blob: 0eee86e8c0d18ea7453d0c911483b15b7b0a85cf [file] [log] [blame]
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +00003<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencer3921c742004-08-26 20:44:00 +00006 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
Eric Christopher6c7e8a02009-12-05 02:46:03 +00008 <meta name="description"
Reid Spencer3921c742004-08-26 20:44:00 +00009 content="LLVM Assembly Language Reference Manual.">
Misha Brukman9d0919f2003-11-08 01:05:38 +000010 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattnerd7923912004-05-23 21:06:01 +000012
Misha Brukman9d0919f2003-11-08 01:05:38 +000013<body>
Chris Lattnerd7923912004-05-23 21:06:01 +000014
Chris Lattner261efe92003-11-25 01:02:51 +000015<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner00950542001-06-06 20:29:01 +000016<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +000017 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000020 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Bill Wendling3d10a5a2009-07-20 01:03:30 +000023 <li><a href="#linkage">Linkage Types</a>
24 <ol>
Bill Wendling987e7eb2009-07-20 02:41:50 +000025 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
Bill Wendling5e721d72010-07-01 21:55:59 +000027 <li><a href="#linkage_linker_private_weak">'<tt>linker_private_weak</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +000028 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
29 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
30 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
31 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
32 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
33 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
34 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner5a2d8752009-10-10 18:26:06 +000035 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +000036 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
37 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendling3d10a5a2009-07-20 01:03:30 +000040 </ol>
41 </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +000042 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnere7886e42009-01-11 20:53:49 +000043 <li><a href="#namedtypes">Named Types</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000044 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +000045 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +000046 <li><a href="#aliasstructure">Aliases</a></li>
Devang Patelcd1fd252010-01-11 19:35:55 +000047 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
Reid Spencerca86e162006-12-31 07:07:53 +000048 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel2c9c3e72008-09-26 23:51:19 +000049 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +000050 <li><a href="#gc">Garbage Collector Names</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +000051 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencerde151942007-02-19 23:54:10 +000052 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman556ca272009-07-27 18:07:55 +000053 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +000054 <li><a href="#volatile">Volatile Memory Accesses</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000055 </ol>
56 </li>
Chris Lattner00950542001-06-06 20:29:01 +000057 <li><a href="#typesystem">Type System</a>
58 <ol>
Chris Lattner4f69f462008-01-04 04:32:38 +000059 <li><a href="#t_classifications">Type Classifications</a></li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +000060 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner261efe92003-11-25 01:02:51 +000061 <ol>
Nick Lewyckyec38da42009-09-27 00:45:11 +000062 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner4f69f462008-01-04 04:32:38 +000063 <li><a href="#t_floating">Floating Point Types</a></li>
64 <li><a href="#t_void">Void Type</a></li>
65 <li><a href="#t_label">Label Type</a></li>
Nick Lewycky7a0370f2009-05-30 05:06:04 +000066 <li><a href="#t_metadata">Metadata Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000067 </ol>
68 </li>
Chris Lattner00950542001-06-06 20:29:01 +000069 <li><a href="#t_derived">Derived Types</a>
70 <ol>
Chris Lattnerfdfeb692010-02-12 20:49:41 +000071 <li><a href="#t_aggregate">Aggregate Types</a>
72 <ol>
73 <li><a href="#t_array">Array Type</a></li>
74 <li><a href="#t_struct">Structure Type</a></li>
75 <li><a href="#t_pstruct">Packed Structure Type</a></li>
76 <li><a href="#t_union">Union Type</a></li>
77 <li><a href="#t_vector">Vector Type</a></li>
78 </ol>
79 </li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000080 <li><a href="#t_function">Function Type</a></li>
81 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner69c11bb2005-04-25 17:34:15 +000082 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000083 </ol>
84 </li>
Chris Lattner242d61d2009-02-02 07:32:36 +000085 <li><a href="#t_uprefs">Type Up-references</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000086 </ol>
87 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +000088 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +000089 <ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +000090 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner70882792009-02-28 18:32:25 +000091 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +000092 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
93 <li><a href="#undefvalues">Undefined Values</a></li>
Dan Gohmanfff6c532010-04-22 23:14:21 +000094 <li><a href="#trapvalues">Trap Values</a></li>
Chris Lattnerf9d078e2009-10-27 21:19:13 +000095 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +000096 <li><a href="#constantexprs">Constant Expressions</a></li>
Chris Lattnerc3f59762004-12-09 17:30:23 +000097 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000098 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +000099 <li><a href="#othervalues">Other Values</a>
100 <ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000101 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Devang Patelcd1fd252010-01-11 19:35:55 +0000102 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000103 </ol>
104 </li>
Chris Lattner857755c2009-07-20 05:55:19 +0000105 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
106 <ol>
107 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner401e10c2009-07-20 06:14:25 +0000108 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
109 Global Variable</a></li>
Chris Lattner857755c2009-07-20 05:55:19 +0000110 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
111 Global Variable</a></li>
112 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
113 Global Variable</a></li>
114 </ol>
115 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000116 <li><a href="#instref">Instruction Reference</a>
117 <ol>
118 <li><a href="#terminators">Terminator Instructions</a>
119 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000120 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
121 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000122 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerab21db72009-10-28 00:19:10 +0000123 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000124 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000125 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000126 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000127 </ol>
128 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000129 <li><a href="#binaryops">Binary Operations</a>
130 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000131 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000132 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000133 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000134 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000135 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000136 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000137 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
138 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
139 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000140 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
141 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
142 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000143 </ol>
144 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000145 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
146 <ol>
Reid Spencer8e11bf82007-02-02 13:57:07 +0000147 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
148 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
149 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000150 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000151 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000152 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000153 </ol>
154 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000155 <li><a href="#vectorops">Vector Operations</a>
156 <ol>
157 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
158 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
159 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000160 </ol>
161 </li>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000162 <li><a href="#aggregateops">Aggregate Operations</a>
163 <ol>
164 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
165 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
166 </ol>
167 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000168 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000169 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000170 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000171 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
172 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
173 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000174 </ol>
175 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000176 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000177 <ol>
178 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
179 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
180 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
181 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
182 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000183 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
184 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
185 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
186 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000187 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
188 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000189 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000190 </ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000191 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000192 <li><a href="#otherops">Other Operations</a>
193 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000194 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
195 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000196 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000197 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000198 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000199 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000200 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000201 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000202 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000203 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000204 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000205 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000206 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
207 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000208 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
209 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
210 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000211 </ol>
212 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000213 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
214 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000215 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
216 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
217 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000218 </ol>
219 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000220 <li><a href="#int_codegen">Code Generator Intrinsics</a>
221 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000222 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
223 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
224 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
225 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
226 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
227 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohman31f1af12010-05-26 21:56:15 +0000228 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000229 </ol>
230 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000231 <li><a href="#int_libc">Standard C Library Intrinsics</a>
232 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000233 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
234 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
236 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
237 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman91c284c2007-10-15 20:30:11 +0000238 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
239 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000241 </ol>
242 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000243 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000244 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000245 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000246 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
247 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
248 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000249 </ol>
250 </li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000251 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
252 <ol>
Bill Wendlingda01af72009-02-08 04:04:40 +0000253 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
254 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
255 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
256 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
257 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendling41b485c2009-02-08 23:00:09 +0000258 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000259 </ol>
260 </li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000261 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
262 <ol>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000263 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
264 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000265 </ol>
266 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000267 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000268 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sandsf7331b32007-09-11 14:10:23 +0000269 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands36397f52007-07-27 12:58:54 +0000270 <ol>
271 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands36397f52007-07-27 12:58:54 +0000272 </ol>
273 </li>
Bill Wendling3c44f5b2008-11-18 22:10:53 +0000274 <li><a href="#int_atomics">Atomic intrinsics</a>
275 <ol>
276 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
277 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
278 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
279 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
280 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
281 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
282 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
283 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
284 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
285 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
286 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
287 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
288 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
289 </ol>
290 </li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000291 <li><a href="#int_memorymarkers">Memory Use Markers</a>
292 <ol>
293 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
294 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
295 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
296 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
297 </ol>
298 </li>
Reid Spencer20677642007-07-20 19:59:11 +0000299 <li><a href="#int_general">General intrinsics</a>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000300 <ol>
Reid Spencer20677642007-07-20 19:59:11 +0000301 <li><a href="#int_var_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000302 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000303 <li><a href="#int_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000304 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000305 <li><a href="#int_trap">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000306 '<tt>llvm.trap</tt>' Intrinsic</a></li>
307 <li><a href="#int_stackprotector">
308 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher0e671492009-11-30 08:03:53 +0000309 <li><a href="#int_objectsize">
