blame]800 <dt><tt>noreturn</tt></dt>
801 <dd>This function attribute indicates that the function never returns. This
802 indicates to LLVM that every call to this function should be treated as if
803 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
Reid Spencer67606122007-03-22 02:02:11 +0000[diff] [blame]804 <dt><tt>nounwind</tt></dt>
805 <dd>This function attribute indicates that the function type does not use
806 the unwind instruction and does not allow stack unwinding to propagate
807 through it.</dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]808 </dl>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]809
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]810</div>
811
812<!-- ======================================================================= -->
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]813<div class="doc_subsection">
Chris Lattner1eeeb0c2006-04-08 04:40:53 +0000[diff] [blame]814 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]815</div>
816
817<div class="doc_text">
818<p>
819Modules may contain "module-level inline asm" blocks, which corresponds to the
820GCC "file scope inline asm" blocks. These blocks are internally concatenated by
821LLVM and treated as a single unit, but may be separated in the .ll file if
822desired. The syntax is very simple:
823</p>
824
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]825<div class="doc_code">
826<pre>
827module asm "inline asm code goes here"
828module asm "more can go here"
829</pre>
830</div>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]831
832<p>The strings can contain any character by escaping non-printable characters.
833 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
834 for the number.
835</p>
836
837<p>
838 The inline asm code is simply printed to the machine code .s file when
839 assembly code is generated.
840</p>
841</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]842
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]843<!-- ======================================================================= -->
844<div class="doc_subsection">
845 <a name="datalayout">Data Layout</a>
846</div>
847
848<div class="doc_text">
849<p>A module may specify a target specific data layout string that specifies how
Reid Spencerc8910842007-04-11 23:49:50 +0000[diff] [blame]850data is to be laid out in memory. The syntax for the data layout is simply:</p>
851<pre> target datalayout = "<i>layout specification</i>"</pre>
852<p>The <i>layout specification</i> consists of a list of specifications
853separated by the minus sign character ('-'). Each specification starts with a
854letter and may include other information after the letter to define some
855aspect of the data layout. The specifications accepted are as follows: </p>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]856<dl>
857 <dt><tt>E</tt></dt>
858 <dd>Specifies that the target lays out data in big-endian form. That is, the
859 bits with the most significance have the lowest address location.</dd>
860 <dt><tt>e</tt></dt>
861 <dd>Specifies that hte target lays out data in little-endian form. That is,
862 the bits with the least significance have the lowest address location.</dd>
863 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
864 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
865 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
866 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
867 too.</dd>
868 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
869 <dd>This specifies the alignment for an integer type of a given bit
870 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
871 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
872 <dd>This specifies the alignment for a vector type of a given bit
873 <i>size</i>.</dd>
874 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
875 <dd>This specifies the alignment for a floating point type of a given bit
876 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
877 (double).</dd>
878 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
879 <dd>This specifies the alignment for an aggregate type of a given bit
880 <i>size</i>.</dd>
881</dl>
882<p>When constructing the data layout for a given target, LLVM starts with a
883default set of specifications which are then (possibly) overriden by the
884specifications in the <tt>datalayout</tt> keyword. The default specifications
885are given in this list:</p>
886<ul>
887 <li><tt>E</tt> - big endian</li>
888 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
889 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
890 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
891 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
892 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
893 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
894 alignment of 64-bits</li>
895 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
896 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
897 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
898 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
899 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
900</ul>
901<p>When llvm is determining the alignment for a given type, it uses the
902following rules:
903<ol>
904 <li>If the type sought is an exact match for one of the specifications, that
905 specification is used.</li>
906 <li>If no match is found, and the type sought is an integer type, then the
907 smallest integer type that is larger than the bitwidth of the sought type is
908 used. If none of the specifications are larger than the bitwidth then the the
909 largest integer type is used. For example, given the default specifications
910 above, the i7 type will use the alignment of i8 (next largest) while both
911 i65 and i256 will use the alignment of i64 (largest specified).</li>
912 <li>If no match is found, and the type sought is a vector type, then the
913 largest vector type that is smaller than the sought vector type will be used
914 as a fall back. This happens because <128 x double> can be implemented in
915 terms of 64 <2 x double>, for example.</li>
916</ol>
917</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]918
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]919<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]920<div class="doc_section"> <a name="typesystem">Type System</a> </div>
921<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]922
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]923<div class="doc_text">
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]924
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]925<p>The LLVM type system is one of the most important features of the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]926intermediate representation. Being typed enables a number of
927optimizations to be performed on the IR directly, without having to do
928extra analyses on the side before the transformation. A strong type
929system makes it easier to read the generated code and enables novel
930analyses and transformations that are not feasible to perform on normal
931three address code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]932
933</div>
934
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]935<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]936<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]937<div class="doc_text">
John Criswell4457dc92004-04-09 16:48:45 +0000[diff] [blame]938<p>The primitive types are the fundamental building blocks of the LLVM
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]939system. The current set of primitive types is as follows:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]940
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]941<table class="layout">
942 <tr class="layout">
943 <td class="left">
944 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]945 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]946 <tr><th>Type</th><th>Description</th></tr>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]947 <tr><td><tt><a name="t_void">void</a></tt></td><td>No value</td></tr>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]948 <tr><td><tt>label</tt></td><td>Branch destination</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]949 </tbody>
950 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]951 </td>
952 <td class="right">
953 <table>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]954 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]955 <tr><th>Type</th><th>Description</th></tr>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]956 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]957 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]958 </tbody>
959 </table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]960 </td>
961 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]962</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]963</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]964
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]965<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]966<div class="doc_subsubsection"> <a name="t_classifications">Type
967Classifications</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]968<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]969<p>These different primitive types fall into a few useful
970classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]971
972<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]973 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]974 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]975 <tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]976 <td><a name="t_integer">integer</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]977 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]978 </tr>
979 <tr>
980 <td><a name="t_floating">floating point</a></td>
981 <td><tt>float, double</tt></td>
982 </tr>
983 <tr>
984 <td><a name="t_firstclass">first class</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]985 <td><tt>i1, ..., float, double, <br/>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]986 <a href="#t_pointer">pointer</a>,<a href="#t_vector">vector</a></tt>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]987 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]988 </tr>
989 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]990</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]991
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]992<p>The <a href="#t_firstclass">first class</a> types are perhaps the
993most important. Values of these types are the only ones which can be
994produced by instructions, passed as arguments, or used as operands to
995instructions. This means that all structures and arrays must be
996manipulated either by pointer or by component.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]997</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]998
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]999<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1000<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1001
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1002<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1003
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1004<p>The real power in LLVM comes from the derived types in the system.
1005This is what allows a programmer to represent arrays, functions,
1006pointers, and other useful types. Note that these derived types may be
1007recursive: For example, it is possible to have a two dimensional array.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1008
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1009</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1010
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1011<!-- _______________________________________________________________________ -->
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1012<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1013
1014<div class="doc_text">
1015
1016<h5>Overview:</h5>
1017<p>The integer type is a very simple derived type that simply specifies an
1018arbitrary bit width for the integer type desired. Any bit width from 1 bit to
10192^23-1 (about 8 million) can be specified.</p>
1020
1021<h5>Syntax:</h5>
1022
1023<pre>
1024 iN
1025</pre>
1026
1027<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1028value.</p>
1029
1030<h5>Examples:</h5>
1031<table class="layout">
1032 <tr class="layout">
1033 <td class="left">
1034 <tt>i1</tt><br/>
1035 <tt>i4</tt><br/>
1036 <tt>i8</tt><br/>
1037 <tt>i16</tt><br/>
1038 <tt>i32</tt><br/>
1039 <tt>i42</tt><br/>
1040 <tt>i64</tt><br/>
1041 <tt>i1942652</tt><br/>
1042 </td>
1043 <td class="left">
1044 A boolean integer of 1 bit<br/>
1045 A nibble sized integer of 4 bits.<br/>
1046 A byte sized integer of 8 bits.<br/>
1047 A half word sized integer of 16 bits.<br/>
1048 A word sized integer of 32 bits.<br/>
1049 An integer whose bit width is the answer. <br/>
1050 A double word sized integer of 64 bits.<br/>
1051 A really big integer of over 1 million bits.<br/>
1052 </td>
1053 </tr>
1054</table>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1055</div>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1056
1057<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1058<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1059
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1060<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1061
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1062<h5>Overview:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1063
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1064<p>The array type is a very simple derived type that arranges elements
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1065sequentially in memory. The array type requires a size (number of
1066elements) and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1067
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1068<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1069
1070<pre>
1071 [&lt;# elements&gt; x &lt;elementtype&gt;]
1072</pre>
1073
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1074<p>The number of elements is a constant integer value; elementtype may
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1075be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1076
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1077<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1078<table class="layout">
1079 <tr class="layout">
1080 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1081 <tt>[40 x i32 ]</tt><br/>
1082 <tt>[41 x i32 ]</tt><br/>
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1083 <tt>[40 x i8]</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1084 </td>
1085 <td class="left">
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1086 Array of 40 32-bit integer values.<br/>
1087 Array of 41 32-bit integer values.<br/>
1088 Array of 40 8-bit integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1089 </td>
1090 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1091</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1092<p>Here are some examples of multidimensional arrays:</p>
1093<table class="layout">
1094 <tr class="layout">
1095 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1096 <tt>[3 x [4 x i32]]</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1097 <tt>[12 x [10 x float]]</tt><br/>
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1098 <tt>[2 x [3 x [4 x i16]]]</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1099 </td>
1100 <td class="left">
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1101 3x4 array of 32-bit integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1102 12x10 array of single precision floating point values.<br/>
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1103 2x3x4 array of 16-bit integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1104 </td>
1105 </tr>
1106</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1107
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1108<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1109length array. Normally, accesses past the end of an array are undefined in
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1110LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1111As a special case, however, zero length arrays are recognized to be variable
1112length. This allows implementation of 'pascal style arrays' with the LLVM
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1113type "{ i32, [0 x float]}", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1114
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1115</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1116
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1117<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1118<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1119<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1120<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1121<p>The function type can be thought of as a function signature. It
1122consists of a return type and a list of formal parameter types.
John Criswell009900b2003-11-25 21:45:46 +0000[diff] [blame]1123Function types are usually used to build virtual function tables
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1124(which are structures of pointers to functions), for indirect function
1125calls, and when defining a function.</p>
John Criswell009900b2003-11-25 21:45:46 +0000[diff] [blame]1126<p>
1127The return type of a function type cannot be an aggregate type.
1128</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1129<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1130<pre> &lt;returntype&gt; (&lt;parameter list&gt;)<br></pre>
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1131<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]1132specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1133which indicates that the function takes a variable number of arguments.
1134Variable argument functions can access their arguments with the <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1135 href="#int_varargs">variable argument handling intrinsic</a> functions.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1136<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1137<table class="layout">
1138 <tr class="layout">
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1139 <td class="left"><tt>i32 (i32)</tt></td>
1140 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1141 </td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1142 </tr><tr class="layout">
Reid Spencer9445e9a2007-07-19 23:13:04 +0000[diff] [blame]1143 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencerf17a0b72006-12-31 07:20:23 +0000[diff] [blame]1144 </tt></td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1145 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1146 an <tt>i16</tt> that should be sign extended and a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1147 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1148 <tt>float</tt>.
1149 </td>
1150 </tr><tr class="layout">
1151 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1152 <td class="left">A vararg function that takes at least one
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1153 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1154 which returns an integer. This is the signature for <tt>printf</tt> in
1155 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1156 </td>
1157 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1158</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1159
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1160</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1161<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1162<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1163<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1164<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1165<p>The structure type is used to represent a collection of data members
1166together in memory. The packing of the field types is defined to match
1167the ABI of the underlying processor. The elements of a structure may
1168be any type that has a size.</p>
1169<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1170and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1171field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1172instruction.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1173<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1174<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1175<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1176<table class="layout">
1177 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]1178 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1179 <td class="left">A triple of three <tt>i32</tt> values</td>
1180 </tr><tr class="layout">
1181 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1182 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1183 second element is a <a href="#t_pointer">pointer</a> to a
1184 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1185 an <tt>i32</tt>.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1186 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1187</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1188</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1189
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1190<!-- _______________________________________________________________________ -->
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]1191<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1192</div>
1193<div class="doc_text">
1194<h5>Overview:</h5>
1195<p>The packed structure type is used to represent a collection of data members
1196together in memory. There is no padding between fields. Further, the alignment
1197of a packed structure is 1 byte. The elements of a packed structure may
1198be any type that has a size.</p>
1199<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1200and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1201field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1202instruction.</p>
1203<h5>Syntax:</h5>
1204<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1205<h5>Examples:</h5>
1206<table class="layout">
1207 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]1208 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1209 <td class="left">A triple of three <tt>i32</tt> values</td>
1210 </tr><tr class="layout">
1211 <td class="left"><tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}&nbsp;&gt;</tt></td>
1212 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1213 second element is a <a href="#t_pointer">pointer</a> to a
1214 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1215 an <tt>i32</tt>.</td>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]1216 </tr>
1217</table>
1218</div>
1219
1220<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1221<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1222<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1223<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1224<p>As in many languages, the pointer type represents a pointer or
1225reference to another object, which must live in memory.</p>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1226<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1227<pre> &lt;type&gt; *<br></pre>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1228<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1229<table class="layout">
1230 <tr class="layout">
1231 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1232 <tt>[4x i32]*</tt><br/>
1233 <tt>i32 (i32 *) *</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1234 </td>
1235 <td class="left">
1236 A <a href="#t_pointer">pointer</a> to <a href="#t_array">array</a> of
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1237 four <tt>i32</tt> values<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1238 A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1239 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
1240 <tt>i32</tt>.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1241 </td>
1242 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1243</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1244</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1245
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1246<!-- _______________________________________________________________________ -->
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1247<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1248<div class="doc_text">
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1249
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1250<h5>Overview:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1251
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1252<p>A vector type is a simple derived type that represents a vector
1253of elements. Vector types are used when multiple primitive data
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1254are operated in parallel using a single instruction (SIMD).
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1255A vector type requires a size (number of
Chris Lattnerb8d172f2005-11-10 01:44:22 +0000[diff] [blame]1256elements) and an underlying primitive data type. Vectors must have a power
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1257of two length (1, 2, 4, 8, 16 ...). Vector types are
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1258considered <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1259
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1260<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1261
1262<pre>
1263 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1264</pre>
1265
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1266<p>The number of elements is a constant integer value; elementtype may
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]1267be any integer or floating point type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1268
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1269<h5>Examples:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1270
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1271<table class="layout">
1272 <tr class="layout">
1273 <td class="left">
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1274 <tt>&lt;4 x i32&gt;</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1275 <tt>&lt;8 x float&gt;</tt><br/>
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1276 <tt>&lt;2 x i64&gt;</tt><br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1277 </td>
1278 <td class="left">
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1279 Vector of 4 32-bit integer values.<br/>
1280 Vector of 8 floating-point values.<br/>
1281 Vector of 2 64-bit integer values.<br/>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1282 </td>
1283 </tr>
1284</table>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1285</div>
1286
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]1287<!-- _______________________________________________________________________ -->
1288<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1289<div class="doc_text">
1290
1291<h5>Overview:</h5>
1292
1293<p>Opaque types are used to represent unknown types in the system. This
1294corresponds (for example) to the C notion of a foward declared structure type.
1295In LLVM, opaque types can eventually be resolved to any type (not just a
1296structure type).</p>
1297
1298<h5>Syntax:</h5>
1299
1300<pre>
1301 opaque
1302</pre>
1303
1304<h5>Examples:</h5>
1305
1306<table class="layout">
1307 <tr class="layout">
1308 <td class="left">
1309 <tt>opaque</tt>
1310 </td>
1311 <td class="left">
1312 An opaque type.<br/>
1313 </td>
1314 </tr>
1315</table>
1316</div>
1317
1318
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1319<!-- *********************************************************************** -->
1320<div class="doc_section"> <a name="constants">Constants</a> </div>
1321<!-- *********************************************************************** -->
1322
1323<div class="doc_text">
1324
1325<p>LLVM has several different basic types of constants. This section describes
1326them all and their syntax.</p>
1327
1328</div>
1329
1330<!-- ======================================================================= -->
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1331<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1332
1333<div class="doc_text">
1334
1335<dl>
1336 <dt><b>Boolean constants</b></dt>
1337
1338 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]1339 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1340 </dd>
1341
1342 <dt><b>Integer constants</b></dt>
1343
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1344 <dd>Standard integers (such as '4') are constants of the <a
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]1345 href="#t_integer">integer</a> type. Negative numbers may be used with
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1346 integer types.
1347 </dd>
1348
1349 <dt><b>Floating point constants</b></dt>
1350
1351 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1352 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1353 notation (see below). Floating point constants must have a <a
1354 href="#t_floating">floating point</a> type. </dd>
1355
1356 <dt><b>Null pointer constants</b></dt>
1357
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1358 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1359 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1360
1361</dl>
1362
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1363<p>The one non-intuitive notation for constants is the optional hexadecimal form
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1364of floating point constants. For example, the form '<tt>double
13650x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
13664.5e+15</tt>'. The only time hexadecimal floating point constants are required
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]1367(and the only time that they are generated by the disassembler) is when a
1368floating point constant must be emitted but it cannot be represented as a
1369decimal floating point number. For example, NaN's, infinities, and other
1370special values are represented in their IEEE hexadecimal format so that
1371assembly and disassembly do not cause any bits to change in the constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1372
1373</div>
1374
1375<!-- ======================================================================= -->
1376<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1377</div>
1378
1379<div class="doc_text">
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]1380<p>Aggregate constants arise from aggregation of simple constants
1381and smaller aggregate constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1382
1383<dl>
1384 <dt><b>Structure constants</b></dt>
1385
1386 <dd>Structure constants are represented with notation similar to structure
1387 type definitions (a comma separated list of elements, surrounded by braces
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1388 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* %G }</tt>",
Chris Lattner3e63a9d2007-07-13 20:01:46 +0000[diff] [blame]1389 where "<tt>%G</tt>" is declared as "<tt>@G = external global i32</tt>". Structure constants
Chris Lattnerd4f6b172005-03-07 22:13:59 +0000[diff] [blame]1390 must have <a href="#t_struct">structure type</a>, and the number and
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1391 types of elements must match those specified by the type.
1392 </dd>
1393
1394 <dt><b>Array constants</b></dt>
1395
1396 <dd>Array constants are represented with notation similar to array type
1397 definitions (a comma separated list of elements, surrounded by square brackets
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1398 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1399 constants must have <a href="#t_array">array type</a>, and the number and
1400 types of elements must match those specified by the type.
1401 </dd>
1402
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1403 <dt><b>Vector constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1404
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1405 <dd>Vector constants are represented with notation similar to vector type
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1406 definitions (a comma separated list of elements, surrounded by
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1407 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]1408 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1409 href="#t_vector">vector type</a>, and the number and types of elements must
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1410 match those specified by the type.
1411 </dd>
1412
1413 <dt><b>Zero initialization</b></dt>
1414
1415 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1416 value to zero of <em>any</em> type, including scalar and aggregate types.
1417 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]1418 large arrays) and is always exactly equivalent to using explicit zero
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1419 initializers.
1420 </dd>
1421</dl>
1422
1423</div>
1424
1425<!-- ======================================================================= -->
1426<div class="doc_subsection">
1427 <a name="globalconstants">Global Variable and Function Addresses</a>
1428</div>
1429
1430<div class="doc_text">
1431
1432<p>The addresses of <a href="#globalvars">global variables</a> and <a
1433href="#functionstructure">functions</a> are always implicitly valid (link-time)
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1434constants. These constants are explicitly referenced when the <a
1435href="#identifiers">identifier for the global</a> is used and always have <a
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1436href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1437file:</p>
1438
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1439<div class="doc_code">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1440<pre>
Chris Lattnera18a4242007-06-06 18:28:13 +0000[diff] [blame]1441@X = global i32 17
1442@Y = global i32 42
1443@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1444</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1445</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1446
1447</div>
1448
1449<!-- ======================================================================= -->
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1450<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1451<div class="doc_text">
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1452 <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]1453 no specific value. Undefined values may be of any type and be used anywhere
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1454 a constant is permitted.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1455
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1456 <p>Undefined values indicate to the compiler that the program is well defined
1457 no matter what value is used, giving the compiler more freedom to optimize.
1458 </p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1459</div>
1460
1461<!-- ======================================================================= -->
1462<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1463</div>
1464
1465<div class="doc_text">
1466
1467<p>Constant expressions are used to allow expressions involving other constants
1468to be used as constants. Constant expressions may be of any <a
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1469href="#t_firstclass">first class</a> type and may involve any LLVM operation
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1470that does not have side effects (e.g. load and call are not supported). The
1471following is the syntax for constant expressions:</p>
1472
1473<dl>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1474 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1475 <dd>Truncate a constant to another type. The bit size of CST must be larger
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]1476 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1477
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1478 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1479 <dd>Zero extend a constant to another type. The bit size of CST must be
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]1480 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1481
1482 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1483 <dd>Sign extend a constant to another type. The bit size of CST must be
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]1484 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1485
1486 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1487 <dd>Truncate a floating point constant to another floating point type. The
1488 size of CST must be larger than the size of TYPE. Both types must be
1489 floating point.</dd>
1490
1491 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1492 <dd>Floating point extend a constant to another type. The size of CST must be
1493 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1494
1495 <dt><b><tt>fp2uint ( CST to TYPE )</tt></b></dt>
1496 <dd>Convert a floating point constant to the corresponding unsigned integer
1497 constant. TYPE must be an integer type. CST must be floating point. If the
1498 value won't fit in the integer type, the results are undefined.</dd>
1499
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]1500 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1501 <dd>Convert a floating point constant to the corresponding signed integer
1502 constant. TYPE must be an integer type. CST must be floating point. If the
1503 value won't fit in the integer type, the results are undefined.</dd>
1504
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]1505 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1506 <dd>Convert an unsigned integer constant to the corresponding floating point
1507 constant. TYPE must be floating point. CST must be of integer type. If the
Jeff Cohencb757312007-04-22 14:56:37 +0000[diff] [blame]1508 value won't fit in the floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1509
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]1510 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1511 <dd>Convert a signed integer constant to the corresponding floating point
1512 constant. TYPE must be floating point. CST must be of integer type. If the
Jeff Cohencb757312007-04-22 14:56:37 +0000[diff] [blame]1513 value won't fit in the floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1514
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]1515 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1516 <dd>Convert a pointer typed constant to the corresponding integer constant
1517 TYPE must be an integer type. CST must be of pointer type. The CST value is
1518 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1519
1520 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1521 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1522 pointer type. CST must be of integer type. The CST value is zero extended,
1523 truncated, or unchanged to make it fit in a pointer size. This one is
1524 <i>really</i> dangerous!</dd>
1525
1526 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1527 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1528 identical (same number of bits). The conversion is done as if the CST value
1529 was stored to memory and read back as TYPE. In other words, no bits change
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]1530 with this operator, just the type. This can be used for conversion of
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1531 vector types to any other type, as long as they have the same bit width. For
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]1532 pointers it is only valid to cast to another pointer type.
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]1533 </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1534
1535 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1536
1537 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1538 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1539 instruction, the index list may have zero or more indexes, which are required
1540 to make sense for the type of "CSTPTR".</dd>
1541
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]1542 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1543
1544 <dd>Perform the <a href="#i_select">select operation</a> on
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]1545 constants.</dd>
1546
1547 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1548 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1549
1550 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1551 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]1552
1553 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1554
1555 <dd>Perform the <a href="#i_extractelement">extractelement
1556 operation</a> on constants.
1557
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]1558 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1559
1560 <dd>Perform the <a href="#i_insertelement">insertelement
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]1561 operation</a> on constants.</dd>
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]1562
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]1563
1564 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1565
1566 <dd>Perform the <a href="#i_shufflevector">shufflevector
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]1567 operation</a> on constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]1568
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1569 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1570
Reid Spencer2dc45b82004-12-09 18:13:12 +0000[diff] [blame]1571 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1572 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1573 binary</a> operations. The constraints on operands are the same as those for
1574 the corresponding instruction (e.g. no bitwise operations on floating point
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1575 values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1576</dl>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1577</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000[diff] [blame]1578
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1579<!-- *********************************************************************** -->
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1580<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1581<!-- *********************************************************************** -->
1582
1583<!-- ======================================================================= -->
1584<div class="doc_subsection">
1585<a name="inlineasm">Inline Assembler Expressions</a>
1586</div>
1587
1588<div class="doc_text">
1589
1590<p>
1591LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1592Module-Level Inline Assembly</a>) through the use of a special value. This
1593value represents the inline assembler as a string (containing the instructions
1594to emit), a list of operand constraints (stored as a string), and a flag that
1595indicates whether or not the inline asm expression has side effects. An example
1596inline assembler expression is:
1597</p>
1598
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1599<div class="doc_code">
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1600<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1601i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1602</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1603</div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1604
1605<p>
1606Inline assembler expressions may <b>only</b> be used as the callee operand of
1607a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1608</p>
1609
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1610<div class="doc_code">
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1611<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1612%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1613</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1614</div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1615
1616<p>
1617Inline asms with side effects not visible in the constraint list must be marked
1618as having side effects. This is done through the use of the
1619'<tt>sideeffect</tt>' keyword, like so:
1620</p>
1621
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1622<div class="doc_code">
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1623<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1624call void asm sideeffect "eieio", ""()
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1625</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1626</div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]1627
1628<p>TODO: The format of the asm and constraints string still need to be
1629documented here. Constraints on what can be done (e.g. duplication, moving, etc
1630need to be documented).
1631</p>
1632
1633</div>
1634
1635<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1636<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1637<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1638
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1639<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1640
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1641<p>The LLVM instruction set consists of several different
1642classifications of instructions: <a href="#terminators">terminator
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1643instructions</a>, <a href="#binaryops">binary instructions</a>,
1644<a href="#bitwiseops">bitwise binary instructions</a>, <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1645 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1646instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1647
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1648</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1649
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1650<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1651<div class="doc_subsection"> <a name="terminators">Terminator
1652Instructions</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1653
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1654<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1655
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1656<p>As mentioned <a href="#functionstructure">previously</a>, every
1657basic block in a program ends with a "Terminator" instruction, which
1658indicates which block should be executed after the current block is
1659finished. These terminator instructions typically yield a '<tt>void</tt>'
1660value: they produce control flow, not values (the one exception being
1661the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1662<p>There are six different terminator instructions: the '<a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1663 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1664instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1665the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1666 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1667 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1668
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1669</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1670
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1671<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1672<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1673Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1674<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1675<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1676<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]1677 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1678</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1679<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1680<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]1681value) from a function back to the caller.</p>
John Criswell4457dc92004-04-09 16:48:45 +0000[diff] [blame]1682<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1683returns a value and then causes control flow, and one that just causes
1684control flow to occur.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1685<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1686<p>The '<tt>ret</tt>' instruction may return any '<a
1687 href="#t_firstclass">first class</a>' type. Notice that a function is
1688not <a href="#wellformed">well formed</a> if there exists a '<tt>ret</tt>'
1689instruction inside of the function that returns a value that does not
1690match the return type of the function.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1691<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1692<p>When the '<tt>ret</tt>' instruction is executed, control flow
1693returns back to the calling function's context. If the caller is a "<a
John Criswellfa081872004-06-25 15:16:57 +0000[diff] [blame]1694 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1695the instruction after the call. If the caller was an "<a
1696 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1697at the beginning of the "normal" destination block. If the instruction
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1698returns a value, that value shall set the call or invoke instruction's
1699return value.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1700<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1701<pre> ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1702 ret void <i>; Return from a void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1703</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1704</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1705<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1706<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1707<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1708<h5>Syntax:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]1709<pre> br i1 &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]1710</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1711<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1712<p>The '<tt>br</tt>' instruction is used to cause control flow to
1713transfer to a different basic block in the current function. There are
1714two forms of this instruction, corresponding to a conditional branch
1715and an unconditional branch.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1716<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1717<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]1718single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1719unconditional form of the '<tt>br</tt>' instruction takes a single
1720'<tt>label</tt>' value as a target.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1721<h5>Semantics:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]1722<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1723argument is evaluated. If the value is <tt>true</tt>, control flows
1724to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1725control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1726<h5>Example:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]1727<pre>Test:<br> %cond = <a href="#i_icmp">icmp</a> eq, i32 %a, %b<br> br i1 %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1728 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1729</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1730<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1731<div class="doc_subsubsection">
1732 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1733</div>
1734
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1735<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1736<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1737
1738<pre>
1739 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1740</pre>
1741
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1742<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1743
1744<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1745several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1746instruction, allowing a branch to occur to one of many possible
1747destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1748
1749
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1750<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1751
1752<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1753comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1754an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1755table is not allowed to contain duplicate constant entries.</p>
1756
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1757<h5>Semantics:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1758
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1759<p>The <tt>switch</tt> instruction specifies a table of values and
1760destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswell84114752004-06-25 16:05:06 +0000[diff] [blame]1761table is searched for the given value. If the value is found, control flow is
1762transfered to the corresponding destination; otherwise, control flow is
1763transfered to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1764
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1765<h5>Implementation:</h5>
1766
1767<p>Depending on properties of the target machine and the particular
1768<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswell84114752004-06-25 16:05:06 +0000[diff] [blame]1769ways. For example, it could be generated as a series of chained conditional
1770branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1771
1772<h5>Example:</h5>
1773
1774<pre>
1775 <i>; Emulate a conditional br instruction</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]1776 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1777 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1778
1779 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1780 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]1781
1782 <i>; Implement a jump table:</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1783 switch i32 %val, label %otherwise [ i32 0, label %onzero
1784 i32 1, label %onone
1785 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1786</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1787</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1788
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1789<!-- _______________________________________________________________________ -->
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1790<div class="doc_subsubsection">
1791 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1792</div>
1793
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1794<div class="doc_text">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1795
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1796<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1797
1798<pre>
1799 &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]1800 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1801</pre>
1802
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]1803<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1804
1805<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1806function, with the possibility of control flow transfer to either the
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1807'<tt>normal</tt>' label or the
1808'<tt>exception</tt>' label. If the callee function returns with the
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1809"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1810"normal" label. If the callee (or any indirect callees) returns with the "<a
John Criswelle4c57cc2005-05-12 16:52:32 +0000[diff] [blame]1811href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
1812continued at the dynamically nearest "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1813
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1814<h5>Arguments:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1815
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1816<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1817
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1818<ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1819 <li>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]1820 The optional "cconv" marker indicates which <a href="#callingconv">calling
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1821 convention</a> the call should use. If none is specified, the call defaults
1822 to using C calling conventions.
1823 </li>
1824 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1825 function value being invoked. In most cases, this is a direct function
1826 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1827 an arbitrary pointer to function value.