310 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000311 </ol>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000312 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000313 </ol>
314 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000315</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000316
317<div class="doc_author">
318 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
319 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000320</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000321
Chris Lattner00950542001-06-06 20:29:01 +0000322<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000323<div class="doc_section"> <a name="abstract">Abstract </a></div>
324<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000325
Misha Brukman9d0919f2003-11-08 01:05:38 +0000326<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000327
328<p>This document is a reference manual for the LLVM assembly language. LLVM is
329 a Static Single Assignment (SSA) based representation that provides type
330 safety, low-level operations, flexibility, and the capability of representing
331 'all' high-level languages cleanly. It is the common code representation
332 used throughout all phases of the LLVM compilation strategy.</p>
333
Misha Brukman9d0919f2003-11-08 01:05:38 +0000334</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000335
Chris Lattner00950542001-06-06 20:29:01 +0000336<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000337<div class="doc_section"> <a name="introduction">Introduction</a> </div>
338<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000339
Misha Brukman9d0919f2003-11-08 01:05:38 +0000340<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000341
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000342<p>The LLVM code representation is designed to be used in three different forms:
343 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
344 for fast loading by a Just-In-Time compiler), and as a human readable
345 assembly language representation. This allows LLVM to provide a powerful
346 intermediate representation for efficient compiler transformations and
347 analysis, while providing a natural means to debug and visualize the
348 transformations. The three different forms of LLVM are all equivalent. This
349 document describes the human readable representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000350
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000351<p>The LLVM representation aims to be light-weight and low-level while being
352 expressive, typed, and extensible at the same time. It aims to be a
353 "universal IR" of sorts, by being at a low enough level that high-level ideas
354 may be cleanly mapped to it (similar to how microprocessors are "universal
355 IR's", allowing many source languages to be mapped to them). By providing
356 type information, LLVM can be used as the target of optimizations: for
357 example, through pointer analysis, it can be proven that a C automatic
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000358 variable is never accessed outside of the current function, allowing it to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000359 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000360
Misha Brukman9d0919f2003-11-08 01:05:38 +0000361</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000362
Chris Lattner00950542001-06-06 20:29:01 +0000363<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000364<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000365
Misha Brukman9d0919f2003-11-08 01:05:38 +0000366<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000367
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000368<p>It is important to note that this document describes 'well formed' LLVM
369 assembly language. There is a difference between what the parser accepts and
370 what is considered 'well formed'. For example, the following instruction is
371 syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000372
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000373<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000374%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000375</pre>
376
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000377<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
378 LLVM infrastructure provides a verification pass that may be used to verify
379 that an LLVM module is well formed. This pass is automatically run by the
380 parser after parsing input assembly and by the optimizer before it outputs
381 bitcode. The violations pointed out by the verifier pass indicate bugs in
382 transformation passes or input to the parser.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000383
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000384</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000385
Chris Lattnercc689392007-10-03 17:34:29 +0000386<!-- Describe the typesetting conventions here. -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000387
Chris Lattner00950542001-06-06 20:29:01 +0000388<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000389<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000390<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000391
Misha Brukman9d0919f2003-11-08 01:05:38 +0000392<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000393
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000394<p>LLVM identifiers come in two basic types: global and local. Global
395 identifiers (functions, global variables) begin with the <tt>'@'</tt>
396 character. Local identifiers (register names, types) begin with
397 the <tt>'%'</tt> character. Additionally, there are three different formats
398 for identifiers, for different purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000399
Chris Lattner00950542001-06-06 20:29:01 +0000400<ol>
Reid Spencer2c452282007-08-07 14:34:28 +0000401 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000402 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
403 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
404 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
405 other characters in their names can be surrounded with quotes. Special
406 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
407 ASCII code for the character in hexadecimal. In this way, any character
408 can be used in a name value, even quotes themselves.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000409
Reid Spencer2c452282007-08-07 14:34:28 +0000410 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000411 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000412
Reid Spencercc16dc32004-12-09 18:02:53 +0000413 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000414 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000415</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000416
Reid Spencer2c452282007-08-07 14:34:28 +0000417<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000418 don't need to worry about name clashes with reserved words, and the set of
419 reserved words may be expanded in the future without penalty. Additionally,
420 unnamed identifiers allow a compiler to quickly come up with a temporary
421 variable without having to avoid symbol table conflicts.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000422
Chris Lattner261efe92003-11-25 01:02:51 +0000423<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000424 languages. There are keywords for different opcodes
425 ('<tt><a href="#i_add">add</a></tt>',
426 '<tt><a href="#i_bitcast">bitcast</a></tt>',
427 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
428 ('<tt><a href="#t_void">void</a></tt>',
429 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
430 reserved words cannot conflict with variable names, because none of them
431 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000432
433<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000434 '<tt>%X</tt>' by 8:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000435
Misha Brukman9d0919f2003-11-08 01:05:38 +0000436<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000437
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000438<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000439%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000440</pre>
441
Misha Brukman9d0919f2003-11-08 01:05:38 +0000442<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000443
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000444<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000445%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000446</pre>
447
Misha Brukman9d0919f2003-11-08 01:05:38 +0000448<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000449
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000450<pre class="doc_code">
Gabor Greifec58f752009-10-28 13:05:07 +0000451%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
452%1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000453%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000454</pre>
455
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000456<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
457 lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000458
Chris Lattner00950542001-06-06 20:29:01 +0000459<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000460 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000461 line.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000462
463 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000464 assigned to a named value.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000465
Misha Brukman9d0919f2003-11-08 01:05:38 +0000466 <li>Unnamed temporaries are numbered sequentially</li>
467</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000468
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000469<p>It also shows a convention that we follow in this document. When
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000470 demonstrating instructions, we will follow an instruction with a comment that
471 defines the type and name of value produced. Comments are shown in italic
472 text.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000473
Misha Brukman9d0919f2003-11-08 01:05:38 +0000474</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000475
476<!-- *********************************************************************** -->
477<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
478<!-- *********************************************************************** -->
479
480<!-- ======================================================================= -->
481<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
482</div>
483
484<div class="doc_text">
485
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000486<p>LLVM programs are composed of "Module"s, each of which is a translation unit
487 of the input programs. Each module consists of functions, global variables,
488 and symbol table entries. Modules may be combined together with the LLVM
489 linker, which merges function (and global variable) definitions, resolves
490 forward declarations, and merges symbol table entries. Here is an example of
491 the "hello world" module:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000492
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000493<pre class="doc_code">
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000494<i>; Declare the string constant as a global constant.</i>
495<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a 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 +0000496
497<i>; External declaration of the puts function</i>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000498<a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000499
500<i>; Definition of main function</i>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000501define i32 @main() { <i>; i32()* </i>
502 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000503 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
Chris Lattnerfa730212004-12-09 16:11:40 +0000504
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000505 <i>; Call puts function to write out the string to stdout.</i>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000506 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
Devang Patelcd1fd252010-01-11 19:35:55 +0000507 <a href="#i_ret">ret</a> i32 0<br>}
508
509<i>; Named metadata</i>
510!1 = metadata !{i32 41}
511!foo = !{!1, null}
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000512</pre>
Chris Lattnerfa730212004-12-09 16:11:40 +0000513
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000514<p>This example is made up of a <a href="#globalvars">global variable</a> named
Devang Patelcd1fd252010-01-11 19:35:55 +0000515 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000516 a <a href="#functionstructure">function definition</a> for
Devang Patelcd1fd252010-01-11 19:35:55 +0000517 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
518 "<tt>foo"</tt>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000519
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000520<p>In general, a module is made up of a list of global values, where both
521 functions and global variables are global values. Global values are
522 represented by a pointer to a memory location (in this case, a pointer to an
523 array of char, and a pointer to a function), and have one of the
524 following <a href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000525
Chris Lattnere5d947b2004-12-09 16:36:40 +0000526</div>
527
528<!-- ======================================================================= -->
529<div class="doc_subsection">
530 <a name="linkage">Linkage Types</a>
531</div>
532
533<div class="doc_text">
534
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000535<p>All Global Variables and Functions have one of the following types of
536 linkage:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000537
538<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000539 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000540 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
541 by objects in the current module. In particular, linking code into a
542 module with an private global value may cause the private to be renamed as
543 necessary to avoid collisions. Because the symbol is private to the
544 module, all references can be updated. This doesn't show up in any symbol
545 table in the object file.</dd>
Rafael Espindolabb46f522009-01-15 20:18:42 +0000546
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000547 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000548 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
549 assembler and evaluated by the linker. Unlike normal strong symbols, they
550 are removed by the linker from the final linked image (executable or
551 dynamic library).</dd>
552
553 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
554 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
555 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
556 linker. The symbols are removed by the linker from the final linked image
557 (executable or dynamic library).</dd>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000558
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000559 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling07d31772010-06-29 22:34:52 +0000560 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000561 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
562 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000563
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000564 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000565 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000566 into the object file corresponding to the LLVM module. They exist to
567 allow inlining and other optimizations to take place given knowledge of
568 the definition of the global, which is known to be somewhere outside the
569 module. Globals with <tt>available_externally</tt> linkage are allowed to
570 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
571 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000572
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000573 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattner4887bd82007-01-14 06:51:48 +0000574 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner873187c2010-01-09 19:15:14 +0000575 the same name when linkage occurs. This can be used to implement
576 some forms of inline functions, templates, or other code which must be
577 generated in each translation unit that uses it, but where the body may
578 be overridden with a more definitive definition later. Unreferenced
579 <tt>linkonce</tt> globals are allowed to be discarded. Note that
580 <tt>linkonce</tt> linkage does not actually allow the optimizer to
581 inline the body of this function into callers because it doesn't know if
582 this definition of the function is the definitive definition within the
583 program or whether it will be overridden by a stronger definition.