1828 </li>
1829
1830 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1831 function to be invoked. </li>
1832
1833 <li>'<tt>function args</tt>': argument list whose types match the function
1834 signature argument types. If the function signature indicates the function
1835 accepts a variable number of arguments, the extra arguments can be
1836 specified. </li>
1837
1838 <li>'<tt>normal label</tt>': the label reached when the called function
1839 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
1840
1841 <li>'<tt>exception label</tt>': the label reached when a callee returns with
1842 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
1843
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1844</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1845
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1846<h5>Semantics:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1847
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1848<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1849href="#i_call">call</a></tt>' instruction in most regards. The primary
1850difference is that it establishes an association with a label, which is used by
1851the runtime library to unwind the stack.</p>
1852
1853<p>This instruction is used in languages with destructors to ensure that proper
1854cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
1855exception. Additionally, this is important for implementation of
1856'<tt>catch</tt>' clauses in high-level languages that support them.</p>
1857
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1858<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]1859<pre>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]1860 %retval = invoke i32 %Test(i32 15) to label %Continue
1861 unwind label %TestCleanup <i>; {i32}:retval set</i>
1862 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Test(i32 15) to label %Continue
1863 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1864</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1865</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1866
1867
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1868<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1869
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1870<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
1871Instruction</a> </div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1872
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1873<div class="doc_text">
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1874
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1875<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1876<pre>
1877 unwind
1878</pre>
1879
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1880<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1881
1882<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
1883at the first callee in the dynamic call stack which used an <a
1884href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
1885primarily used to implement exception handling.</p>
1886
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]1887<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1888
1889<p>The '<tt>unwind</tt>' intrinsic causes execution of the current function to
1890immediately halt. The dynamic call stack is then searched for the first <a
1891href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
1892execution continues at the "exceptional" destination block specified by the
1893<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
1894dynamic call chain, undefined behavior results.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1895</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]1896
1897<!-- _______________________________________________________________________ -->
1898
1899<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
1900Instruction</a> </div>
1901
1902<div class="doc_text">
1903
1904<h5>Syntax:</h5>
1905<pre>
1906 unreachable
1907</pre>
1908
1909<h5>Overview:</h5>
1910
1911<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
1912instruction is used to inform the optimizer that a particular portion of the
1913code is not reachable. This can be used to indicate that the code after a
1914no-return function cannot be reached, and other facts.</p>
1915
1916<h5>Semantics:</h5>
1917
1918<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
1919</div>
1920
1921
1922
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1923<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1924<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1925<div class="doc_text">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1926<p>Binary operators are used to do most of the computation in a
1927program. They require two operands, execute an operation on them, and
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]1928produce a single value. The operands might represent
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1929multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1930The result value of a binary operator is not
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1931necessarily the same type as its operands.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1932<p>There are several different binary operators:</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1933</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1934<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1935<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
1936Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1937<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1938<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1939<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]1940</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1941<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1942<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1943<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1944<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1945 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1946 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1947Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1948<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1949<p>The value produced is the integer or floating point sum of the two
1950operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1951<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1952<pre> &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1953</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1954</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1955<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1956<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
1957Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1958<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1959<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1960<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]1961</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1962<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1963<p>The '<tt>sub</tt>' instruction returns the difference of its two
1964operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1965<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
1966instruction present in most other intermediate representations.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1967<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1968<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1969 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1970values.
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1971This instruction can also take <a href="#t_vector">vector</a> versions of the values.
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1972Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1973<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1974<p>The value produced is the integer or floating point difference of
1975the two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1976<h5>Example:</h5>
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]1977<pre>
1978 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1979 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1980</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1981</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1982<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1983<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
1984Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1985<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1986<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1987<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]1988</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1989<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1990<p>The '<tt>mul</tt>' instruction returns the product of its two
1991operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1992<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1993<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1994 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1995values.
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]1996This instruction can also take <a href="#t_vector">vector</a> versions of the values.
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]1997Both arguments must have identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1998<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1999<p>The value produced is the integer or floating point product of the
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2000two operands.</p>
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]2001<p>Because the operands are the same width, the result of an integer
2002multiplication is the same whether the operands should be deemed unsigned or
2003signed.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2004<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2005<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2006</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2007</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2008<!-- _______________________________________________________________________ -->
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2009<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2010</a></div>
2011<div class="doc_text">
2012<h5>Syntax:</h5>
2013<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2014</pre>
2015<h5>Overview:</h5>
2016<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2017operands.</p>
2018<h5>Arguments:</h5>
2019<p>The two arguments to the '<tt>udiv</tt>' instruction must be
2020<a href="#t_integer">integer</a> values. Both arguments must have identical
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2021types. This instruction can also take <a href="#t_vector">vector</a> versions
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2022of the values in which case the elements must be integers.</p>
2023<h5>Semantics:</h5>
2024<p>The value produced is the unsigned integer quotient of the two operands. This
2025instruction always performs an unsigned division operation, regardless of
2026whether the arguments are unsigned or not.</p>
2027<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2028<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2029</pre>
2030</div>
2031<!-- _______________________________________________________________________ -->
2032<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2033</a> </div>
2034<div class="doc_text">
2035<h5>Syntax:</h5>
2036<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2037</pre>
2038<h5>Overview:</h5>
2039<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2040operands.</p>
2041<h5>Arguments:</h5>
2042<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2043<a href="#t_integer">integer</a> values. Both arguments must have identical
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2044types. This instruction can also take <a href="#t_vector">vector</a> versions
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2045of the values in which case the elements must be integers.</p>
2046<h5>Semantics:</h5>
2047<p>The value produced is the signed integer quotient of the two operands. This
2048instruction always performs a signed division operation, regardless of whether
2049the arguments are signed or not.</p>
2050<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2051<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2052</pre>
2053</div>
2054<!-- _______________________________________________________________________ -->
2055<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2056Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2057<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2058<h5>Syntax:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2059<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]2060</pre>
2061<h5>Overview:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2062<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2063operands.</p>
2064<h5>Arguments:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2065<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2066<a href="#t_floating">floating point</a> values. Both arguments must have
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2067identical types. This instruction can also take <a href="#t_vector">vector</a>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2068versions of floating point values.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2069<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2070<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2071<h5>Example:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]2072<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]2073</pre>
2074</div>
2075<!-- _______________________________________________________________________ -->
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2076<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2077</div>
2078<div class="doc_text">
2079<h5>Syntax:</h5>
2080<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2081</pre>
2082<h5>Overview:</h5>
2083<p>The '<tt>urem</tt>' instruction returns the remainder from the
2084unsigned division of its two arguments.</p>
2085<h5>Arguments:</h5>
2086<p>The two arguments to the '<tt>urem</tt>' instruction must be
2087<a href="#t_integer">integer</a> values. Both arguments must have identical
2088types.</p>
2089<h5>Semantics:</h5>
2090<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
2091This instruction always performs an unsigned division to get the remainder,
2092regardless of whether the arguments are unsigned or not.</p>
2093<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2094<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2095</pre>
2096
2097</div>
2098<!-- _______________________________________________________________________ -->
2099<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2100Instruction</a> </div>
2101<div class="doc_text">
2102<h5>Syntax:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2103<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]2104</pre>
2105<h5>Overview:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2106<p>The '<tt>srem</tt>' instruction returns the remainder from the
2107signed division of its two operands.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2108<h5>Arguments:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2109<p>The two arguments to the '<tt>srem</tt>' instruction must be
2110<a href="#t_integer">integer</a> values. Both arguments must have identical
2111types.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2112<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2113<p>This instruction returns the <i>remainder</i> of a division (where the result
Reid Spencerc9fdfc82007-03-24 22:23:39 +0000[diff] [blame]2114has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2115operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2116a value. For more information about the difference, see <a
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2117 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
Reid Spencerc9fdfc82007-03-24 22:23:39 +0000[diff] [blame]2118Math Forum</a>. For a table of how this is implemented in various languages,
Reid Spencer64f5c6c2007-03-24 22:40:44 +0000[diff] [blame]2119please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
Reid Spencerc9fdfc82007-03-24 22:23:39 +0000[diff] [blame]2120Wikipedia: modulo operation</a>.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2121<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2122<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]2123</pre>
2124
2125</div>
2126<!-- _______________________________________________________________________ -->
2127<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2128Instruction</a> </div>
2129<div class="doc_text">
2130<h5>Syntax:</h5>
2131<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2132</pre>
2133<h5>Overview:</h5>
2134<p>The '<tt>frem</tt>' instruction returns the remainder from the
2135division of its two operands.</p>
2136<h5>Arguments:</h5>
2137<p>The two arguments to the '<tt>frem</tt>' instruction must be
2138<a href="#t_floating">floating point</a> values. Both arguments must have
2139identical types.</p>
2140<h5>Semantics:</h5>
2141<p>This instruction returns the <i>remainder</i> of a division.</p>
2142<h5>Example:</h5>
2143<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]2144</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2145</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]2146
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]2147<!-- ======================================================================= -->
2148<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2149Operations</a> </div>
2150<div class="doc_text">
2151<p>Bitwise binary operators are used to do various forms of
2152bit-twiddling in a program. They are generally very efficient
2153instructions and can commonly be strength reduced from other
2154instructions. They require two operands, execute an operation on them,
2155and produce a single value. The resulting value of the bitwise binary
2156operators is always the same type as its first operand.</p>
2157</div>
2158
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]2159<!-- _______________________________________________________________________ -->
2160<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2161Instruction</a> </div>
2162<div class="doc_text">
2163<h5>Syntax:</h5>
2164<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2165</pre>
2166<h5>Overview:</h5>
2167<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2168the left a specified number of bits.</p>
2169<h5>Arguments:</h5>
2170<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
2171 href="#t_integer">integer</a> type.</p>
2172<h5>Semantics:</h5>
2173<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup>.</p>
2174<h5>Example:</h5><pre>
2175 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2176 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2177 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
2178</pre>
2179</div>
2180<!-- _______________________________________________________________________ -->
2181<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2182Instruction</a> </div>
2183<div class="doc_text">
2184<h5>Syntax:</h5>
2185<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2186</pre>
2187
2188<h5>Overview:</h5>
2189<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2190operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]2191
2192<h5>Arguments:</h5>
2193<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
2194<a href="#t_integer">integer</a> type.</p>
2195
2196<h5>Semantics:</h5>
2197<p>This instruction always performs a logical shift right operation. The most
2198significant bits of the result will be filled with zero bits after the
2199shift.</p>
2200
2201<h5>Example:</h5>
2202<pre>
2203 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2204 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2205 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2206 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
2207</pre>
2208</div>
2209
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]2210<!-- _______________________________________________________________________ -->
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]2211<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2212Instruction</a> </div>
2213<div class="doc_text">
2214
2215<h5>Syntax:</h5>
2216<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2217</pre>
2218
2219<h5>Overview:</h5>
2220<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2221operand shifted to the right a specified number of bits with sign extension.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]2222
2223<h5>Arguments:</h5>
2224<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
2225<a href="#t_integer">integer</a> type.</p>
2226
2227<h5>Semantics:</h5>
2228<p>This instruction always performs an arithmetic shift right operation,
2229The most significant bits of the result will be filled with the sign bit
2230of <tt>var1</tt>.</p>
2231
2232<h5>Example:</h5>
2233<pre>
2234 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2235 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2236 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2237 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
2238</pre>
2239</div>
2240
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2241<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2242<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2243Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2244<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2245<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2246<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]2247</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2248<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2249<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2250its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2251<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2252<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]2253 href="#t_integer">integer</a> values. Both arguments must have
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2254identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2255<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2256<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2257<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2258<div style="align: center">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2259<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2260 <tbody>
2261 <tr>
2262 <td>In0</td>
2263 <td>In1</td>
2264 <td>Out</td>
2265 </tr>
2266 <tr>
2267 <td>0</td>
2268 <td>0</td>
2269 <td>0</td>
2270 </tr>
2271 <tr>
2272 <td>0</td>
2273 <td>1</td>
2274 <td>0</td>
2275 </tr>
2276 <tr>
2277 <td>1</td>
2278 <td>0</td>
2279 <td>0</td>
2280 </tr>
2281 <tr>
2282 <td>1</td>
2283 <td>1</td>
2284 <td>1</td>
2285 </tr>
2286 </tbody>
2287</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2288</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2289<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2290<pre> &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
2291 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2292 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2293</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2294</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2295<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2296<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2297<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2298<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2299<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]2300</pre>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2301<h5>Overview:</h5>
2302<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2303or of its two operands.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2304<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2305<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]2306 href="#t_integer">integer</a> values. Both arguments must have
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2307identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2308<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2309<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2310<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2311<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2312<table border="1" cellspacing="0" cellpadding="4">
2313 <tbody>
2314 <tr>
2315 <td>In0</td>
2316 <td>In1</td>
2317 <td>Out</td>
2318 </tr>
2319 <tr>
2320 <td>0</td>
2321 <td>0</td>
2322 <td>0</td>
2323 </tr>
2324 <tr>
2325 <td>0</td>
2326 <td>1</td>
2327 <td>1</td>
2328 </tr>
2329 <tr>
2330 <td>1</td>
2331 <td>0</td>
2332 <td>1</td>
2333 </tr>
2334 <tr>
2335 <td>1</td>
2336 <td>1</td>
2337 <td>1</td>
2338 </tr>
2339 </tbody>
2340</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2341</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2342<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2343<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2344 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2345 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2346</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2347</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2348<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2349<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2350Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2351<div class="doc_text">
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2352<h5>Syntax:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2353<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]2354</pre>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2355<h5>Overview:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2356<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2357or of its two operands. The <tt>xor</tt> is used to implement the
2358"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2359<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2360<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]2361 href="#t_integer">integer</a> values. Both arguments must have
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2362identical types.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2363<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2364<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2365<p> </p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2366<div style="align: center">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2367<table border="1" cellspacing="0" cellpadding="4">
2368 <tbody>
2369 <tr>
2370 <td>In0</td>
2371 <td>In1</td>
2372 <td>Out</td>
2373 </tr>
2374 <tr>
2375 <td>0</td>
2376 <td>0</td>
2377 <td>0</td>
2378 </tr>
2379 <tr>
2380 <td>0</td>
2381 <td>1</td>
2382 <td>1</td>
2383 </tr>
2384 <tr>
2385 <td>1</td>
2386 <td>0</td>
2387 <td>1</td>
2388 </tr>
2389 <tr>
2390 <td>1</td>
2391 <td>1</td>
2392 <td>0</td>
2393 </tr>
2394 </tbody>
2395</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2396</div>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2397<p> </p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2398<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2399<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2400 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2401 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2402 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2403</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2404</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2405
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2406<!-- ======================================================================= -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2407<div class="doc_subsection">
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2408 <a name="vectorops">Vector Operations</a>
2409</div>
2410
2411<div class="doc_text">
2412
2413<p>LLVM supports several instructions to represent vector operations in a
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2414target-independent manner. These instructions cover the element-access and
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2415vector-specific operations needed to process vectors effectively. While LLVM
2416does directly support these vector operations, many sophisticated algorithms
2417will want to use target-specific intrinsics to take full advantage of a specific
2418target.</p>
2419
2420</div>
2421
2422<!-- _______________________________________________________________________ -->
2423<div class="doc_subsubsection">
2424 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2425</div>
2426
2427<div class="doc_text">
2428
2429<h5>Syntax:</h5>
2430
2431<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2432 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2433</pre>
2434
2435<h5>Overview:</h5>
2436
2437<p>
2438The '<tt>extractelement</tt>' instruction extracts a single scalar
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2439element from a vector at a specified index.
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2440</p>
2441
2442
2443<h5>Arguments:</h5>
2444
2445<p>
2446The first operand of an '<tt>extractelement</tt>' instruction is a
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2447value of <a href="#t_vector">vector</a> type. The second operand is
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2448an index indicating the position from which to extract the element.
2449The index may be a variable.</p>
2450
2451<h5>Semantics:</h5>
2452
2453<p>
2454The result is a scalar of the same type as the element type of
2455<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2456<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2457results are undefined.
2458</p>
2459
2460<h5>Example:</h5>
2461
2462<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2463 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2464</pre>
2465</div>
2466
2467
2468<!-- _______________________________________________________________________ -->
2469<div class="doc_subsubsection">
2470 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2471</div>
2472
2473<div class="doc_text">
2474
2475<h5>Syntax:</h5>
2476
2477<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2478 &lt;result&gt; = insertelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt, i32 &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2479</pre>
2480
2481<h5>Overview:</h5>
2482
2483<p>
2484The '<tt>insertelement</tt>' instruction inserts a scalar
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2485element into a vector at a specified index.
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2486</p>
2487
2488
2489<h5>Arguments:</h5>
2490
2491<p>
2492The first operand of an '<tt>insertelement</tt>' instruction is a
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2493value of <a href="#t_vector">vector</a> type. The second operand is a
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2494scalar value whose type must equal the element type of the first
2495operand. The third operand is an index indicating the position at
2496which to insert the value. The index may be a variable.</p>
2497
2498<h5>Semantics:</h5>
2499
2500<p>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2501The result is a vector of the same type as <tt>val</tt>. Its
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2502element values are those of <tt>val</tt> except at position
2503<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2504exceeds the length of <tt>val</tt>, the results are undefined.
2505</p>
2506
2507<h5>Example:</h5>
2508
2509<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2510 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2511</pre>
2512</div>
2513
2514<!-- _______________________________________________________________________ -->
2515<div class="doc_subsubsection">
2516 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2517</div>
2518
2519<div class="doc_text">
2520
2521<h5>Syntax:</h5>
2522
2523<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2524 &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 i32&gt; &lt;mask&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2525</pre>
2526
2527<h5>Overview:</h5>
2528
2529<p>
2530The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2531from two input vectors, returning a vector of the same type.
2532</p>
2533
2534<h5>Arguments:</h5>
2535
2536<p>
2537The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2538with types that match each other and types that match the result of the
2539instruction. The third argument is a shuffle mask, which has the same number
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2540of elements as the other vector type, but whose element type is always 'i32'.
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2541</p>
2542
2543<p>
2544The shuffle mask operand is required to be a constant vector with either
2545constant integer or undef values.
2546</p>
2547
2548<h5>Semantics:</h5>
2549
2550<p>
2551The elements of the two input vectors are numbered from left to right across
2552both of the vectors. The shuffle mask operand specifies, for each element of
2553the result vector, which element of the two input registers the result element
2554gets. The element selector may be undef (meaning "don't care") and the second
2555operand may be undef if performing a shuffle from only one vector.
2556</p>
2557
2558<h5>Example:</h5>
2559
2560<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2561 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2562 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2563 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2564 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2565</pre>
2566</div>
2567
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]2568
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]2569<!-- ======================================================================= -->
2570<div class="doc_subsection">
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]2571 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2572</div>
2573
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2574<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2575
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2576<p>A key design point of an SSA-based representation is how it
2577represents memory. In LLVM, no memory locations are in SSA form, which
2578makes things very simple. This section describes how to read, write,
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2579allocate, and free memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2580
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2581</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2582
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2583<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2584<div class="doc_subsubsection">
2585 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2586</div>
2587
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2588<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2589
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2590<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2591
2592<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2593 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2594</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2595
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2596<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2597
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2598<p>The '<tt>malloc</tt>' instruction allocates memory from the system
2599heap and returns a pointer to it.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2600
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2601<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2602
2603<p>The '<tt>malloc</tt>' instruction allocates
2604<tt>sizeof(&lt;type&gt;)*NumElements</tt>
John Criswell6e4ca612004-02-24 16:13:56 +0000[diff] [blame]2605bytes of memory from the operating system and returns a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2606appropriate type to the program. If "NumElements" is specified, it is the
2607number of elements allocated. If an alignment is specified, the value result
2608of the allocation is guaranteed to be aligned to at least that boundary. If
2609not specified, or if zero, the target can choose to align the allocation on any
2610convenient boundary.</p>
2611
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2612<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2613
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2614<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2615
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2616<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
2617a pointer is returned.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2618
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2619<h5>Example:</h5>
2620
2621<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2622 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2623
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]2624 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2625 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2626 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2627 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2628 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2629</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2630</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2631
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2632<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2633<div class="doc_subsubsection">
2634 <a name="i_free">'<tt>free</tt>' Instruction</a>
2635</div>
2636
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2637<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2638
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2639<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2640
2641<pre>
2642 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2643</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2644
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2645<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2646
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2647<p>The '<tt>free</tt>' instruction returns memory back to the unused
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2648memory heap to be reallocated in the future.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2649
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2650<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2651
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2652<p>'<tt>value</tt>' shall be a pointer value that points to a value
2653that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2654instruction.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2655
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2656<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2657
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2658<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2659after this instruction executes.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2660
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2661<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2662
2663<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2664 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2665 free [4 x i8]* %array
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2666</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2667</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2668
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2669<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2670<div class="doc_subsubsection">
2671 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2672</div>
2673
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2674<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2675
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2676<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2677
2678<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2679 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2680</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2681
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2682<h5>Overview:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2683
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2684<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
2685currently executing function, to be automatically released when this function
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2686returns to its caller.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2687
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2688<h5>Arguments:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2689
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2690<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2691bytes of memory on the runtime stack, returning a pointer of the
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2692appropriate type to the program. If "NumElements" is specified, it is the
2693number of elements allocated. If an alignment is specified, the value result
2694of the allocation is guaranteed to be aligned to at least that boundary. If
2695not specified, or if zero, the target can choose to align the allocation on any
2696convenient boundary.</p>
2697
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2698<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2699
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2700<h5>Semantics:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2701
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2702<p>Memory is allocated; a pointer is returned. '<tt>alloca</tt>'d
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2703memory is automatically released when the function returns. The '<tt>alloca</tt>'
2704instruction is commonly used to represent automatic variables that must
2705have an address available. When the function returns (either with the <tt><a
John Criswelldae2e932005-05-12 16:55:34 +0000[diff] [blame]2706 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]2707instructions), the memory is reclaimed.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2708
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2709<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2710
2711<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2712 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2713 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
2714 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2715 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2716</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2717</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]2718
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2719<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2720<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
2721Instruction</a> </div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2722<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2723<h5>Syntax:</h5>
Christopher Lamb2330e4d2007-04-21 08:16:25 +0000[diff] [blame]2724<pre> &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]<br> &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]<br></pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2725<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2726<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2727<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2728<p>The argument to the '<tt>load</tt>' instruction specifies the memory
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2729address from which to load. The pointer must point to a <a
Chris Lattnere53e5082004-06-03 22:57:15 +0000[diff] [blame]2730 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2731marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2732the number or order of execution of this <tt>load</tt> with other
2733volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
2734instructions. </p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2735<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2736<p>The location of memory pointed to is loaded.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2737<h5>Examples:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2738<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2739 <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2740 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2741 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2742</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2743</div>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2744<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2745<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
2746Instruction</a> </div>
Reid Spencer035ab572006-11-09 21:18:01 +0000[diff] [blame]2747<div class="doc_text">
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2748<h5>Syntax:</h5>
Christopher Lamb2330e4d2007-04-21 08:16:25 +0000[diff] [blame]2749<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
2750 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2751</pre>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2752<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2753<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2754<h5>Arguments:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2755<p>There are two arguments to the '<tt>store</tt>' instruction: a value
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2756to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2757operand must be a pointer to the type of the '<tt>&lt;value&gt;</tt>'
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2758operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2759optimizer is not allowed to modify the number or order of execution of
2760this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
2761 href="#i_store">store</a></tt> instructions.</p>
2762<h5>Semantics:</h5>
2763<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
2764at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2765<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2766<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2767 <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2768 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
2769 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2770</pre>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]2771</div>
2772
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]2773<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2774<div class="doc_subsubsection">
2775 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
2776</div>
2777
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2778<div class="doc_text">
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2779<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2780<pre>
2781 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
2782</pre>
2783
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2784<h5>Overview:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2785
2786<p>
2787The '<tt>getelementptr</tt>' instruction is used to get the address of a
2788subelement of an aggregate data structure.</p>
2789
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2790<h5>Arguments:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2791
Reid Spencer85f5b5b2006-12-04 21:29:24 +0000[diff] [blame]2792<p>This instruction takes a list of integer operands that indicate what
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2793elements of the aggregate object to index to. The actual types of the arguments
2794provided depend on the type of the first pointer argument. The
2795'<tt>getelementptr</tt>' instruction is used to index down through the type
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]2796levels of a structure or to a specific index in an array. When indexing into a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2797structure, only <tt>i32</tt> integer constants are allowed. When indexing
Reid Spencer85f5b5b2006-12-04 21:29:24 +0000[diff] [blame]2798into an array or pointer, only integers of 32 or 64 bits are allowed, and will
2799be sign extended to 64-bit values.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2800
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2801<p>For example, let's consider a C code fragment and how it gets
2802compiled to LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2803
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2804<div class="doc_code">
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2805<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2806struct RT {
2807 char A;
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]2808 int B[10][20];
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2809 char C;
2810};
2811struct ST {
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]2812 int X;
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2813 double Y;
2814 struct RT Z;
2815};
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2816
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]2817int *foo(struct ST *s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2818 return &amp;s[1].Z.B[5][13];
2819}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2820</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2821</div>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2822
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2823<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2824
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2825<div class="doc_code">
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2826<pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2827%RT = type { i8 , [10 x [20 x i32]], i8 }
2828%ST = type { i32, double, %RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2829
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2830define i32* %foo(%ST* %s) {
2831entry:
2832 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
2833 ret i32* %reg
2834}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2835</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2836</div>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2837
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2838<h5>Semantics:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2839
2840<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
John Criswellc1f786c2005-05-13 22:25:59 +0000[diff] [blame]2841on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
Reid Spencer85f5b5b2006-12-04 21:29:24 +0000[diff] [blame]2842and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
Reid Spencer42ddd842006-12-03 16:53:48 +0000[diff] [blame]2843<a href="#t_integer">integer</a> type but the value will always be sign extended
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2844to 64-bits. <a href="#t_struct">Structure</a> types require <tt>i32</tt>
Reid Spencer42ddd842006-12-03 16:53:48 +0000[diff] [blame]2845<b>constants</b>.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2846
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2847<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2848type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2849}</tt>' type, a structure. The second index indexes into the third element of
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2850the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
2851i8 }</tt>' type, another structure. The third index indexes into the second
2852element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2853array. The two dimensions of the array are subscripted into, yielding an
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2854'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
2855to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2856
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]2857<p>Note that it is perfectly legal to index partially through a
2858structure, returning a pointer to an inner element. Because of this,
2859the LLVM code for the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2860
2861<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2862 define i32* %foo(%ST* %s) {
2863 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2864 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
2865 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2866 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
2867 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
2868 ret i32* %t5
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2869 }
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2870</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2871
2872<p>Note that it is undefined to access an array out of bounds: array and
2873pointer indexes must always be within the defined bounds of the array type.
2874The one exception for this rules is zero length arrays. These arrays are
2875defined to be accessible as variable length arrays, which requires access
2876beyond the zero'th element.</p>
2877
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]2878<p>The getelementptr instruction is often confusing. For some more insight
2879into how it works, see <a href="GetElementPtr.html">the getelementptr
2880FAQ</a>.</p>
2881
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2882<h5>Example:</h5>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2883
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2884<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2885 <i>; yields [12 x i8]*:aptr</i>
2886 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2887</pre>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]2888</div>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]2889
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2890<!-- ======================================================================= -->
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]2891<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2892</div>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2893<div class="doc_text">
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]2894<p>The instructions in this category are the conversion instructions (casting)
2895which all take a single operand and a type. They perform various bit conversions
2896on the operand.</p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2897</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2898
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]2899<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]2900<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2901 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
2902</div>
2903<div class="doc_text">
2904
2905<h5>Syntax:</h5>
2906<pre>
2907 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2908</pre>
2909
2910<h5>Overview:</h5>
2911<p>
2912The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
2913</p>
2914
2915<h5>Arguments:</h5>
2916<p>
2917The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
2918be an <a href="#t_integer">integer</a> type, and a type that specifies the size
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]2919and type of the result, which must be an <a href="#t_integer">integer</a>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2920type. The bit size of <tt>value</tt> must be larger than the bit size of
2921<tt>ty2</tt>. Equal sized types are not allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2922
2923<h5>Semantics:</h5>
2924<p>
2925The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2926and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
2927larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
2928It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2929
2930<h5>Example:</h5>
2931<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2932 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]2933 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
2934 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2935</pre>
2936</div>
2937
2938<!-- _______________________________________________________________________ -->
2939<div class="doc_subsubsection">
2940 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
2941</div>
2942<div class="doc_text">
2943
2944<h5>Syntax:</h5>
2945<pre>
2946 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2947</pre>
2948
2949<h5>Overview:</h5>
2950<p>The '<tt>zext</tt>' instruction zero extends its operand to type
2951<tt>ty2</tt>.</p>
2952
2953
2954<h5>Arguments:</h5>
2955<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]2956<a href="#t_integer">integer</a> type, and a type to cast it to, which must
2957also be of <a href="#t_integer">integer</a> type. The bit size of the
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2958<tt>value</tt> must be smaller than the bit size of the destination type,
2959<tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2960
2961<h5>Semantics:</h5>
2962<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Chris Lattnerd1d25172007-05-24 19:13:27 +0000[diff] [blame]2963bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2964
Reid Spencerb5929522007-01-12 15:46:11 +0000[diff] [blame]2965<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2966
2967<h5>Example:</h5>
2968<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2969 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]2970 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2971</pre>
2972</div>
2973
2974<!-- _______________________________________________________________________ -->
2975<div class="doc_subsubsection">
2976 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
2977</div>
2978<div class="doc_text">
2979
2980<h5>Syntax:</h5>
2981<pre>
2982 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
2983</pre>
2984
2985<h5>Overview:</h5>
2986<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
2987
2988<h5>Arguments:</h5>
2989<p>
2990The '<tt>sext</tt>' instruction takes a value to cast, which must be of
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]2991<a href="#t_integer">integer</a> type, and a type to cast it to, which must
2992also be of <a href="#t_integer">integer</a> type. The bit size of the
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]2993<tt>value</tt> must be smaller than the bit size of the destination type,
2994<tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2995
2996<h5>Semantics:</h5>
2997<p>
2998The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
2999bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
Chris Lattnerd1d25172007-05-24 19:13:27 +0000[diff] [blame]3000the type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3001
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3002<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3003
3004<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3005<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3006 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3007 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3008</pre>
3009</div>
3010
3011<!-- _______________________________________________________________________ -->
3012<div class="doc_subsubsection">
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]3013 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3014</div>
3015
3016<div class="doc_text">
3017
3018<h5>Syntax:</h5>
3019
3020<pre>
3021 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3022</pre>
3023
3024<h5>Overview:</h5>
3025<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3026<tt>ty2</tt>.</p>
3027
3028
3029<h5>Arguments:</h5>
3030<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3031 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3032cast it to. The size of <tt>value</tt> must be larger than the size of
3033<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3034<i>no-op cast</i>.</p>
3035
3036<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3037<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3038<a href="#t_floating">floating point</a> type to a smaller
3039<a href="#t_floating">floating point</a> type. If the value cannot fit within
3040the destination type, <tt>ty2</tt>, then the results are undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]3041
3042<h5>Example:</h5>
3043<pre>
3044 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3045 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3046</pre>
3047</div>
3048
3049<!-- _______________________________________________________________________ -->
3050<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3051 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3052</div>
3053<div class="doc_text">
3054
3055<h5>Syntax:</h5>
3056<pre>
3057 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3058</pre>
3059
3060<h5>Overview:</h5>
3061<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3062floating point value.</p>
3063
3064<h5>Arguments:</h5>
3065<p>The '<tt>fpext</tt>' instruction takes a
3066<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3067and a <a href="#t_floating">floating point</a> type to cast it to. The source
3068type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3069
3070<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3071<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3072<a href="#t_floating">floating point</a> type to a larger
3073<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3074used to make a <i>no-op cast</i> because it always changes bits. Use
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3075<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3076
3077<h5>Example:</h5>
3078<pre>
3079 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3080 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3081</pre>
3082</div>
3083
3084<!-- _______________________________________________________________________ -->
3085<div class="doc_subsubsection">
Reid Spencer24d6da52007-01-21 00:29:26 +0000[diff] [blame]3086 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3087</div>
3088<div class="doc_text">
3089
3090<h5>Syntax:</h5>
3091<pre>
3092 &lt;result&gt; = fp2uint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3093</pre>
3094
3095<h5>Overview:</h5>
3096<p>The '<tt>fp2uint</tt>' converts a floating point <tt>value</tt> to its
3097unsigned integer equivalent of type <tt>ty2</tt>.