584 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
585 linkage.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000586
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000587 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000588 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
589 <tt>linkonce</tt> linkage, except that unreferenced globals with
590 <tt>weak</tt> linkage may not be discarded. This is used for globals that
591 are declared "weak" in C source code.</dd>
592
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000593 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000594 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
595 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
596 global scope.
597 Symbols with "<tt>common</tt>" linkage are merged in the same way as
598 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000599 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000600 must have a zero initializer, and may not be marked '<a
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000601 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
602 have common linkage.</dd>
Chris Lattner26d054d2009-08-05 05:21:07 +0000603
Chris Lattnere5d947b2004-12-09 16:36:40 +0000604
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000605 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000606 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000607 pointer to array type. When two global variables with appending linkage
608 are linked together, the two global arrays are appended together. This is
609 the LLVM, typesafe, equivalent of having the system linker append together
610 "sections" with identical names when .o files are linked.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000611
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000612 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000613 <dd>The semantics of this linkage follow the ELF object file model: the symbol
614 is weak until linked, if not linked, the symbol becomes null instead of
615 being an undefined reference.</dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000616
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000617 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
618 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000619 <dd>Some languages allow differing globals to be merged, such as two functions
620 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling5e721d72010-07-01 21:55:59 +0000621 that only equivalent globals are ever merged (the "one definition rule"
622 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000623 and <tt>weak_odr</tt> linkage types to indicate that the global will only
624 be merged with equivalent globals. These linkage types are otherwise the
625 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands667d4b82009-03-07 15:45:40 +0000626
Chris Lattnerfa730212004-12-09 16:11:40 +0000627 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000628 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000629 visible, meaning that it participates in linkage and can be used to
630 resolve external symbol references.</dd>
Reid Spencerc8910842007-04-11 23:49:50 +0000631</dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000632
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000633<p>The next two types of linkage are targeted for Microsoft Windows platform
634 only. They are designed to support importing (exporting) symbols from (to)
635 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000636
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000637<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000638 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000639 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000640 or variable via a global pointer to a pointer that is set up by the DLL
641 exporting the symbol. On Microsoft Windows targets, the pointer name is
642 formed by combining <code>__imp_</code> and the function or variable
643 name.</dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000644
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000645 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000646 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000647 pointer to a pointer in a DLL, so that it can be referenced with the
648 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
649 name is formed by combining <code>__imp_</code> and the function or
650 variable name.</dd>
Chris Lattnerfa730212004-12-09 16:11:40 +0000651</dl>
652
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000653<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
654 another module defined a "<tt>.LC0</tt>" variable and was linked with this
655 one, one of the two would be renamed, preventing a collision. Since
656 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
657 declarations), they are accessible outside of the current module.</p>
658
659<p>It is illegal for a function <i>declaration</i> to have any linkage type
660 other than "externally visible", <tt>dllimport</tt>
661 or <tt>extern_weak</tt>.</p>
662
Duncan Sands667d4b82009-03-07 15:45:40 +0000663<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000664 or <tt>weak_odr</tt> linkages.</p>
665
Chris Lattnerfa730212004-12-09 16:11:40 +0000666</div>
667
668<!-- ======================================================================= -->
669<div class="doc_subsection">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000670 <a name="callingconv">Calling Conventions</a>
671</div>
672
673<div class="doc_text">
674
675<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000676 and <a href="#i_invoke">invokes</a> can all have an optional calling
677 convention specified for the call. The calling convention of any pair of
678 dynamic caller/callee must match, or the behavior of the program is
679 undefined. The following calling conventions are supported by LLVM, and more
680 may be added in the future:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000681
682<dl>
683 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000684 <dd>This calling convention (the default if no other calling convention is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000685 specified) matches the target C calling conventions. This calling
686 convention supports varargs function calls and tolerates some mismatch in
687 the declared prototype and implemented declaration of the function (as
688 does normal C).</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000689
690 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000691 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000692 (e.g. by passing things in registers). This calling convention allows the
693 target to use whatever tricks it wants to produce fast code for the
694 target, without having to conform to an externally specified ABI
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000695 (Application Binary Interface).
696 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattner29689432010-03-11 00:22:57 +0000697 when this or the GHC convention is used.</a> This calling convention
698 does not support varargs and requires the prototype of all callees to
699 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000700
701 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000702 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000703 as possible under the assumption that the call is not commonly executed.
704 As such, these calls often preserve all registers so that the call does
705 not break any live ranges in the caller side. This calling convention
706 does not support varargs and requires the prototype of all callees to
707 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000708
Chris Lattner29689432010-03-11 00:22:57 +0000709 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
710 <dd>This calling convention has been implemented specifically for use by the
711 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
712 It passes everything in registers, going to extremes to achieve this by
713 disabling callee save registers. This calling convention should not be
714 used lightly but only for specific situations such as an alternative to
715 the <em>register pinning</em> performance technique often used when
716 implementing functional programming languages.At the moment only X86
717 supports this convention and it has the following limitations:
718 <ul>
719 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
720 floating point types are supported.</li>
721 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
722 6 floating point parameters.</li>
723 </ul>
724 This calling convention supports
725 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
726 requires both the caller and callee are using it.
727 </dd>
728
Chris Lattnercfe6b372005-05-07 01:46:40 +0000729 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000730 <dd>Any calling convention may be specified by number, allowing
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000731 target-specific calling conventions to be used. Target specific calling
732 conventions start at 64.</dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000733</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000734
735<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000736 support Pascal conventions or any other well-known target-independent
737 convention.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000738
739</div>
740
741<!-- ======================================================================= -->
742<div class="doc_subsection">
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000743 <a name="visibility">Visibility Styles</a>
744</div>
745
746<div class="doc_text">
747
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000748<p>All Global Variables and Functions have one of the following visibility
749 styles:</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000750
751<dl>
752 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000753 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000754 that the declaration is visible to other modules and, in shared libraries,
755 means that the declared entity may be overridden. On Darwin, default
756 visibility means that the declaration is visible to other modules. Default
757 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000758
759 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000760 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000761 object if they are in the same shared object. Usually, hidden visibility
762 indicates that the symbol will not be placed into the dynamic symbol
763 table, so no other module (executable or shared library) can reference it
764 directly.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000765
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000766 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000767 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000768 the dynamic symbol table, but that references within the defining module
769 will bind to the local symbol. That is, the symbol cannot be overridden by
770 another module.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000771</dl>
772
773</div>
774
775<!-- ======================================================================= -->
776<div class="doc_subsection">
Chris Lattnere7886e42009-01-11 20:53:49 +0000777 <a name="namedtypes">Named Types</a>
778</div>
779
780<div class="doc_text">
781
782<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000783 it easier to read the IR and make the IR more condensed (particularly when
784 recursive types are involved). An example of a name specification is:</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000785
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000786<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +0000787%mytype = type { %mytype*, i32 }
788</pre>
Chris Lattnere7886e42009-01-11 20:53:49 +0000789
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000790<p>You may give a name to any <a href="#typesystem">type</a> except
791 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
792 is expected with the syntax "%mytype".</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000793
794<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000795 and that you can therefore specify multiple names for the same type. This
796 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
797 uses structural typing, the name is not part of the type. When printing out
798 LLVM IR, the printer will pick <em>one name</em> to render all types of a
799 particular shape. This means that if you have code where two different
800 source types end up having the same LLVM type, that the dumper will sometimes
801 print the "wrong" or unexpected type. This is an important design point and
802 isn't going to change.</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000803
804</div>
805
Chris Lattnere7886e42009-01-11 20:53:49 +0000806<!-- ======================================================================= -->
807<div class="doc_subsection">
Chris Lattnerfa730212004-12-09 16:11:40 +0000808 <a name="globalvars">Global Variables</a>
809</div>
810
811<div class="doc_text">
812
Chris Lattner3689a342005-02-12 19:30:21 +0000813<p>Global variables define regions of memory allocated at compilation time
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000814 instead of run-time. Global variables may optionally be initialized, may
815 have an explicit section to be placed in, and may have an optional explicit
816 alignment specified. A variable may be defined as "thread_local", which
817 means that it will not be shared by threads (each thread will have a
818 separated copy of the variable). A variable may be defined as a global
819 "constant," which indicates that the contents of the variable
820 will <b>never</b> be modified (enabling better optimization, allowing the
821 global data to be placed in the read-only section of an executable, etc).