3098</p>
3099
3100<h5>Arguments:</h5>
3101<p>The '<tt>fp2uint</tt>' instruction takes a value to cast, which must be a
3102<a href="#t_floating">floating point</a> value, and a type to cast it to, which
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]3103must be an <a href="#t_integer">integer</a> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3104
3105<h5>Semantics:</h5>
3106<p> The '<tt>fp2uint</tt>' instruction converts its
3107<a href="#t_floating">floating point</a> operand into the nearest (rounding
3108towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3109the results are undefined.</p>
3110
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3111<p>When converting to i1, the conversion is done as a comparison against
3112zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
3113If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3114
3115<h5>Example:</h5>
3116<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3117 %X = fp2uint double 123.0 to i32 <i>; yields i32:123</i>
3118 %Y = fp2uint float 1.0E+300 to i1 <i>; yields i1:true</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3119 %X = fp2uint float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3120</pre>
3121</div>
3122
3123<!-- _______________________________________________________________________ -->
3124<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3125 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3126</div>
3127<div class="doc_text">
3128
3129<h5>Syntax:</h5>
3130<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3131 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3132</pre>
3133
3134<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3135<p>The '<tt>fptosi</tt>' instruction converts
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3136<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]3137</p>
3138
3139
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3140<h5>Arguments:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3141<p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3142<a href="#t_floating">floating point</a> value, and a type to cast it to, which
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]3143must also be an <a href="#t_integer">integer</a> type.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3144
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3145<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3146<p>The '<tt>fptosi</tt>' instruction converts its
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3147<a href="#t_floating">floating point</a> operand into the nearest (rounding
3148towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3149the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3150
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3151<p>When converting to i1, the conversion is done as a comparison against
3152zero. If the <tt>value</tt> was zero, the i1 result will be <tt>false</tt>.
3153If the <tt>value</tt> was non-zero, the i1 result will be <tt>true</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3154
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]3155<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3156<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3157 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
3158 %Y = fptosi float 1.0E-247 to i1 <i>; yields i1:true</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3159 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3160</pre>
3161</div>
3162
3163<!-- _______________________________________________________________________ -->
3164<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3165 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3166</div>
3167<div class="doc_text">
3168
3169<h5>Syntax:</h5>
3170<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3171 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3172</pre>
3173
3174<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3175<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3176integer and converts that value to the <tt>ty2</tt> type.</p>
3177
3178
3179<h5>Arguments:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3180<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be an
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]3181<a href="#t_integer">integer</a> value, and a type to cast it to, which must
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3182be a <a href="#t_floating">floating point</a> type.</p>
3183
3184<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3185<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3186integer quantity and converts it to the corresponding floating point value. If
Jeff Cohencb757312007-04-22 14:56:37 +0000[diff] [blame]3187the value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3188
3189
3190<h5>Example:</h5>
3191<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3192 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3193 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3194</pre>
3195</div>
3196
3197<!-- _______________________________________________________________________ -->
3198<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3199 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3200</div>
3201<div class="doc_text">
3202
3203<h5>Syntax:</h5>
3204<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3205 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3206</pre>
3207
3208<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3209<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3210integer and converts that value to the <tt>ty2</tt> type.</p>
3211
3212<h5>Arguments:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3213<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be an
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]3214<a href="#t_integer">integer</a> value, and a type to cast it to, which must be
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3215a <a href="#t_floating">floating point</a> type.</p>
3216
3217<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]3218<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3219integer quantity and converts it to the corresponding floating point value. If
Jeff Cohencb757312007-04-22 14:56:37 +0000[diff] [blame]3220the value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3221
3222<h5>Example:</h5>
3223<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3224 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3225 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3226</pre>
3227</div>
3228
3229<!-- _______________________________________________________________________ -->
3230<div class="doc_subsubsection">
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3231 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3232</div>
3233<div class="doc_text">
3234
3235<h5>Syntax:</h5>
3236<pre>
3237 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3238</pre>
3239
3240<h5>Overview:</h5>
3241<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3242the integer type <tt>ty2</tt>.</p>
3243
3244<h5>Arguments:</h5>
3245<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3246must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3247<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3248
3249<h5>Semantics:</h5>
3250<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3251<tt>ty2</tt> by interpreting the pointer value as an integer and either
3252truncating or zero extending that value to the size of the integer type. If
3253<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3254<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3255are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3256change.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3257
3258<h5>Example:</h5>
3259<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3260 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3261 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3262</pre>
3263</div>
3264
3265<!-- _______________________________________________________________________ -->
3266<div class="doc_subsubsection">
3267 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3268</div>
3269<div class="doc_text">
3270
3271<h5>Syntax:</h5>
3272<pre>
3273 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3274</pre>
3275
3276<h5>Overview:</h5>
3277<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3278a pointer type, <tt>ty2</tt>.</p>
3279
3280<h5>Arguments:</h5>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3281<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3282value to cast, and a type to cast it to, which must be a
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]3283<a href="#t_pointer">pointer</a> type.
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3284
3285<h5>Semantics:</h5>
3286<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3287<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3288the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3289size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3290the size of a pointer then a zero extension is done. If they are the same size,
3291nothing is done (<i>no-op cast</i>).</p>
3292
3293<h5>Example:</h5>
3294<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3295 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3296 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3297 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3298</pre>
3299</div>
3300
3301<!-- _______________________________________________________________________ -->
3302<div class="doc_subsubsection">
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3303 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3304</div>
3305<div class="doc_text">
3306
3307<h5>Syntax:</h5>
3308<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3309 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3310</pre>
3311
3312<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3313<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3314<tt>ty2</tt> without changing any bits.</p>
3315
3316<h5>Arguments:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3317<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3318a first class value, and a type to cast it to, which must also be a <a
3319 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
Reid Spencer19b569f2007-01-09 20:08:58 +0000[diff] [blame]3320and the destination type, <tt>ty2</tt>, must be identical. If the source
3321type is a pointer, the destination type must also be a pointer.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3322
3323<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]3324<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]3325<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3326this conversion. The conversion is done as if the <tt>value</tt> had been
3327stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3328converted to other pointer types with this instruction. To convert pointers to
3329other types, use the <a href="#i_inttoptr">inttoptr</a> or
3330<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]3331
3332<h5>Example:</h5>
3333<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3334 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3335 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3336 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]3337</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3338</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3339
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3340<!-- ======================================================================= -->
3341<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3342<div class="doc_text">
3343<p>The instructions in this category are the "miscellaneous"
3344instructions, which defy better classification.</p>
3345</div>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3346
3347<!-- _______________________________________________________________________ -->
3348<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3349</div>
3350<div class="doc_text">
3351<h5>Syntax:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3352<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3353</pre>
3354<h5>Overview:</h5>
3355<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
3356of its two integer operands.</p>
3357<h5>Arguments:</h5>
3358<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3359the condition code indicating the kind of comparison to perform. It is not
3360a value, just a keyword. The possible condition code are:
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3361<ol>
3362 <li><tt>eq</tt>: equal</li>
3363 <li><tt>ne</tt>: not equal </li>
3364 <li><tt>ugt</tt>: unsigned greater than</li>
3365 <li><tt>uge</tt>: unsigned greater or equal</li>
3366 <li><tt>ult</tt>: unsigned less than</li>
3367 <li><tt>ule</tt>: unsigned less or equal</li>
3368 <li><tt>sgt</tt>: signed greater than</li>
3369 <li><tt>sge</tt>: signed greater or equal</li>
3370 <li><tt>slt</tt>: signed less than</li>
3371 <li><tt>sle</tt>: signed less or equal</li>
3372</ol>
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]3373<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Reid Spencer350f8aa2007-01-04 05:19:58 +0000[diff] [blame]3374<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3375<h5>Semantics:</h5>
3376<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3377the condition code given as <tt>cond</tt>. The comparison performed always
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3378yields a <a href="#t_primitive">i1</a> result, as follows:
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3379<ol>
3380 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3381 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3382 </li>
3383 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3384 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3385 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3386 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3387 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3388 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3389 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3390 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3391 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3392 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3393 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3394 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3395 <li><tt>sge</tt>: interprets the operands as signed values and yields
3396 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3397 <li><tt>slt</tt>: interprets the operands as signed values and yields
3398 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3399 <li><tt>sle</tt>: interprets the operands as signed values and yields
3400 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3401</ol>
3402<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3403values are compared as if they were integers.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3404
3405<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3406<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3407 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3408 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3409 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3410 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3411 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3412</pre>
3413</div>
3414
3415<!-- _______________________________________________________________________ -->
3416<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3417</div>
3418<div class="doc_text">
3419<h5>Syntax:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3420<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3421</pre>
3422<h5>Overview:</h5>
3423<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3424of its floating point operands.</p>
3425<h5>Arguments:</h5>
3426<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3427the condition code indicating the kind of comparison to perform. It is not
3428a value, just a keyword. The possible condition code are:
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3429<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3430 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3431 <li><tt>oeq</tt>: ordered and equal</li>
3432 <li><tt>ogt</tt>: ordered and greater than </li>
3433 <li><tt>oge</tt>: ordered and greater than or equal</li>
3434 <li><tt>olt</tt>: ordered and less than </li>
3435 <li><tt>ole</tt>: ordered and less than or equal</li>
3436 <li><tt>one</tt>: ordered and not equal</li>
3437 <li><tt>ord</tt>: ordered (no nans)</li>
3438 <li><tt>ueq</tt>: unordered or equal</li>
3439 <li><tt>ugt</tt>: unordered or greater than </li>
3440 <li><tt>uge</tt>: unordered or greater than or equal</li>
3441 <li><tt>ult</tt>: unordered or less than </li>
3442 <li><tt>ule</tt>: unordered or less than or equal</li>
3443 <li><tt>une</tt>: unordered or not equal</li>
3444 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3445 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3446</ol>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3447<p><i>Ordered</i> means that neither operand is a QNAN while
Reid Spencer93a49852006-12-06 07:08:07 +0000[diff] [blame]3448<i>unordered</i> means that either operand may be a QNAN.</p>
Reid Spencer350f8aa2007-01-04 05:19:58 +0000[diff] [blame]3449<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3450<a href="#t_floating">floating point</a> typed. They must have identical
3451types.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3452<h5>Semantics:</h5>
3453<p>The '<tt>fcmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3454the condition code given as <tt>cond</tt>. The comparison performed always
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3455yields a <a href="#t_primitive">i1</a> result, as follows:
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3456<ol>
3457 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3458 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3459 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3460 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3461 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3462 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3463 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3464 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3465 <tt>var1</tt> is less than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3466 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3467 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3468 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3469 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3470 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3471 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3472 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3473 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3474 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3475 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3476 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3477 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3478 <tt>var1</tt> is less than <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3479 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3480 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3481 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3482 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]3483 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3484 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3485</ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]3486
3487<h5>Example:</h5>
3488<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3489 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3490 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3491 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3492</pre>
3493</div>
3494
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3495<!-- _______________________________________________________________________ -->
3496<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3497Instruction</a> </div>
3498<div class="doc_text">
3499<h5>Syntax:</h5>
3500<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3501<h5>Overview:</h5>
3502<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3503the SSA graph representing the function.</p>
3504<h5>Arguments:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3505<p>The type of the incoming values is specified with the first type
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3506field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3507as arguments, with one pair for each predecessor basic block of the
3508current block. Only values of <a href="#t_firstclass">first class</a>
3509type may be used as the value arguments to the PHI node. Only labels
3510may be used as the label arguments.</p>
3511<p>There must be no non-phi instructions between the start of a basic
3512block and the PHI instructions: i.e. PHI instructions must be first in
3513a basic block.</p>
3514<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3515<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3516specified by the pair corresponding to the predecessor basic block that executed
3517just prior to the current block.</p>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3518<h5>Example:</h5>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3519<pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add i32 %indvar, 1<br> br label %Loop<br></pre>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]3520</div>
3521
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3522<!-- _______________________________________________________________________ -->
3523<div class="doc_subsubsection">
3524 <a name="i_select">'<tt>select</tt>' Instruction</a>
3525</div>
3526
3527<div class="doc_text">
3528
3529<h5>Syntax:</h5>
3530
3531<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3532 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3533</pre>
3534
3535<h5>Overview:</h5>
3536
3537<p>
3538The '<tt>select</tt>' instruction is used to choose one value based on a
3539condition, without branching.
3540</p>
3541
3542
3543<h5>Arguments:</h5>
3544
3545<p>
3546The '<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.
3547</p>
3548
3549<h5>Semantics:</h5>
3550
3551<p>
3552If the boolean condition evaluates to true, the instruction returns the first
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3553value argument; otherwise, it returns the second value argument.
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3554</p>
3555
3556<h5>Example:</h5>
3557
3558<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3559 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]3560</pre>
3561</div>
3562
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]3563
3564<!-- _______________________________________________________________________ -->
3565<div class="doc_subsubsection">
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3566 <a name="i_call">'<tt>call</tt>' Instruction</a>
3567</div>
3568
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3569<div class="doc_text">
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3570
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3571<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3572<pre>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3573 &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]3574</pre>
3575
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3576<h5>Overview:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3577
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3578<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3579
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3580<h5>Arguments:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3581
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3582<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3583
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3584<ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3585 <li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3586 <p>The optional "tail" marker indicates whether the callee function accesses
3587 any allocas or varargs in the caller. If the "tail" marker is present, the
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3588 function call is eligible for tail call optimization. Note that calls may
3589 be marked "tail" even if they do not occur before a <a
3590 href="#i_ret"><tt>ret</tt></a> instruction.
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3591 </li>
3592 <li>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]3593 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3594 convention</a> the call should use. If none is specified, the call defaults
3595 to using C calling conventions.
3596 </li>
3597 <li>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3598 <p>'<tt>ty</tt>': shall be the signature of the pointer to function value
3599 being invoked. The argument types must match the types implied by this
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3600 signature. This type can be omitted if the function is not varargs and
3601 if the function type does not return a pointer to a function.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3602 </li>
3603 <li>
3604 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
3605 be invoked. In most cases, this is a direct function invocation, but
3606 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3607 to function value.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3608 </li>
3609 <li>
3610 <p>'<tt>function args</tt>': argument list whose types match the
Reid Spencera7e302a2005-05-01 22:22:57 +0000[diff] [blame]3611 function signature argument types. All arguments must be of
3612 <a href="#t_firstclass">first class</a> type. If the function signature
3613 indicates the function accepts a variable number of arguments, the extra
3614 arguments can be specified.</p>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3615 </li>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3616</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3617
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3618<h5>Semantics:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3619
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3620<p>The '<tt>call</tt>' instruction is used to cause control flow to
3621transfer to a specified function, with its incoming arguments bound to
3622the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
3623instruction in the called function, control flow continues with the
3624instruction after the function call, and the return value of the
3625function is bound to the result argument. This is a simpler case of
3626the <a href="#i_invoke">invoke</a> instruction.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3627
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3628<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3629
3630<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3631 %retval = call i32 %test(i32 %argc)
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3632 call i32(i8 *, ...) *%printf(i8 * %msg, i32 12, i8 42);
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3633 %X = tail call i32 %foo()
3634 %Y = tail call <a href="#callingconv">fastcc</a> i32 %foo()
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]3635</pre>
3636
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3637</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3638
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3639<!-- _______________________________________________________________________ -->
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3640<div class="doc_subsubsection">
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]3641 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3642</div>
3643
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3644<div class="doc_text">
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3645
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3646<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3647
3648<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3649 &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]3650</pre>
3651
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3652<h5>Overview:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3653
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3654<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3655the "variable argument" area of a function call. It is used to implement the
3656<tt>va_arg</tt> macro in C.</p>
3657
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3658<h5>Arguments:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3659
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3660<p>This instruction takes a <tt>va_list*</tt> value and the type of
3661the argument. It returns a value of the specified argument type and
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3662increments the <tt>va_list</tt> to point to the next argument. The
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3663actual type of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3664
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3665<h5>Semantics:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3666
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3667<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
3668type from the specified <tt>va_list</tt> and causes the
3669<tt>va_list</tt> to point to the next argument. For more information,
3670see the variable argument handling <a href="#int_varargs">Intrinsic
3671Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3672
3673<p>It is legal for this instruction to be called in a function which does not
3674take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3675function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3676
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3677<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3678href="#intrinsics">intrinsic function</a> because it takes a type as an
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3679argument.</p>
3680
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]3681<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]3682
3683<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
3684
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3685</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3686
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3687<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3688<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
3689<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3690
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3691<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3692
3693<p>LLVM supports the notion of an "intrinsic function". These functions have
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]3694well known names and semantics and are required to follow certain restrictions.
3695Overall, these intrinsics represent an extension mechanism for the LLVM
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3696language that does not require changing all of the transformations in LLVM when
Gabor Greif04367bf2007-07-06 22:07:22 +0000[diff] [blame]3697adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3698
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]3699<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3700prefix is reserved in LLVM for intrinsic names; thus, function names may not
3701begin with this prefix. Intrinsic functions must always be external functions:
3702you cannot define the body of intrinsic functions. Intrinsic functions may
3703only be used in call or invoke instructions: it is illegal to take the address
3704of an intrinsic function. Additionally, because intrinsic functions are part
3705of the LLVM language, it is required if any are added that they be documented
3706here.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3707
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3708<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]3709a family of functions that perform the same operation but on different data
3710types. This is most frequent with the integer types. Since LLVM can represent
3711over 8 million different integer types, there is a way to declare an intrinsic
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3712that can be overloaded based on its arguments. Such an intrinsic will have the
3713names of its argument types encoded into its function name, each
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]3714preceded by a period. For example, the <tt>llvm.ctpop</tt> function can take an
3715integer of any width. This leads to a family of functions such as
3716<tt>i32 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i32 @llvm.ctpop.i29(i29 %val)</tt>.
3717</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3718
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]3719
3720<p>To learn how to add an intrinsic function, please see the
3721<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3722</p>
3723
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3724</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3725
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3726<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3727<div class="doc_subsection">
3728 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
3729</div>
3730
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3731<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3732
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3733<p>Variable argument support is defined in LLVM with the <a
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]3734 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3735intrinsic functions. These functions are related to the similarly
3736named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3737
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3738<p>All of these functions operate on arguments that use a
3739target-specific value type "<tt>va_list</tt>". The LLVM assembly
3740language reference manual does not define what this type is, so all
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3741transformations should be prepared to handle these functions regardless of
3742the type used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3743
Chris Lattner374ab302006-05-15 17:26:46 +0000[diff] [blame]3744<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]3745instruction and the variable argument handling intrinsic functions are
3746used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3747
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]3748<div class="doc_code">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3749<pre>
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]3750define i32 @test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3751 ; Initialize variable argument processing
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3752 %ap = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3753 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]3754 call void @llvm.va_start(i8* %ap2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3755
3756 ; Read a single integer argument
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3757 %tmp = va_arg i8** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3758
3759 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3760 %aq = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3761 %aq2 = bitcast i8** %aq to i8*
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3762 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]3763 call void @llvm.va_end(i8* %aq2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3764
3765 ; Stop processing of arguments.
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]3766 call void @llvm.va_end(i8* %ap2)
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3767 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3768}
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]3769
3770declare void @llvm.va_start(i8*)
3771declare void @llvm.va_copy(i8*, i8*)
3772declare void @llvm.va_end(i8*)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3773</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3774</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3775
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]3776</div>
3777
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3778<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3779<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3780 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3781</div>
3782
3783
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3784<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3785<h5>Syntax:</h5>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3786<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3787<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3788<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
3789<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
3790href="#i_va_arg">va_arg</a></tt>.</p>
3791
3792<h5>Arguments:</h5>
3793
3794<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
3795
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3796<h5>Semantics:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3797
3798<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
3799macro available in C. In a target-dependent way, it initializes the
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3800<tt>va_list</tt> element to which the argument points, so that the next call to
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3801<tt>va_arg</tt> will produce the first variable argument passed to the function.
3802Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3803last argument of the function as the compiler can figure that out.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3804
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3805</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3806
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3807<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3808<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3809 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3810</div>
3811
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3812<div class="doc_text">
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3813<h5>Syntax:</h5>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]3814<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3815<h5>Overview:</h5>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3816
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3817<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3818which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3819or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3820
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3821<h5>Arguments:</h5>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3822
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3823<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3824
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3825<h5>Semantics:</h5>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3826
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3827<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3828macro available in C. In a target-dependent way, it destroys the
3829<tt>va_list</tt> element to which the argument points. Calls to <a
3830href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
3831<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
3832<tt>llvm.va_end</tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]3833
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3834</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3835
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3836<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3837<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3838 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3839</div>
3840
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3841<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3842
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3843<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3844
3845<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]3846 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3847</pre>
3848
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3849<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3850
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3851<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
3852from the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3853
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3854<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3855
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]3856<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]3857The 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]3858
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3859
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]3860<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3861
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]3862<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
3863macro available in C. In a target-dependent way, it copies the source
3864<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
3865intrinsic is necessary because the <tt><a href="#int_va_start">
3866llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
3867example, memory allocation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3868
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3869</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]3870
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]3871<!-- ======================================================================= -->
3872<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3873 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
3874</div>
3875
3876<div class="doc_text">
3877
3878<p>
3879LLVM support for <a href="GarbageCollection.html">Accurate Garbage
3880Collection</a> requires the implementation and generation of these intrinsics.
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3881These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3882stack</a>, as well as garbage collector implementations that require <a
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3883href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3884Front-ends for type-safe garbage collected languages should generate these
3885intrinsics to make use of the LLVM garbage collectors. For more details, see <a
3886href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
3887</p>
3888</div>
3889
3890<!-- _______________________________________________________________________ -->
3891<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3892 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3893</div>
3894
3895<div class="doc_text">
3896
3897<h5>Syntax:</h5>
3898
3899<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]3900 declare void @llvm.gcroot(&lt;ty&gt;** %ptrloc, &lt;ty2&gt;* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3901</pre>
3902
3903<h5>Overview:</h5>
3904
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]3905<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]3906the code generator, and allows some metadata to be associated with it.</p>
3907
3908<h5>Arguments:</h5>
3909
3910<p>The first argument specifies the address of a stack object that contains the
3911root pointer. The second pointer (which must be either a constant or a global
3912value address) contains the meta-data to be associated with the root.</p>
3913
3914<h5>Semantics:</h5>
3915
3916<p>At runtime, a call to this intrinsics stores a null pointer into the "ptrloc"
3917location. At compile-time, the code generator generates information to allow
3918the runtime to find the pointer at GC safe points.
3919</p>
3920
3921</div>
3922
3923
3924<!-- _______________________________________________________________________ -->
3925<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3926 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3927</div>
3928
3929<div class="doc_text">
3930
3931<h5>Syntax:</h5>
3932
3933<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]3934 declare i8 * @llvm.gcread(i8 * %ObjPtr, i8 ** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3935</pre>
3936
3937<h5>Overview:</h5>
3938
3939<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
3940locations, allowing garbage collector implementations that require read
3941barriers.</p>
3942
3943<h5>Arguments:</h5>
3944
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3945<p>The second argument is the address to read from, which should be an address
3946allocated from the garbage collector. The first object is a pointer to the
3947start of the referenced object, if needed by the language runtime (otherwise
3948null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3949
3950<h5>Semantics:</h5>
3951
3952<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
3953instruction, but may be replaced with substantially more complex code by the
3954garbage collector runtime, as needed.</p>
3955
3956</div>
3957
3958
3959<!-- _______________________________________________________________________ -->
3960<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]3961 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3962</div>
3963
3964<div class="doc_text">
3965
3966<h5>Syntax:</h5>
3967
3968<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]3969 declare void @llvm.gcwrite(i8 * %P1, i8 * %Obj, i8 ** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3970</pre>
3971
3972<h5>Overview:</h5>
3973
3974<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
3975locations, allowing garbage collector implementations that require write
3976barriers (such as generational or reference counting collectors).</p>
3977
3978<h5>Arguments:</h5>
3979
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]3980<p>The first argument is the reference to store, the second is the start of the
3981object to store it to, and the third is the address of the field of Obj to
3982store to. If the runtime does not require a pointer to the object, Obj may be
3983null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]3984
3985<h5>Semantics:</h5>
3986
3987<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
3988instruction, but may be replaced with substantially more complex code by the
3989garbage collector runtime, as needed.</p>
3990
3991</div>
3992
3993
3994
3995<!-- ======================================================================= -->
3996<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]3997 <a name="int_codegen">Code Generator Intrinsics</a>
3998</div>
3999
4000<div class="doc_text">
4001<p>
4002These intrinsics are provided by LLVM to expose special features that may only
4003be implemented with code generator support.
4004</p>
4005
4006</div>
4007
4008<!-- _______________________________________________________________________ -->
4009<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4010 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4011</div>
4012
4013<div class="doc_text">
4014
4015<h5>Syntax:</h5>
4016<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4017 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4018</pre>
4019
4020<h5>Overview:</h5>
4021
4022<p>
Chris Lattner32b5d712006-10-15 20:05:59 +0000[diff] [blame]4023The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4024target-specific value indicating the return address of the current function
4025or one of its callers.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4026</p>
4027
4028<h5>Arguments:</h5>
4029
4030<p>
4031The argument to this intrinsic indicates which function to return the address
4032for. Zero indicates the calling function, one indicates its caller, etc. The
4033argument is <b>required</b> to be a constant integer value.
4034</p>
4035
4036<h5>Semantics:</h5>
4037
4038<p>
4039The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4040the return address of the specified call frame, or zero if it cannot be
4041identified. The value returned by this intrinsic is likely to be incorrect or 0
4042for arguments other than zero, so it should only be used for debugging purposes.
4043</p>
4044
4045<p>
4046Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +0000[diff] [blame]4047aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4048source-language caller.
4049</p>
4050</div>
4051
4052
4053<!-- _______________________________________________________________________ -->
4054<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4055 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4056</div>
4057
4058<div class="doc_text">
4059
4060<h5>Syntax:</h5>
4061<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4062 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4063</pre>
4064
4065<h5>Overview:</h5>
4066
4067<p>
Chris Lattner32b5d712006-10-15 20:05:59 +0000[diff] [blame]4068The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4069target-specific frame pointer value for the specified stack frame.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4070</p>
4071
4072<h5>Arguments:</h5>
4073
4074<p>
4075The argument to this intrinsic indicates which function to return the frame
4076pointer for. Zero indicates the calling function, one indicates its caller,
4077etc. The argument is <b>required</b> to be a constant integer value.
4078</p>
4079
4080<h5>Semantics:</h5>
4081
4082<p>
4083The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4084the frame address of the specified call frame, or zero if it cannot be
4085identified. The value returned by this intrinsic is likely to be incorrect or 0
4086for arguments other than zero, so it should only be used for debugging purposes.
4087</p>
4088
4089<p>
4090Note that calling this intrinsic does not prevent function inlining or other
Chris Lattnerb40bb382005-03-07 20:30:51 +0000[diff] [blame]4091aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4092source-language caller.
4093</p>
4094</div>
4095
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]4096<!-- _______________________________________________________________________ -->
4097<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4098 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4099</div>
4100
4101<div class="doc_text">
4102
4103<h5>Syntax:</h5>
4104<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4105 declare i8 *@llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4106</pre>
4107
4108<h5>Overview:</h5>
4109
4110<p>
4111The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4112the function stack, for use with <a href="#int_stackrestore">
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4113<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4114features like scoped automatic variable sized arrays in C99.
4115</p>
4116
4117<h5>Semantics:</h5>
4118
4119<p>
4120This intrinsic returns a opaque pointer value that can be passed to <a
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4121href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4122<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4123<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4124state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4125practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4126that were allocated after the <tt>llvm.stacksave</tt> was executed.
4127</p>
4128
4129</div>
4130
4131<!-- _______________________________________________________________________ -->
4132<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4133 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4134</div>
4135
4136<div class="doc_text">
4137
4138<h5>Syntax:</h5>
4139<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4140 declare void @llvm.stackrestore(i8 * %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4141</pre>
4142
4143<h5>Overview:</h5>
4144
4145<p>
4146The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4147the function stack to the state it was in when the corresponding <a
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4148href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4149useful for implementing language features like scoped automatic variable sized
4150arrays in C99.
4151</p>
4152
4153<h5>Semantics:</h5>
4154
4155<p>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4156See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]4157</p>
4158
4159</div>
4160
4161
4162<!-- _______________________________________________________________________ -->
4163<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4164 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]4165</div>
4166
4167<div class="doc_text">
4168
4169<h5>Syntax:</h5>
4170<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4171 declare void @llvm.prefetch(i8 * &lt;address&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4172 i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]4173</pre>
4174
4175<h5>Overview:</h5>
4176
4177
4178<p>
4179The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]4180a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4181no
4182effect on the behavior of the program but can change its performance
Chris Lattner2a615362005-02-28 19:47:14 +0000[diff] [blame]4183characteristics.
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]4184</p>
4185
4186<h5>Arguments:</h5>
4187
4188<p>
4189<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4190determining if the fetch should be for a read (0) or write (1), and
4191<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Chris Lattneraeffb4a2005-03-07 20:31:38 +0000[diff] [blame]4192locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]4193<tt>locality</tt> arguments must be constant integers.
4194</p>
4195
4196<h5>Semantics:</h5>
4197
4198<p>
4199This intrinsic does not modify the behavior of the program. In particular,
4200prefetches cannot trap and do not produce a value. On targets that support this
4201intrinsic, the prefetch can provide hints to the processor cache for better
4202performance.