822 Note that variables that need runtime initialization cannot be marked
823 "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000824
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000825<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
826 constant, even if the final definition of the global is not. This capability
827 can be used to enable slightly better optimization of the program, but
828 requires the language definition to guarantee that optimizations based on the
829 'constantness' are valid for the translation units that do not include the
830 definition.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000831
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000832<p>As SSA values, global variables define pointer values that are in scope
833 (i.e. they dominate) all basic blocks in the program. Global variables
834 always define a pointer to their "content" type because they describe a
835 region of memory, and all memory objects in LLVM are accessed through
836 pointers.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000837
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000838<p>A global variable may be declared to reside in a target-specific numbered
839 address space. For targets that support them, address spaces may affect how
840 optimizations are performed and/or what target instructions are used to
841 access the variable. The default address space is zero. The address space
842 qualifier must precede any other attributes.</p>
Christopher Lamb284d9922007-12-11 09:31:00 +0000843
Chris Lattner88f6c462005-11-12 00:45:07 +0000844<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000845 supports it, it will emit globals to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000846
Chris Lattnerce99fa92010-04-28 00:13:42 +0000847<p>An explicit alignment may be specified for a global, which must be a power
848 of 2. If not present, or if the alignment is set to zero, the alignment of
849 the global is set by the target to whatever it feels convenient. If an
850 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner2d4b8ee2010-04-28 00:31:12 +0000851 alignment. Targets and optimizers are not allowed to over-align the global
852 if the global has an assigned section. In this case, the extra alignment
853 could be observable: for example, code could assume that the globals are
854 densely packed in their section and try to iterate over them as an array,
855 alignment padding would break this iteration.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000856
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000857<p>For example, the following defines a global in a numbered address space with
858 an initializer, section, and alignment:</p>
Chris Lattner68027ea2007-01-14 00:27:09 +0000859
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000860<pre class="doc_code">
Dan Gohman398873c2009-01-11 00:40:00 +0000861@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner68027ea2007-01-14 00:27:09 +0000862</pre>
863
Chris Lattnerfa730212004-12-09 16:11:40 +0000864</div>
865
866
867<!-- ======================================================================= -->
868<div class="doc_subsection">
869 <a name="functionstructure">Functions</a>
870</div>
871
872<div class="doc_text">
873
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000874<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000875 optional <a href="#linkage">linkage type</a>, an optional
876 <a href="#visibility">visibility style</a>, an optional
877 <a href="#callingconv">calling convention</a>, a return type, an optional
878 <a href="#paramattrs">parameter attribute</a> for the return type, a function
879 name, a (possibly empty) argument list (each with optional
880 <a href="#paramattrs">parameter attributes</a>), optional
881 <a href="#fnattrs">function attributes</a>, an optional section, an optional
882 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
883 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000884
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000885<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
886 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000887 <a href="#visibility">visibility style</a>, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000888 <a href="#callingconv">calling convention</a>, a return type, an optional
889 <a href="#paramattrs">parameter attribute</a> for the return type, a function
890 name, a possibly empty list of arguments, an optional alignment, and an
891 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000892
Chris Lattnerd3eda892008-08-05 18:29:16 +0000893<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000894 (Control Flow Graph) for the function. Each basic block may optionally start
895 with a label (giving the basic block a symbol table entry), contains a list
896 of instructions, and ends with a <a href="#terminators">terminator</a>
897 instruction (such as a branch or function return).</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000898
Chris Lattner4a3c9012007-06-08 16:52:14 +0000899<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000900 executed on entrance to the function, and it is not allowed to have
901 predecessor basic blocks (i.e. there can not be any branches to the entry
902 block of a function). Because the block can have no predecessors, it also
903 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000904
Chris Lattner88f6c462005-11-12 00:45:07 +0000905<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000906 supports it, it will emit functions to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000907
Chris Lattner2cbdc452005-11-06 08:02:57 +0000908<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000909 the alignment is set to zero, the alignment of the function is set by the
910 target to whatever it feels convenient. If an explicit alignment is
911 specified, the function is forced to have at least that much alignment. All
912 alignments must be a power of 2.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000913
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000914<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000915<pre class="doc_code">
Chris Lattner50ad45c2008-10-13 16:55:18 +0000916define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000917 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
918 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
919 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
920 [<a href="#gc">gc</a>] { ... }
921</pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000922
Chris Lattnerfa730212004-12-09 16:11:40 +0000923</div>
924
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000925<!-- ======================================================================= -->
926<div class="doc_subsection">
927 <a name="aliasstructure">Aliases</a>
928</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000929
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000930<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000931
932<p>Aliases act as "second name" for the aliasee value (which can be either
933 function, global variable, another alias or bitcast of global value). Aliases
934 may have an optional <a href="#linkage">linkage type</a>, and an
935 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000936
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000937<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000938<pre class="doc_code">
Duncan Sands0b23ac12008-09-12 20:48:21 +0000939@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendlingaac388b2007-05-29 09:42:13 +0000940</pre>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000941
942</div>
943
Chris Lattner4e9aba72006-01-23 23:23:47 +0000944<!-- ======================================================================= -->
Devang Patelcd1fd252010-01-11 19:35:55 +0000945<div class="doc_subsection">
946 <a name="namedmetadatastructure">Named Metadata</a>
947</div>
948
949<div class="doc_text">
950
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000951<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
952 nodes</a> (but not metadata strings) and null are the only valid operands for
953 a named metadata.</p>
Devang Patelcd1fd252010-01-11 19:35:55 +0000954
955<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000956<pre class="doc_code">
Devang Patelcd1fd252010-01-11 19:35:55 +0000957!1 = metadata !{metadata !"one"}
958!name = !{null, !1}
959</pre>
Devang Patelcd1fd252010-01-11 19:35:55 +0000960
961</div>
962
963<!-- ======================================================================= -->
Reid Spencerca86e162006-12-31 07:07:53 +0000964<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Reid Spencerca86e162006-12-31 07:07:53 +0000965
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000966<div class="doc_text">
967
968<p>The return type and each parameter of a function type may have a set of
969 <i>parameter attributes</i> associated with them. Parameter attributes are
970 used to communicate additional information about the result or parameters of
971 a function. Parameter attributes are considered to be part of the function,
972 not of the function type, so functions with different parameter attributes
973 can have the same function type.</p>
974
975<p>Parameter attributes are simple keywords that follow the type specified. If
976 multiple parameter attributes are needed, they are space separated. For
977 example:</p>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000978
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000979<pre class="doc_code">
Nick Lewyckyb6a7d252009-02-15 23:06:14 +0000980declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattner66d922c2008-10-04 18:33:34 +0000981declare i32 @atoi(i8 zeroext)
982declare signext i8 @returns_signed_char()
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000983</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000984
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000985<p>Note that any attributes for the function result (<tt>nounwind</tt>,
986 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerca86e162006-12-31 07:07:53 +0000987
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000988<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner47507de2008-01-11 06:20:47 +0000989
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000990<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000991 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000992 <dd>This indicates to the code generator that the parameter or return value
993 should be zero-extended to a 32-bit value by the caller (for a parameter)
994 or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000995
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000996 <dt><tt><b>signext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000997 <dd>This indicates to the code generator that the parameter or return value
998 should be sign-extended to a 32-bit value by the caller (for a parameter)
999 or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +00001000
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001001 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001002 <dd>This indicates that this parameter or return value should be treated in a
1003 special target-dependent fashion during while emitting code for a function
1004 call or return (usually, by putting it in a register as opposed to memory,
1005 though some targets use it to distinguish between two different kinds of
1006 registers). Use of this attribute is target-specific.</dd>
Chris Lattner47507de2008-01-11 06:20:47 +00001007
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001008 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001009 <dd>This indicates that the pointer parameter should really be passed by value
1010 to the function. The attribute implies that a hidden copy of the pointee
1011 is made between the caller and the callee, so the callee is unable to
1012 modify the value in the callee. This attribute is only valid on LLVM
1013 pointer arguments. It is generally used to pass structs and arrays by
1014 value, but is also valid on pointers to scalars. The copy is considered
1015 to belong to the caller not the callee (for example,
1016 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1017 <tt>byval</tt> parameters). This is not a valid attribute for return
1018 values. The byval attribute also supports specifying an alignment with
1019 the align attribute. This has a target-specific effect on the code
1020 generator that usually indicates a desired alignment for the synthesized
1021 stack slot.</dd>
1022
Dan Gohmanff235352010-07-02 23:18:08 +00001023 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001024 <dd>This indicates that the pointer parameter specifies the address of a
1025 structure that is the return value of the function in the source program.
1026 This pointer must be guaranteed by the caller to be valid: loads and
1027 stores to the structure may be assumed by the callee to not to trap. This
1028 may only be applied to the first parameter. This is not a valid attribute
1029 for return values. </dd>
1030
Dan Gohmanff235352010-07-02 23:18:08 +00001031 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohman1e109622010-07-02 18:41:32 +00001032 <dd>This indicates that pointer values
1033 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmanefca7f92010-07-02 23:46:54 +00001034 value do not alias pointer values which are not <i>based</i> on it,
1035 ignoring certain "irrelevant" dependencies.
1036 For a call to the parent function, dependencies between memory
1037 references from before or after the call and from those during the call
1038 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1039 return value used in that call.
Dan Gohman1e109622010-07-02 18:41:32 +00001040 The caller shares the responsibility with the callee for ensuring that
1041 these requirements are met.
1042 For further details, please see the discussion of the NoAlias response in
Dan Gohmanff70fe42010-07-06 15:26:33 +00001043 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1044<br>
John McCall191d4ee2010-07-06 21:07:14 +00001045 Note that this definition of <tt>noalias</tt> is intentionally
1046 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner211244a2010-07-06 20:51:35 +00001047 arguments, though it is slightly weaker.
Dan Gohmanff70fe42010-07-06 15:26:33 +00001048<br>
1049 For function return values, C99's <tt>restrict</tt> is not meaningful,
1050 while LLVM's <tt>noalias</tt> is.
1051 </dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001052
Dan Gohmanff235352010-07-02 23:18:08 +00001053 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001054 <dd>This indicates that the callee does not make any copies of the pointer
1055 that outlive the callee itself. This is not a valid attribute for return
1056 values.</dd>
1057
Dan Gohmanff235352010-07-02 23:18:08 +00001058 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001059 <dd>This indicates that the pointer parameter can be excised using the
1060 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1061 attribute for return values.</dd>
1062</dl>
Reid Spencerca86e162006-12-31 07:07:53 +00001063
Reid Spencerca86e162006-12-31 07:07:53 +00001064</div>
1065
1066<!-- ======================================================================= -->
Chris Lattner4e9aba72006-01-23 23:23:47 +00001067<div class="doc_subsection">
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001068 <a name="gc">Garbage Collector Names</a>
1069</div>
1070
1071<div class="doc_text">
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001072
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001073<p>Each function may specify a garbage collector name, which is simply a
1074 string:</p>
1075
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001076<pre class="doc_code">
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001077define void @f() gc "name" { ... }
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001078</pre>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001079
1080<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001081 collector which will cause the compiler to alter its output in order to
1082 support the named garbage collection algorithm.</p>
1083
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001084</div>
1085
1086<!-- ======================================================================= -->
1087<div class="doc_subsection">
Devang Patel2c9c3e72008-09-26 23:51:19 +00001088 <a name="fnattrs">Function Attributes</a>
Devang Patelf8b94812008-09-04 23:05:13 +00001089</div>
1090
1091<div class="doc_text">
Devang Patel2c9c3e72008-09-26 23:51:19 +00001092
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001093<p>Function attributes are set to communicate additional information about a
1094 function. Function attributes are considered to be part of the function, not
1095 of the function type, so functions with different parameter attributes can
1096 have the same function type.</p>
Devang Patel2c9c3e72008-09-26 23:51:19 +00001097
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001098<p>Function attributes are simple keywords that follow the type specified. If
1099 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelf8b94812008-09-04 23:05:13 +00001100
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001101<pre class="doc_code">
Devang Patel2c9c3e72008-09-26 23:51:19 +00001102define void @f() noinline { ... }
1103define void @f() alwaysinline { ... }
1104define void @f() alwaysinline optsize { ... }
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001105define void @f() optsize { ... }
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001106</pre>
Devang Patelf8b94812008-09-04 23:05:13 +00001107
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001108<dl>
Charles Davis1e063d12010-02-12 00:31:15 +00001109 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1110 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1111 the backend should forcibly align the stack pointer. Specify the
1112 desired alignment, which must be a power of two, in parentheses.
1113
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001114 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001115 <dd>This attribute indicates that the inliner should attempt to inline this
1116 function into callers whenever possible, ignoring any active inlining size
1117 threshold for this caller.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001118
Jakob Stoklund Olesen570a4a52010-02-06 01:16:28 +00001119 <dt><tt><b>inlinehint</b></tt></dt>
1120 <dd>This attribute indicates that the source code contained a hint that inlining
1121 this function is desirable (such as the "inline" keyword in C/C++). It
1122 is just a hint; it imposes no requirements on the inliner.</dd>
1123
Nick Lewycky76ec37a2010-07-06 18:24:09 +00001124 <dt><tt><b>naked</b></tt></dt>
1125 <dd>This attribute disables prologue / epilogue emission for the function.
1126 This can have very system-specific consequences.</dd>
1127
1128 <dt><tt><b>noimplicitfloat</b></tt></dt>
1129 <dd>This attributes disables implicit floating point instructions.</dd>
1130
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001131 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001132 <dd>This attribute indicates that the inliner should never inline this
1133 function in any situation. This attribute may not be used together with
1134 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +00001135
Nick Lewycky76ec37a2010-07-06 18:24:09 +00001136 <dt><tt><b>noredzone</b></tt></dt>
1137 <dd>This attribute indicates that the code generator should not use a red
1138 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +00001139
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001140 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001141 <dd>This function attribute indicates that the function never returns
1142 normally. This produces undefined behavior at runtime if the function
1143 ever does dynamically return.</dd>
Bill Wendling31359ba2008-11-13 01:02:51 +00001144
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001145 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001146 <dd>This function attribute indicates that the function never returns with an
1147 unwind or exceptional control flow. If the function does unwind, its
1148 runtime behavior is undefined.</dd>
Bill Wendlingfbaa7ed2008-11-26 19:07:40 +00001149
Nick Lewycky76ec37a2010-07-06 18:24:09 +00001150 <dt><tt><b>optsize</b></tt></dt>
1151 <dd>This attribute suggests that optimization passes and code generator passes
1152 make choices that keep the code size of this function low, and otherwise
1153 do optimizations specifically to reduce code size.</dd>
1154
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001155 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001156 <dd>This attribute indicates that the function computes its result (or decides
1157 to unwind an exception) based strictly on its arguments, without
1158 dereferencing any pointer arguments or otherwise accessing any mutable
1159 state (e.g. memory, control registers, etc) visible to caller functions.
1160 It does not write through any pointer arguments
1161 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1162 changes any state visible to callers. This means that it cannot unwind
1163 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1164 could use the <tt>unwind</tt> instruction.</dd>
Devang Patel5d96fda2009-06-12 19:45:19 +00001165
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001166 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001167 <dd>This attribute indicates that the function does not write through any
1168 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1169 arguments) or otherwise modify any state (e.g. memory, control registers,
1170 etc) visible to caller functions. It may dereference pointer arguments
1171 and read state that may be set in the caller. A readonly function always
1172 returns the same value (or unwinds an exception identically) when called
1173 with the same set of arguments and global state. It cannot unwind an
1174 exception by calling the <tt>C++</tt> exception throwing methods, but may
1175 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovc5ec8a72009-07-17 18:07:26 +00001176
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001177 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001178 <dd>This attribute indicates that the function should emit a stack smashing
1179 protector. It is in the form of a "canary"&mdash;a random value placed on
1180 the stack before the local variables that's checked upon return from the
1181 function to see if it has been overwritten. A heuristic is used to
1182 determine if a function needs stack protectors or not.<br>
1183<br>
1184 If a function that has an <tt>ssp</tt> attribute is inlined into a
1185 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1186 function will have an <tt>ssp</tt> attribute.</dd>
1187
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001188 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001189 <dd>This attribute indicates that the function should <em>always</em> emit a
1190 stack smashing protector. This overrides
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001191 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1192<br>
1193 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1194 function that doesn't have an <tt>sspreq</tt> attribute or which has
1195 an <tt>ssp</tt> attribute, then the resulting function will have
1196 an <tt>sspreq</tt> attribute.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001197</dl>
1198
Devang Patelf8b94812008-09-04 23:05:13 +00001199</div>
1200
1201<!-- ======================================================================= -->
1202<div class="doc_subsection">
Chris Lattner1eeeb0c2006-04-08 04:40:53 +00001203 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001204</div>
1205
1206<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001207
1208<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1209 the GCC "file scope inline asm" blocks. These blocks are internally
1210 concatenated by LLVM and treated as a single unit, but may be separated in
1211 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001212
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001213<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00001214module asm "inline asm code goes here"
1215module asm "more can go here"
1216</pre>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001217
1218<p>The strings can contain any character by escaping non-printable characters.