4203</p>
4204
4205</div>
4206
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4207<!-- _______________________________________________________________________ -->
4208<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4209 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4210</div>
4211
4212<div class="doc_text">
4213
4214<h5>Syntax:</h5>
4215<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4216 declare void @llvm.pcmarker( i32 &lt;id&gt; )
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4217</pre>
4218
4219<h5>Overview:</h5>
4220
4221
4222<p>
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]4223The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4224(PC) in a region of
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4225code to simulators and other tools. The method is target specific, but it is
4226expected that the marker will use exported symbols to transmit the PC of the marker.
Jeff Cohen25d4f7e2005-11-11 02:15:27 +0000[diff] [blame]4227The marker makes no guarantees that it will remain with any specific instruction
Chris Lattnerd07c3f42005-11-15 06:07:55 +0000[diff] [blame]4228after optimizations. It is possible that the presence of a marker will inhibit
Chris Lattnerb3e7afd2006-03-24 07:16:10 +0000[diff] [blame]4229optimizations. The intended use is to be inserted after optimizations to allow
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]4230correlations of simulation runs.
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]4231</p>
4232
4233<h5>Arguments:</h5>
4234
4235<p>
4236<tt>id</tt> is a numerical id identifying the marker.
4237</p>
4238
4239<h5>Semantics:</h5>
4240
4241<p>
4242This intrinsic does not modify the behavior of the program. Backends that do not
4243support this intrinisic may ignore it.
4244</p>
4245
4246</div>
4247
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]4248<!-- _______________________________________________________________________ -->
4249<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4250 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]4251</div>
4252
4253<div class="doc_text">
4254
4255<h5>Syntax:</h5>
4256<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4257 declare i64 @llvm.readcyclecounter( )
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]4258</pre>
4259
4260<h5>Overview:</h5>
4261
4262
4263<p>
4264The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4265counter register (or similar low latency, high accuracy clocks) on those targets
4266that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4267As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4268should only be used for small timings.
4269</p>
4270
4271<h5>Semantics:</h5>
4272
4273<p>
4274When directly supported, reading the cycle counter should not modify any memory.
4275Implementations are allowed to either return a application specific value or a
4276system wide value. On backends without support, this is lowered to a constant 0.
4277</p>
4278
4279</div>
4280
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4281<!-- ======================================================================= -->
4282<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4283 <a name="int_libc">Standard C Library Intrinsics</a>
4284</div>
4285
4286<div class="doc_text">
4287<p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4288LLVM provides intrinsics for a few important standard C library functions.
4289These intrinsics allow source-language front-ends to pass information about the
4290alignment of the pointer arguments to the code generator, providing opportunity
4291for more efficient code generation.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4292</p>
4293
4294</div>
4295
4296<!-- _______________________________________________________________________ -->
4297<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4298 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4299</div>
4300
4301<div class="doc_text">
4302
4303<h5>Syntax:</h5>
4304<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4305 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4306 i32 &lt;len&gt;, i32 &lt;align&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4307 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4308 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4309</pre>
4310
4311<h5>Overview:</h5>
4312
4313<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4314The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4315location to the destination location.
4316</p>
4317
4318<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4319Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4320intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4321</p>
4322
4323<h5>Arguments:</h5>
4324
4325<p>
4326The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4327the source. The third argument is an integer argument
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4328specifying the number of bytes to copy, and the fourth argument is the alignment
4329of the source and destination locations.
4330</p>
4331
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4332<p>
4333If 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]4334the caller guarantees that both the source and destination pointers are aligned
4335to that boundary.
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4336</p>
4337
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4338<h5>Semantics:</h5>
4339
4340<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4341The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]4342location to the destination location, which are not allowed to overlap. It
4343copies "len" bytes of memory over. If the argument is known to be aligned to
4344some boundary, this can be specified as the fourth argument, otherwise it should
4345be set to 0 or 1.
4346</p>
4347</div>
4348
4349
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4350<!-- _______________________________________________________________________ -->
4351<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4352 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4353</div>
4354
4355<div class="doc_text">
4356
4357<h5>Syntax:</h5>
4358<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4359 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4360 i32 &lt;len&gt;, i32 &lt;align&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4361 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4362 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4363</pre>
4364
4365<h5>Overview:</h5>
4366
4367<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4368The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4369location to the destination location. It is similar to the
4370'<tt>llvm.memcmp</tt>' intrinsic but allows the two memory locations to overlap.
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4371</p>
4372
4373<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4374Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4375intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4376</p>
4377
4378<h5>Arguments:</h5>
4379
4380<p>
4381The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4382the source. The third argument is an integer argument
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4383specifying the number of bytes to copy, and the fourth argument is the alignment
4384of the source and destination locations.
4385</p>
4386
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4387<p>
4388If 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]4389the caller guarantees that the source and destination pointers are aligned to
4390that boundary.
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]4391</p>
4392
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4393<h5>Semantics:</h5>
4394
4395<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4396The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]4397location to the destination location, which may overlap. It
4398copies "len" bytes of memory over. If the argument is known to be aligned to
4399some boundary, this can be specified as the fourth argument, otherwise it should
4400be set to 0 or 1.
4401</p>
4402</div>
4403
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]4404
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4405<!-- _______________________________________________________________________ -->
4406<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4407 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4408</div>
4409
4410<div class="doc_text">
4411
4412<h5>Syntax:</h5>
4413<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4414 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4415 i32 &lt;len&gt;, i32 &lt;align&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4416 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4417 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4418</pre>
4419
4420<h5>Overview:</h5>
4421
4422<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4423The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4424byte value.
4425</p>
4426
4427<p>
4428Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4429does not return a value, and takes an extra alignment argument.
4430</p>
4431
4432<h5>Arguments:</h5>
4433
4434<p>
4435The first argument is a pointer to the destination to fill, the second is the
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4436byte value to fill it with, the third argument is an integer
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4437argument specifying the number of bytes to fill, and the fourth argument is the
4438known alignment of destination location.
4439</p>
4440
4441<p>
4442If 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]4443the caller guarantees that the destination pointer is aligned to that boundary.
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4444</p>
4445
4446<h5>Semantics:</h5>
4447
4448<p>
Chris Lattner5b310c32006-03-03 00:07:20 +0000[diff] [blame]4449The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4450the
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]4451destination location. If the argument is known to be aligned to some boundary,
4452this can be specified as the fourth argument, otherwise it should be set to 0 or
44531.
4454</p>
4455</div>
4456
4457
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]4458<!-- _______________________________________________________________________ -->
4459<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4460 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4461</div>
4462
4463<div class="doc_text">
4464
4465<h5>Syntax:</h5>
4466<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4467 declare float @llvm.sqrt.f32(float %Val)
4468 declare double @llvm.sqrt.f64(double %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4469</pre>
4470
4471<h5>Overview:</h5>
4472
4473<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4474The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4475returning the same value as the libm '<tt>sqrt</tt>' function would. Unlike
4476<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
4477negative numbers (which allows for better optimization).
4478</p>
4479
4480<h5>Arguments:</h5>
4481
4482<p>
4483The argument and return value are floating point numbers of the same type.
4484</p>
4485
4486<h5>Semantics:</h5>
4487
4488<p>
Dan Gohmand6257fe2007-07-16 14:37:41 +0000[diff] [blame]4489This function returns the sqrt of the specified operand if it is a nonnegative
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]4490floating point number.
4491</p>
4492</div>
4493
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4494<!-- _______________________________________________________________________ -->
4495<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4496 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4497</div>
4498
4499<div class="doc_text">
4500
4501<h5>Syntax:</h5>
4502<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4503 declare float @llvm.powi.f32(float %Val, i32 %power)
4504 declare double @llvm.powi.f64(double %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]4505</pre>
4506
4507<h5>Overview:</h5>
4508
4509<p>
4510The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
4511specified (positive or negative) power. The order of evaluation of
4512multiplications is not defined.
4513</p>
4514
4515<h5>Arguments:</h5>
4516
4517<p>
4518The second argument is an integer power, and the first is a value to raise to
4519that power.
4520</p>
4521
4522<h5>Semantics:</h5>
4523
4524<p>
4525This function returns the first value raised to the second power with an
4526unspecified sequence of rounding operations.</p>
4527</div>
4528
4529
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4530<!-- ======================================================================= -->
4531<div class="doc_subsection">
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4532 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4533</div>
4534
4535<div class="doc_text">
4536<p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4537LLVM provides intrinsics for a few important bit manipulation operations.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4538These allow efficient code generation for some algorithms.
4539</p>
4540
4541</div>
4542
4543<!-- _______________________________________________________________________ -->
4544<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4545 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4546</div>
4547
4548<div class="doc_text">
4549
4550<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4551<p>This is an overloaded intrinsic function. You can use bswap on any integer
4552type that is an even number of bytes (i.e. BitWidth % 16 == 0). Note the suffix
4553that includes the type for the result and the operand.
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4554<pre>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4555 declare i16 @llvm.bswap.i16.i16(i16 &lt;id&gt;)
4556 declare i32 @llvm.bswap.i32.i32(i32 &lt;id&gt;)
Reid Spencer543ab1d2007-04-02 00:19:52 +0000[diff] [blame]4557 declare i64 @llvm.bswap.i64.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4558</pre>
4559
4560<h5>Overview:</h5>
4561
4562<p>
Reid Spencer338ea092007-04-02 02:25:19 +0000[diff] [blame]4563The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4564values with an even number of bytes (positive multiple of 16 bits). These are
4565useful for performing operations on data that is not in the target's native
4566byte order.
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4567</p>
4568
4569<h5>Semantics:</h5>
4570
4571<p>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4572The <tt>llvm.bswap.16.i16</tt> intrinsic returns an i16 value that has the high
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4573and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
4574intrinsic returns an i32 value that has the four bytes of the input i32
4575swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4576i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48.i48</tt>,
4577<tt>llvm.bswap.i64.i64</tt> and other intrinsics extend this concept to
4578additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]4579</p>
4580
4581</div>
4582
4583<!-- _______________________________________________________________________ -->
4584<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4585 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4586</div>
4587
4588<div class="doc_text">
4589
4590<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4591<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
4592width. Not all targets support all bit widths however.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4593<pre>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4594 declare i32 @llvm.ctpop.i8 (i8 &lt;src&gt;)
4595 declare i32 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4596 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4597 declare i32 @llvm.ctpop.i64(i64 &lt;src&gt;)
4598 declare i32 @llvm.ctpop.i256(i256 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4599</pre>
4600
4601<h5>Overview:</h5>
4602
4603<p>
Chris Lattnerec6cb612006-01-16 22:38:59 +0000[diff] [blame]4604The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
4605value.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4606</p>
4607
4608<h5>Arguments:</h5>
4609
4610<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]4611The only argument is the value to be counted. The argument may be of any
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]4612integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4613</p>
4614
4615<h5>Semantics:</h5>
4616
4617<p>
4618The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
4619</p>
4620</div>
4621
4622<!-- _______________________________________________________________________ -->
4623<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]4624 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4625</div>
4626
4627<div class="doc_text">
4628
4629<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4630<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
4631integer bit width. Not all targets support all bit widths however.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4632<pre>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4633 declare i32 @llvm.ctlz.i8 (i8 &lt;src&gt;)
4634 declare i32 @llvm.ctlz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4635 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4636 declare i32 @llvm.ctlz.i64(i64 &lt;src&gt;)
4637 declare i32 @llvm.ctlz.i256(i256 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4638</pre>
4639
4640<h5>Overview:</h5>
4641
4642<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4643The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
4644leading zeros in a variable.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4645</p>
4646
4647<h5>Arguments:</h5>
4648
4649<p>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]4650The only argument is the value to be counted. The argument may be of any
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]4651integer type. The return type must match the argument type.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4652</p>
4653
4654<h5>Semantics:</h5>
4655
4656<p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4657The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
4658in a variable. If the src == 0 then the result is the size in bits of the type
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4659of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]4660</p>
4661</div>
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]4662
4663
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4664
4665<!-- _______________________________________________________________________ -->
4666<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]4667 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4668</div>
4669
4670<div class="doc_text">
4671
4672<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4673<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
4674integer bit width. Not all targets support all bit widths however.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4675<pre>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4676 declare i32 @llvm.cttz.i8 (i8 &lt;src&gt;)
4677 declare i32 @llvm.cttz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]4678 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]4679 declare i32 @llvm.cttz.i64(i64 &lt;src&gt;)
4680 declare i32 @llvm.cttz.i256(i256 &lt;src&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4681</pre>
4682
4683<h5>Overview:</h5>
4684
4685<p>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]4686The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
4687trailing zeros.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4688</p>
4689
4690<h5>Arguments:</h5>
4691
4692<p>
4693The only argument is the value to be counted. The argument may be of any
Reid Spencera5173382007-01-04 16:43:23 +0000[diff] [blame]4694integer type. The return type must match the argument type.
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]4695</p>
4696
4697<h5>Semantics:</h5>
4698
4699<p>
4700The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
4701in a variable. If the src == 0 then the result is the size in bits of the type
4702of src. For example, <tt>llvm.cttz(2) = 1</tt>.
4703</p>
4704</div>
4705
Reid Spencer497d93e2007-04-01 08:27:01 +0000[diff] [blame]4706<!-- _______________________________________________________________________ -->
4707<div class="doc_subsubsection">
Reid Spencerbeacf662007-04-10 02:51:31 +0000[diff] [blame]4708 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4709</div>
4710
4711<div class="doc_text">
4712
4713<h5>Syntax:</h5>
Reid Spencerbeacf662007-04-10 02:51:31 +0000[diff] [blame]4714<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4715on any integer bit width.
4716<pre>
Reid Spencerbeacf662007-04-10 02:51:31 +0000[diff] [blame]4717 declare i17 @llvm.part.select.i17.i17 (i17 %val, i32 %loBit, i32 %hiBit)
4718 declare i29 @llvm.part.select.i29.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4719</pre>
4720
4721<h5>Overview:</h5>
Reid Spencerbeacf662007-04-10 02:51:31 +0000[diff] [blame]4722<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4723range of bits from an integer value and returns them in the same bit width as
4724the original value.</p>
4725
4726<h5>Arguments:</h5>
4727<p>The first argument, <tt>%val</tt> and the result may be integer types of
4728any bit width but they must have the same bit width. The second and third
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4729arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4730
4731<h5>Semantics:</h5>
Reid Spencerbeacf662007-04-10 02:51:31 +0000[diff] [blame]4732<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]4733of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
4734<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
4735operates in forward mode.</p>
4736<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
4737right by <tt>%loBit</tt> bits and then ANDing it with a mask with
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4738only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
4739<ol>
4740 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
4741 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
4742 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
4743 to determine the number of bits to retain.</li>
4744 <li>A mask of the retained bits is created by shifting a -1 value.</li>
4745 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
4746</ol>
Reid Spencerd6a85b52007-05-14 16:14:57 +0000[diff] [blame]4747<p>In reverse mode, a similar computation is made except that the bits are
4748returned in the reverse order. So, for example, if <tt>X</tt> has the value
4749<tt>i16 0x0ACF (101011001111)</tt> and we apply
4750<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
4751<tt>i16 0x0026 (000000100110)</tt>.</p>
Reid Spencera13ba7d2007-04-01 19:00:37 +0000[diff] [blame]4752</div>
4753
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]4754<div class="doc_subsubsection">
4755 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
4756</div>
4757
4758<div class="doc_text">
4759
4760<h5>Syntax:</h5>
4761<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
4762on any integer bit width.
4763<pre>
4764 declare i17 @llvm.part.set.i17.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
4765 declare i29 @llvm.part.set.i29.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
4766</pre>
4767
4768<h5>Overview:</h5>
4769<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
4770of bits in an integer value with another integer value. It returns the integer
4771with the replaced bits.</p>
4772
4773<h5>Arguments:</h5>
4774<p>The first argument, <tt>%val</tt> and the result may be integer types of
4775any bit width but they must have the same bit width. <tt>%val</tt> is the value
4776whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
4777integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
4778type since they specify only a bit index.</p>
4779
4780<h5>Semantics:</h5>
4781<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
4782of operation: forwards and reverse. If <tt>%lo</tt> is greater than
4783<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
4784operates in forward mode.</p>
4785<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
4786truncating it down to the size of the replacement area or zero extending it
4787up to that size.</p>
4788<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
4789are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
4790in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
4791to the <tt>%hi</tt>th bit.
Reid Spencerc6749c42007-05-14 16:50:20 +0000[diff] [blame]4792<p>In reverse mode, a similar computation is made except that the bits are
4793reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
4794<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]4795<h5>Examples:</h5>
4796<pre>
Reid Spencerf0dbf642007-04-12 01:03:03 +0000[diff] [blame]4797 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
Reid Spencerc6749c42007-05-14 16:50:20 +0000[diff] [blame]4798 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
4799 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
4800 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
Reid Spencerf0dbf642007-04-12 01:03:03 +0000[diff] [blame]4801 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
Reid Spencerc8910842007-04-11 23:49:50 +0000[diff] [blame]4802</pre>
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]4803</div>
4804
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]4805<!-- ======================================================================= -->
4806<div class="doc_subsection">
4807 <a name="int_debugger">Debugger Intrinsics</a>
4808</div>
4809
4810<div class="doc_text">
4811<p>
4812The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
4813are described in the <a
4814href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
4815Debugging</a> document.
4816</p>
4817</div>
4818
4819
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]4820<!-- ======================================================================= -->
4821<div class="doc_subsection">
4822 <a name="int_eh">Exception Handling Intrinsics</a>
4823</div>
4824
4825<div class="doc_text">
4826<p> The LLVM exception handling intrinsics (which all start with
4827<tt>llvm.eh.</tt> prefix), are described in the <a
4828href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
4829Handling</a> document. </p>
4830</div>
4831
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]4832<!-- ======================================================================= -->
4833<div class="doc_subsection">
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4834 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
4835</div>
4836
4837<div class="doc_text">
4838<p>
4839 These intrinsic functions expand the "universal IR" of LLVM to represent
4840 hardware constructs for atomic operations and memory synchronization. This
4841 provides an interface to the hardware, not an interface to the programmer. It
4842 is aimed at a low enough level to allow any programming models or APIs which
4843 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
4844 hardware behavior. Just as hardware provides a "unviresal IR" for source
4845 languages, it also provides a starting point for developing a "universal"
4846 atomic operation and synchronization IR.
4847</p>
4848<p>
4849 These do <em>not</em> form an API such as high-level threading libraries,
4850 software transaction memory systems, atomic primitives, and intrinsic
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]4851 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4852 application libraries. The hardware interface provided by LLVM should allow
4853 a clean implementation of all of these APIs and parallel programming models.
4854 No one model or paradigm should be selected above others unless the hardware
4855 itself ubiquitously does so.
4856</p>
4857</div>
4858
4859<!-- _______________________________________________________________________ -->
4860<div class="doc_subsubsection">
4861 <a name="int_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a>
4862</div>
4863<div class="doc_text">
4864<h5>Syntax:</h5>
4865<p>
4866 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lcs</tt> on any
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]4867 integer bit width. Not all targets support all bit widths however.</p>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4868<pre>
4869declare i8 @llvm.atomic.lcs.i8.i8p.i8.i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
4870declare i16 @llvm.atomic.lcs.i16.i16p.i16.i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
4871declare i32 @llvm.atomic.lcs.i32.i32p.i32.i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
4872declare i64 @llvm.atomic.lcs.i64.i64p.i64.i64( i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt; )
4873</pre>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4874<h5>Overview:</h5>
4875<p>
4876 This loads a value in shared memory and compares it to a given value. If they
4877 are equal, it stores a new value into the shared memory.
4878</p>
4879<h5>Arguments:</h5>
4880<p>
4881 The <tt>llvm.atomic.lcs</tt> intrinsic takes three arguments. The result as
4882 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
4883 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
4884 this integer type. While any bit width integer may be used, targets may only
4885 lower representations they support in hardware.
4886</p>
4887<h5>Semantics:</h5>
4888<p>
4889 This entire intrinsic must be executed atomically. It first loads the value
4890 in shared memory pointed to by <tt>ptr</tt> and compares it with the value
4891 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the shared
4892 memory. The loaded value is yielded in all cases. This provides the
4893 equivalent of an atomic compare-and-swap operation within the SSA framework.
4894</p>
4895<h5>Examples:</h5>
4896<pre>
4897%ptr = malloc i32
4898 store i32 4, %ptr
4899
4900%val1 = add i32 4, 4
4901%result1 = call i32 @llvm.atomic.lcs( i32* %ptr, i32 4, %val1 )
4902 <i>; yields {i32}:result1 = 4</i>
4903%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
4904%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
4905
4906%val2 = add i32 1, 1
4907%result2 = call i32 @llvm.atomic.lcs( i32* %ptr, i32 5, %val2 )
4908 <i>; yields {i32}:result2 = 8</i>
4909%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
4910%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
4911</pre>
4912</div>
4913
4914<!-- _______________________________________________________________________ -->
4915<div class="doc_subsubsection">
4916 <a name="int_ls">'<tt>llvm.atomic.ls.*</tt>' Intrinsic</a>
4917</div>
4918<div class="doc_text">
4919<h5>Syntax:</h5>
4920<p>
4921 This is an overloaded intrinsic. You can use <tt>llvm.atomic.ls</tt> on any
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]4922 integer bit width. Not all targets support all bit widths however.</p>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4923<pre>
4924declare i8 @llvm.atomic.ls.i8.i8p.i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
4925declare i16 @llvm.atomic.ls.i16.i16p.i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
4926declare i32 @llvm.atomic.ls.i32.i32p.i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
4927declare i64 @llvm.atomic.ls.i64.i64p.i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
4928</pre>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4929<h5>Overview:</h5>
4930<p>
4931 This intrinsic loads the value stored in shared memory at <tt>ptr</tt> and
4932 yields the value from memory. It then stores the value in <tt>val</tt> in the
4933 shared memory at <tt>ptr</tt>.
4934</p>
4935<h5>Arguments:</h5>
4936<p>
4937 The <tt>llvm.atomic.ls</tt> intrinsic takes two arguments. Both the
4938 <tt>val</tt> argument and the result must be integers of the same bit width.
4939 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
4940 integer type. The targets may only lower integer representations they
4941 support.
4942</p>
4943<h5>Semantics:</h5>
4944<p>
4945 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
4946 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
4947 equivalent of an atomic swap operation within the SSA framework.
4948</p>
4949<h5>Examples:</h5>
4950<pre>
4951%ptr = malloc i32
4952 store i32 4, %ptr
4953
4954%val1 = add i32 4, 4
4955%result1 = call i32 @llvm.atomic.ls( i32* %ptr, i32 %val1 )
4956 <i>; yields {i32}:result1 = 4</i>
4957%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
4958%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
4959
4960%val2 = add i32 1, 1
4961%result2 = call i32 @llvm.atomic.ls( i32* %ptr, i32 %val2 )
4962 <i>; yields {i32}:result2 = 8</i>
4963%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
4964%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
4965</pre>
4966 </div>
4967
4968<!-- _______________________________________________________________________ -->
4969<div class="doc_subsubsection">
4970 <a name="int_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a>
4971</div>
4972<div class="doc_text">
4973<h5>Syntax:</h5>
4974<p>
4975 This is an overloaded intrinsic. You can use <tt>llvm.atomic.las</tt> on any
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]4976 integer bit width. Not all targets support all bit widths however.</p>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4977<pre>
4978declare i8 @llvm.atomic.las.i8.i8p.i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
4979declare i16 @llvm.atomic.las.i16.i16p.i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
4980declare i32 @llvm.atomic.las.i32.i32p.i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
4981declare i64 @llvm.atomic.las.i64.i64p.i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
4982</pre>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]4983<h5>Overview:</h5>
4984<p>
4985 This intrinsic adds <tt>delta</tt> to the value stored in shared memory at
4986 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
4987</p>
4988<h5>Arguments:</h5>
4989<p>
4990 The intrinsic takes two arguments, the first a pointer to an integer value
4991 and the second an integer value. The result is also an integer value. These
4992 integer types can have any bit width, but they must all have the same bit
4993 width. The targets may only lower integer representations they support.
4994</p>
4995<h5>Semantics:</h5>
4996<p>
4997 This intrinsic does a series of operations atomically. It first loads the
4998 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
4999 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
5000</p>
5001<h5>Examples:</h5>
5002<pre>
5003%ptr = malloc i32
5004 store i32 4, %ptr
5005%result1 = call i32 @llvm.atomic.las( i32* %ptr, i32 4 )
5006 <i>; yields {i32}:result1 = 4</i>
5007%result2 = call i32 @llvm.atomic.las( i32* %ptr, i32 2 )
5008 <i>; yields {i32}:result2 = 8</i>
5009%result3 = call i32 @llvm.atomic.las( i32* %ptr, i32 5 )
5010 <i>; yields {i32}:result3 = 10</i>
5011%memval = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
5012</pre>
5013</div>
5014
5015<!-- _______________________________________________________________________ -->
5016<div class="doc_subsubsection">
5017 <a name="int_lss">'<tt>llvm.atomic.lss.*</tt>' Intrinsic</a>
5018</div>
5019<div class="doc_text">
5020<h5>Syntax:</h5>
5021<p>
5022 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lss</tt> on any
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame^]5023 integer bit width. Not all targets support all bit widths however.</p>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]5024<pre>
5025declare i8 @llvm.atomic.lss.i8.i8.i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
5026declare i16 @llvm.atomic.lss.i16.i16.i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
5027declare i32 @llvm.atomic.lss.i32.i32.i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
5028declare i64 @llvm.atomic.lss.i64.i64.i64( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
5029</pre>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]5030<h5>Overview:</h5>
5031<p>
5032 This intrinsic subtracts <tt>delta</tt> from the value stored in shared
5033 memory at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5034</p>
5035<h5>Arguments:</h5>
5036<p>
5037 The intrinsic takes two arguments, the first a pointer to an integer value
5038 and the second an integer value. The result is also an integer value. These
5039 integer types can have any bit width, but they must all have the same bit
5040 width. The targets may only lower integer representations they support.
5041</p>
5042<h5>Semantics:</h5>
5043<p>
5044 This intrinsic does a series of operations atomically. It first loads the
5045 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>,
5046 stores the result to <tt>ptr</tt>. It yields the original value stored
5047 at <tt>ptr</tt>.
5048</p>
5049<h5>Examples:</h5>
5050<pre>
5051%ptr = malloc i32
5052 store i32 32, %ptr
5053%result1 = call i32 @llvm.atomic.lss( i32* %ptr, i32 4 )
5054 <i>; yields {i32}:result1 = 32</i>
5055%result2 = call i32 @llvm.atomic.lss( i32* %ptr, i32 2 )
5056 <i>; yields {i32}:result2 = 28</i>
5057%result3 = call i32 @llvm.atomic.lss( i32* %ptr, i32 5 )
5058 <i>; yields {i32}:result3 = 26</i>
5059%memval = load i32* %ptr <i>; yields {i32}:memval1 = 21</i>
5060</pre>
5061</div>
5062
5063<!-- _______________________________________________________________________ -->
5064<div class="doc_subsubsection">
5065 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5066</div>
5067<div class="doc_text">
5068<h5>Syntax:</h5>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]5069<pre>
5070declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt; )
5071</pre>
Chandler Carruth2eb93b32007-07-20 19:34:37 +0000[diff] [blame]5072<h5>Overview:</h5>
5073<p>
5074 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5075 specific pairs of memory access types.
5076</p>
5077<h5>Arguments:</h5>
5078<p>
5079 The <tt>llvm.memory.barrier</tt> intrinsic requires four boolean arguments.
5080 Each argument enables a specific barrier as listed below.
5081 <ul>
5082 <li><tt>ll</tt>: load-load barrier</li>
5083 <li><tt>ls</tt>: load-store barrier</li>
5084 <li><tt>sl</tt>: store-load barrier</li>
5085 <li><tt>ss</tt>: store-store barrier</li>
5086 </ul>
5087</p>
5088<h5>Semantics:</h5>
5089<p>
5090 This intrinsic causes the system to enforce some ordering constraints upon
5091 the loads and stores of the program. This barrier does not indicate
5092 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5093 which they occur. For any of the specified pairs of load and store operations
5094 (f.ex. load-load, or store-load), all of the first operations preceding the
5095 barrier will complete before any of the second operations succeeding the
5096 barrier begin. Specifically the semantics for each pairing is as follows:
5097 <ul>
5098 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5099 after the barrier begins.</li>
5100 <li><tt>ls</tt>: All loads before the barrier must complete before any
5101 store after the barrier begins.</li>
5102 <li><tt>ss</tt>: All stores before the barrier must complete before any
5103 store after the barrier begins.</li>
5104 <li><tt>sl</tt>: All stores before the barrier must complete before any
5105 load after the barrier begins.</li>
5106 </ul>
5107 These semantics are applied with a logical "and" behavior when more than one
5108 is enabled in a single memory barrier intrinsic.
5109</p>
5110<h5>Example:</h5>
5111<pre>
5112%ptr = malloc i32
5113 store i32 4, %ptr
5114
5115%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5116 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5117 <i>; guarantee the above finishes</i>
5118 store i32 8, %ptr <i>; before this begins</i>
5119</pre>
5120</div>
5121
5122<!-- ======================================================================= -->
5123<div class="doc_subsection">
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]5124 <a name="int_general">General Intrinsics</a>
5125</div>
5126
5127<div class="doc_text">
5128<p> This class of intrinsics is designed to be generic and has
5129no specific purpose. </p>
5130</div>
5131
5132<!-- _______________________________________________________________________ -->
5133<div class="doc_subsubsection">
5134 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5135</div>
5136
5137<div class="doc_text">
5138
5139<h5>Syntax:</h5>
5140<pre>
Tanya Lattnerd2e84422007-06-18 23:42:37 +0000[diff] [blame]5141 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]5142</pre>
5143
5144<h5>Overview:</h5>
5145
5146<p>
5147The '<tt>llvm.var.annotation</tt>' intrinsic
5148</p>
5149
5150<h5>Arguments:</h5>
5151
5152<p>
Tanya Lattnerd2e84422007-06-18 23:42:37 +0000[diff] [blame]5153The first argument is a pointer to a value, the second is a pointer to a
5154global string, the third is a pointer to a global string which is the source
5155file name, and the last argument is the line number.
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]5156</p>
5157
5158<h5>Semantics:</h5>
5159
5160<p>
5161This intrinsic allows annotation of local variables with arbitrary strings.
5162This can be useful for special purpose optimizations that want to look for these
5163 annotations. These have no other defined use, they are ignored by code
5164 generation and optimization.
5165</div>
5166
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]5167
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5168<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5169<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]5170<address>
5171 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
5172 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
5173 <a href="http://validator.w3.org/check/referer"><img
5174 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
5175
5176 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +0000[diff] [blame]5177 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]5178 Last modified: $Date$
5179</address>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5180</body>
5181</html>