1219 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001220 for the number.</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001221
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001222<p>The inline asm code is simply printed to the machine code .s file when
1223 assembly code is generated.</p>
1224
Chris Lattner4e9aba72006-01-23 23:23:47 +00001225</div>
Chris Lattnerfa730212004-12-09 16:11:40 +00001226
Reid Spencerde151942007-02-19 23:54:10 +00001227<!-- ======================================================================= -->
1228<div class="doc_subsection">
1229 <a name="datalayout">Data Layout</a>
1230</div>
1231
1232<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001233
Reid Spencerde151942007-02-19 23:54:10 +00001234<p>A module may specify a target specific data layout string that specifies how
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001235 data is to be laid out in memory. The syntax for the data layout is
1236 simply:</p>
1237
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001238<pre class="doc_code">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001239target datalayout = "<i>layout specification</i>"
1240</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001241
1242<p>The <i>layout specification</i> consists of a list of specifications
1243 separated by the minus sign character ('-'). Each specification starts with
1244 a letter and may include other information after the letter to define some
1245 aspect of the data layout. The specifications accepted are as follows:</p>
1246
Reid Spencerde151942007-02-19 23:54:10 +00001247<dl>
1248 <dt><tt>E</tt></dt>
1249 <dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001250 bits with the most significance have the lowest address location.</dd>
1251
Reid Spencerde151942007-02-19 23:54:10 +00001252 <dt><tt>e</tt></dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +00001253 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001254 the bits with the least significance have the lowest address
1255 location.</dd>
1256
Reid Spencerde151942007-02-19 23:54:10 +00001257 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001258 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001259 <i>preferred</i> alignments. All sizes are in bits. Specifying
1260 the <i>pref</i> alignment is optional. If omitted, the
1261 preceding <tt>:</tt> should be omitted too.</dd>
1262
Reid Spencerde151942007-02-19 23:54:10 +00001263 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1264 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001265 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1266
Reid Spencerde151942007-02-19 23:54:10 +00001267 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001268 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001269 <i>size</i>.</dd>
1270
Reid Spencerde151942007-02-19 23:54:10 +00001271 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001272 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesen9d8d2212010-05-28 18:54:47 +00001273 <i>size</i>. Only values of <i>size</i> that are supported by the target
1274 will work. 32 (float) and 64 (double) are supported on all targets;
1275 80 or 128 (different flavors of long double) are also supported on some
1276 targets.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001277
Reid Spencerde151942007-02-19 23:54:10 +00001278 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1279 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001280 <i>size</i>.</dd>
1281
Daniel Dunbar87bde0b2009-06-08 22:17:53 +00001282 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1283 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001284 <i>size</i>.</dd>
Chris Lattnere82bdc42009-11-07 09:35:34 +00001285
1286 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1287 <dd>This specifies a set of native integer widths for the target CPU
1288 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1289 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001290 this set are considered to support most general arithmetic
Chris Lattnere82bdc42009-11-07 09:35:34 +00001291 operations efficiently.</dd>
Reid Spencerde151942007-02-19 23:54:10 +00001292</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001293
Reid Spencerde151942007-02-19 23:54:10 +00001294<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman1c70c002010-04-28 00:36:01 +00001295 default set of specifications which are then (possibly) overridden by the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001296 specifications in the <tt>datalayout</tt> keyword. The default specifications
1297 are given in this list:</p>
1298
Reid Spencerde151942007-02-19 23:54:10 +00001299<ul>
1300 <li><tt>E</tt> - big endian</li>
Dan Gohmanfdf2e8c2010-02-23 02:44:03 +00001301 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
Reid Spencerde151942007-02-19 23:54:10 +00001302 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1303 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1304 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1305 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattnerd3eda892008-08-05 18:29:16 +00001306 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Reid Spencerde151942007-02-19 23:54:10 +00001307 alignment of 64-bits</li>
1308 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1309 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1310 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1311 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1312 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar87bde0b2009-06-08 22:17:53 +00001313 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencerde151942007-02-19 23:54:10 +00001314</ul>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001315
1316<p>When LLVM is determining the alignment for a given type, it uses the
1317 following rules:</p>
1318
Reid Spencerde151942007-02-19 23:54:10 +00001319<ol>
1320 <li>If the type sought is an exact match for one of the specifications, that
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001321 specification is used.</li>
1322
Reid Spencerde151942007-02-19 23:54:10 +00001323 <li>If no match is found, and the type sought is an integer type, then the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001324 smallest integer type that is larger than the bitwidth of the sought type
1325 is used. If none of the specifications are larger than the bitwidth then
1326 the the largest integer type is used. For example, given the default
1327 specifications above, the i7 type will use the alignment of i8 (next
1328 largest) while both i65 and i256 will use the alignment of i64 (largest
1329 specified).</li>
1330
Reid Spencerde151942007-02-19 23:54:10 +00001331 <li>If no match is found, and the type sought is a vector type, then the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001332 largest vector type that is smaller than the sought vector type will be
1333 used as a fall back. This happens because &lt;128 x double&gt; can be
1334 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencerde151942007-02-19 23:54:10 +00001335</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001336
Reid Spencerde151942007-02-19 23:54:10 +00001337</div>
Chris Lattnerfa730212004-12-09 16:11:40 +00001338
Dan Gohman556ca272009-07-27 18:07:55 +00001339<!-- ======================================================================= -->
1340<div class="doc_subsection">
1341 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1342</div>
1343
1344<div class="doc_text">
1345
Andreas Bolka55e459a2009-07-29 00:02:05 +00001346<p>Any memory access must be done through a pointer value associated
Andreas Bolka99a82052009-07-27 20:37:10 +00001347with an address range of the memory access, otherwise the behavior
Dan Gohman556ca272009-07-27 18:07:55 +00001348is undefined. Pointer values are associated with address ranges
1349according to the following rules:</p>
1350
1351<ul>
Dan Gohman1e109622010-07-02 18:41:32 +00001352 <li>A pointer value is associated with the addresses associated with
1353 any value it is <i>based</i> on.
Andreas Bolka55e459a2009-07-29 00:02:05 +00001354 <li>An address of a global variable is associated with the address
Dan Gohman556ca272009-07-27 18:07:55 +00001355 range of the variable's storage.</li>
1356 <li>The result value of an allocation instruction is associated with
1357 the address range of the allocated storage.</li>
1358 <li>A null pointer in the default address-space is associated with
Andreas Bolka55e459a2009-07-29 00:02:05 +00001359 no address.</li>
Dan Gohman556ca272009-07-27 18:07:55 +00001360 <li>An integer constant other than zero or a pointer value returned
1361 from a function not defined within LLVM may be associated with address
1362 ranges allocated through mechanisms other than those provided by
Andreas Bolka55e459a2009-07-29 00:02:05 +00001363 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman556ca272009-07-27 18:07:55 +00001364 allocated by mechanisms provided by LLVM.</li>
Dan Gohman1e109622010-07-02 18:41:32 +00001365</ul>
1366
1367<p>A pointer value is <i>based</i> on another pointer value according
1368 to the following rules:</p>
1369
1370<ul>
1371 <li>A pointer value formed from a
1372 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1373 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1374 <li>The result value of a
1375 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1376 of the <tt>bitcast</tt>.</li>
1377 <li>A pointer value formed by an
1378 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1379 pointer values that contribute (directly or indirectly) to the
1380 computation of the pointer's value.</li>
1381 <li>The "<i>based</i> on" relationship is transitive.</li>
1382</ul>
1383
1384<p>Note that this definition of <i>"based"</i> is intentionally
1385 similar to the definition of <i>"based"</i> in C99, though it is
1386 slightly weaker.</p>
Dan Gohman556ca272009-07-27 18:07:55 +00001387
1388<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +00001389<tt><a href="#i_load">load</a></tt> merely indicates the size and
1390alignment of the memory from which to load, as well as the
Dan Gohmanc22c0f32010-06-17 19:23:50 +00001391interpretation of the value. The first operand type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +00001392<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1393and alignment of the store.</p>
Dan Gohman556ca272009-07-27 18:07:55 +00001394
1395<p>Consequently, type-based alias analysis, aka TBAA, aka
1396<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1397LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1398additional information which specialized optimization passes may use
1399to implement type-based alias analysis.</p>
1400
1401</div>
1402
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00001403<!-- ======================================================================= -->
1404<div class="doc_subsection">
1405 <a name="volatile">Volatile Memory Accesses</a>
1406</div>
1407
1408<div class="doc_text">
1409
1410<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1411href="#i_store"><tt>store</tt></a>s, and <a
1412href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1413The optimizers must not change the number of volatile operations or change their
1414order of execution relative to other volatile operations. The optimizers
1415<i>may</i> change the order of volatile operations relative to non-volatile
1416operations. This is not Java's "volatile" and has no cross-thread
1417synchronization behavior.</p>
1418
1419</div>
1420
Chris Lattner00950542001-06-06 20:29:01 +00001421<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00001422<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1423<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +00001424
Misha Brukman9d0919f2003-11-08 01:05:38 +00001425<div class="doc_text">
Chris Lattnerfa730212004-12-09 16:11:40 +00001426
Misha Brukman9d0919f2003-11-08 01:05:38 +00001427<p>The LLVM type system is one of the most important features of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001428 intermediate representation. Being typed enables a number of optimizations
1429 to be performed on the intermediate representation directly, without having
1430 to do extra analyses on the side before the transformation. A strong type
1431 system makes it easier to read the generated code and enables novel analyses
1432 and transformations that are not feasible to perform on normal three address
1433 code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +00001434
1435</div>
1436
Chris Lattner00950542001-06-06 20:29:01 +00001437<!-- ======================================================================= -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001438<div class="doc_subsection"> <a name="t_classifications">Type
Chris Lattner261efe92003-11-25 01:02:51 +00001439Classifications</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001440
Misha Brukman9d0919f2003-11-08 01:05:38 +00001441<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001442
1443<p>The types fall into a few useful classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001444
1445<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00001446 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001447 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +00001448 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +00001449 <td><a href="#t_integer">integer</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +00001450 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +00001451 </tr>
1452 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +00001453 <td><a href="#t_floating">floating point</a></td>
1454 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +00001455 </tr>
1456 <tr>
1457 <td><a name="t_firstclass">first class</a></td>
Chris Lattner4f69f462008-01-04 04:32:38 +00001458 <td><a href="#t_integer">integer</a>,
1459 <a href="#t_floating">floating point</a>,
1460 <a href="#t_pointer">pointer</a>,
Dan Gohman0066db62008-06-18 18:42:13 +00001461 <a href="#t_vector">vector</a>,
Dan Gohmana334d5f2008-05-12 23:51:09 +00001462 <a href="#t_struct">structure</a>,
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001463 <a href="#t_union">union</a>,
Dan Gohmana334d5f2008-05-12 23:51:09 +00001464 <a href="#t_array">array</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001465 <a href="#t_label">label</a>,
1466 <a href="#t_metadata">metadata</a>.
Reid Spencerca86e162006-12-31 07:07:53 +00001467 </td>
Chris Lattner261efe92003-11-25 01:02:51 +00001468 </tr>
Chris Lattner4f69f462008-01-04 04:32:38 +00001469 <tr>
1470 <td><a href="#t_primitive">primitive</a></td>
1471 <td><a href="#t_label">label</a>,
1472 <a href="#t_void">void</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001473 <a href="#t_floating">floating point</a>,
1474 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner4f69f462008-01-04 04:32:38 +00001475 </tr>
1476 <tr>
1477 <td><a href="#t_derived">derived</a></td>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001478 <td><a href="#t_array">array</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +00001479 <a href="#t_function">function</a>,
1480 <a href="#t_pointer">pointer</a>,
1481 <a href="#t_struct">structure</a>,
1482 <a href="#t_pstruct">packed structure</a>,
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001483 <a href="#t_union">union</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +00001484 <a href="#t_vector">vector</a>,
1485 <a href="#t_opaque">opaque</a>.
Dan Gohman01ac1012008-10-14 16:32:04 +00001486 </td>
Chris Lattner4f69f462008-01-04 04:32:38 +00001487 </tr>
Chris Lattner261efe92003-11-25 01:02:51 +00001488 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001489</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001490
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001491<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1492 important. Values of these types are the only ones which can be produced by
Nick Lewyckyec38da42009-09-27 00:45:11 +00001493 instructions.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001494
Misha Brukman9d0919f2003-11-08 01:05:38 +00001495</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001496
Chris Lattner00950542001-06-06 20:29:01 +00001497<!-- ======================================================================= -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001498<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner8f8c7b72008-01-04 04:34:14 +00001499
Chris Lattner4f69f462008-01-04 04:32:38 +00001500<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001501
Chris Lattner4f69f462008-01-04 04:32:38 +00001502<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001503 system.</p>
Chris Lattner4f69f462008-01-04 04:32:38 +00001504
Chris Lattner8f8c7b72008-01-04 04:34:14 +00001505</div>
1506
Chris Lattner4f69f462008-01-04 04:32:38 +00001507<!-- _______________________________________________________________________ -->
Nick Lewyckyec38da42009-09-27 00:45:11 +00001508<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1509
1510<div class="doc_text">
1511
1512<h5>Overview:</h5>
1513<p>The integer type is a very simple type that simply specifies an arbitrary
1514 bit width for the integer type desired. Any bit width from 1 bit to
1515 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1516
1517<h5>Syntax:</h5>
1518<pre>
1519 iN
1520</pre>
1521
1522<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1523 value.</p>
1524
1525<h5>Examples:</h5>
1526<table class="layout">
1527 <tr class="layout">
1528 <td class="left"><tt>i1</tt></td>
1529 <td class="left">a single-bit integer.</td>
1530 </tr>
1531 <tr class="layout">
1532 <td class="left"><tt>i32</tt></td>
1533 <td class="left">a 32-bit integer.</td>
1534 </tr>
1535 <tr class="layout">
1536 <td class="left"><tt>i1942652</tt></td>
1537 <td class="left">a really big integer of over 1 million bits.</td>
1538 </tr>
1539</table>
1540
Nick Lewyckyec38da42009-09-27 00:45:11 +00001541</div>
1542
1543<!-- _______________________________________________________________________ -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001544<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1545
1546<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001547
1548<table>
1549 <tbody>
1550 <tr><th>Type</th><th>Description</th></tr>
1551 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1552 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1553 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1554 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1555 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1556 </tbody>
1557</table>
1558
Chris Lattner4f69f462008-01-04 04:32:38 +00001559</div>
1560
1561<!-- _______________________________________________________________________ -->
1562<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1563
1564<div class="doc_text">
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001565
Chris Lattner4f69f462008-01-04 04:32:38 +00001566<h5>Overview:</h5>
1567<p>The void type does not represent any value and has no size.</p>
1568
1569<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +00001570<pre>
1571 void
1572</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001573
Chris Lattner4f69f462008-01-04 04:32:38 +00001574</div>
1575
1576<!-- _______________________________________________________________________ -->
1577<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1578
1579<div class="doc_text">
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001580
Chris Lattner4f69f462008-01-04 04:32:38 +00001581<h5>Overview:</h5>
1582<p>The label type represents code labels.</p>
1583
1584<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +00001585<pre>
1586 label
1587</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001588
Chris Lattner4f69f462008-01-04 04:32:38 +00001589</div>
1590
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001591<!-- _______________________________________________________________________ -->
1592<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1593
1594<div class="doc_text">
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001595
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001596<h5>Overview:</h5>
Nick Lewyckyc261df92009-09-27 23:27:42 +00001597<p>The metadata type represents embedded metadata. No derived types may be
1598 created from metadata except for <a href="#t_function">function</a>
1599 arguments.
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001600
1601<h5>Syntax:</h5>
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001602<pre>
1603 metadata
1604</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001605
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001606</div>
1607
Chris Lattner4f69f462008-01-04 04:32:38 +00001608
1609<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001610<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001611
Misha Brukman9d0919f2003-11-08 01:05:38 +00001612<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00001613
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001614<p>The real power in LLVM comes from the derived types in the system. This is
1615 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewyckyec38da42009-09-27 00:45:11 +00001616 useful types. Each of these types contain one or more element types which
1617 may be a primitive type, or another derived type. For example, it is
1618 possible to have a two dimensional array, using an array as the element type
1619 of another array.</p>
Dan Gohmand8791e52009-01-24 15:58:40 +00001620
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001621
1622</div>
1623
1624<!-- _______________________________________________________________________ -->
1625<div class="doc_subsubsection"> <a name="t_aggregate">Aggregate Types</a> </div>
1626
1627<div class="doc_text">
1628
1629<p>Aggregate Types are a subset of derived types that can contain multiple
1630 member types. <a href="#t_array">Arrays</a>,
1631 <a href="#t_struct">structs</a>, <a href="#t_vector">vectors</a> and
1632 <a href="#t_union">unions</a> are aggregate types.</p>
1633
1634</div>
1635
Reid Spencer2b916312007-05-16 18:44:01 +00001636<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001637<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001638
Misha Brukman9d0919f2003-11-08 01:05:38 +00001639<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00001640
Chris Lattner00950542001-06-06 20:29:01 +00001641<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001642<p>The array type is a very simple derived type that arranges elements
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001643 sequentially in memory. The array type requires a size (number of elements)
1644 and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001645
Chris Lattner7faa8832002-04-14 06:13:44 +00001646<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001647<pre>
1648 [&lt;# elements&gt; x &lt;elementtype&gt;]
1649</pre>
1650
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001651<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1652 be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001653
Chris Lattner7faa8832002-04-14 06:13:44 +00001654<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001655<table class="layout">
1656 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001657 <td class="left"><tt>[40 x i32]</tt></td>
1658 <td class="left">Array of 40 32-bit integer values.</td>
1659 </tr>
1660 <tr class="layout">
1661 <td class="left"><tt>[41 x i32]</tt></td>
1662 <td class="left">Array of 41 32-bit integer values.</td>
1663 </tr>
1664 <tr class="layout">
1665 <td class="left"><tt>[4 x i8]</tt></td>
1666 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001667 </tr>
Chris Lattner00950542001-06-06 20:29:01 +00001668</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001669<p>Here are some examples of multidimensional arrays:</p>
1670<table class="layout">
1671 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001672 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1673 <td class="left">3x4 array of 32-bit integer values.</td>
1674 </tr>
1675 <tr class="layout">
1676 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1677 <td class="left">12x10 array of single precision floating point values.</td>
1678 </tr>
1679 <tr class="layout">
1680 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1681 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001682 </tr>
1683</table>
Chris Lattnere67a9512005-06-24 17:22:57 +00001684
Dan Gohman7657f6b2009-11-09 19:01:53 +00001685<p>There is no restriction on indexing beyond the end of the array implied by
1686 a static type (though there are restrictions on indexing beyond the bounds
1687 of an allocated object in some cases). This means that single-dimension
1688 'variable sized array' addressing can be implemented in LLVM with a zero
1689 length array type. An implementation of 'pascal style arrays' in LLVM could
1690 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +00001691
Misha Brukman9d0919f2003-11-08 01:05:38 +00001692</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001693
Chris Lattner00950542001-06-06 20:29:01 +00001694<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001695<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001696
Misha Brukman9d0919f2003-11-08 01:05:38 +00001697<div class="doc_text">
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001698
Chris Lattner00950542001-06-06 20:29:01 +00001699<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001700<p>The function type can be thought of as a function signature. It consists of
1701 a return type and a list of formal parameter types. The return type of a
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001702 function type is a scalar type, a void type, a struct type, or a union
1703 type. If the return type is a struct type then all struct elements must be
1704 of first class types, and the struct must have at least one element.</p>
Devang Patelc3fc6df2008-03-10 20:49:15 +00001705
Chris Lattner00950542001-06-06 20:29:01 +00001706<h5>Syntax:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001707<pre>
Nick Lewycky51386942009-09-27 07:55:32 +00001708 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001709</pre>
1710
John Criswell0ec250c2005-10-24 16:17:18 +00001711<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001712 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1713 which indicates that the function takes a variable number of arguments.
1714 Variable argument functions can access their arguments with
1715 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner0724fbd2010-03-02 06:36:51 +00001716 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewyckyc261df92009-09-27 23:27:42 +00001717 <a href="#t_label">label</a>.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001718
Chris Lattner00950542001-06-06 20:29:01 +00001719<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001720<table class="layout">
1721 <tr class="layout">
Reid Spencer92f82302006-12-31 07:18:34 +00001722 <td class="left"><tt>i32 (i32)</tt></td>
1723 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001724 </td>
Reid Spencer92f82302006-12-31 07:18:34 +00001725 </tr><tr class="layout">
Chris Lattner0724fbd2010-03-02 06:36:51 +00001726 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencerf17a0b72006-12-31 07:20:23 +00001727 </tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001728 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner0724fbd2010-03-02 06:36:51 +00001729 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1730 returning <tt>float</tt>.
Reid Spencer92f82302006-12-31 07:18:34 +00001731 </td>
1732 </tr><tr class="layout">
1733 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001734 <td class="left">A vararg function that takes at least one
1735 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1736 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer92f82302006-12-31 07:18:34 +00001737 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +00001738 </td>
Devang Patela582f402008-03-24 05:35:41 +00001739 </tr><tr class="layout">
1740 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky51386942009-09-27 07:55:32 +00001741 <td class="left">A function taking an <tt>i32</tt>, returning a
1742 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patela582f402008-03-24 05:35:41 +00001743 </td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001744 </tr>
Chris Lattner00950542001-06-06 20:29:01 +00001745</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001746
Misha Brukman9d0919f2003-11-08 01:05:38 +00001747</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001748
Chris Lattner00950542001-06-06 20:29:01 +00001749<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001750<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001751
Misha Brukman9d0919f2003-11-08 01:05:38 +00001752<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001753
Chris Lattner00950542001-06-06 20:29:01 +00001754<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001755<p>The structure type is used to represent a collection of data members together
1756 in memory. The packing of the field types is defined to match the ABI of the
1757 underlying processor. The elements of a structure may be any type that has a
1758 size.</p>
1759
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +00001760<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1761 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1762 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1763 Structures in registers are accessed using the
1764 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1765 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001766<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001767<pre>
1768 { &lt;type list&gt; }
1769</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001770
Chris Lattner00950542001-06-06 20:29:01 +00001771<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001772<table class="layout">
1773 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +00001774 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1775 <td class="left">A triple of three <tt>i32</tt> values</td>
1776 </tr><tr class="layout">
1777 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1778 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1779 second element is a <a href="#t_pointer">pointer</a> to a
1780 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1781 an <tt>i32</tt>.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001782 </tr>
Chris Lattner00950542001-06-06 20:29:01 +00001783</table>
Dan Gohmand8791e52009-01-24 15:58:40 +00001784
Misha Brukman9d0919f2003-11-08 01:05:38 +00001785</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001786
Chris Lattner00950542001-06-06 20:29:01 +00001787<!-- _______________________________________________________________________ -->
Andrew Lenharth75e10682006-12-08 17:13:00 +00001788<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1789</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001790
Andrew Lenharth75e10682006-12-08 17:13:00 +00001791<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001792
Andrew Lenharth75e10682006-12-08 17:13:00 +00001793<h5>Overview:</h5>
1794<p>The packed structure type is used to represent a collection of data members
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001795 together in memory. There is no padding between fields. Further, the
1796 alignment of a packed structure is 1 byte. The elements of a packed
1797 structure may be any type that has a size.</p>
1798
1799<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1800 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1801 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1802
Andrew Lenharth75e10682006-12-08 17:13:00 +00001803<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001804<pre>
1805 &lt; { &lt;type list&gt; } &gt;
1806</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001807
Andrew Lenharth75e10682006-12-08 17:13:00 +00001808<h5>Examples:</h5>
1809<table class="layout">
1810 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +00001811 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1812 <td class="left">A triple of three <tt>i32</tt> values</td>
1813 </tr><tr class="layout">
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001814 <td class="left">
1815<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00001816 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1817 second element is a <a href="#t_pointer">pointer</a> to a
1818 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1819 an <tt>i32</tt>.</td>
Andrew Lenharth75e10682006-12-08 17:13:00 +00001820 </tr>
1821</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001822
Andrew Lenharth75e10682006-12-08 17:13:00 +00001823</div>
1824
1825<!-- _______________________________________________________________________ -->
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001826<div class="doc_subsubsection"> <a name="t_union">Union Type</a> </div>
1827
1828<div class="doc_text">
1829
1830<h5>Overview:</h5>
1831<p>A union type describes an object with size and alignment suitable for
1832 an object of any one of a given set of types (also known as an "untagged"
1833 union). It is similar in concept and usage to a
1834 <a href="#t_struct">struct</a>, except that all members of the union
1835 have an offset of zero. The elements of a union may be any type that has a
1836 size. Unions must have at least one member - empty unions are not allowed.
1837 </p>
1838
1839<p>The size of the union as a whole will be the size of its largest member,
1840 and the alignment requirements of the union as a whole will be the largest
1841 alignment requirement of any member.</p>
1842
Dan Gohman2eddfef2010-02-25 16:51:31 +00001843<p>Union members are accessed using '<tt><a href="#i_load">load</a></tt> and
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001844 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1845 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1846 Since all members are at offset zero, the getelementptr instruction does
1847 not affect the address, only the type of the resulting pointer.</p>
1848
1849<h5>Syntax:</h5>
1850<pre>
1851 union { &lt;type list&gt; }
1852</pre>
1853
1854<h5>Examples:</h5>
1855<table class="layout">
1856 <tr class="layout">
1857 <td class="left"><tt>union { i32, i32*, float }</tt></td>
1858 <td class="left">A union of three types: an <tt>i32</tt>, a pointer to
1859 an <tt>i32</tt>, and a <tt>float</tt>.</td>
1860 </tr><tr class="layout">
1861 <td class="left">
1862 <tt>union {&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1863 <td class="left">A union, where the first element is a <tt>float</tt> and the
1864 second element is a <a href="#t_pointer">pointer</a> to a
1865 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1866 an <tt>i32</tt>.</td>
1867 </tr>
1868</table>
1869
1870</div>
1871
1872<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001873<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Chris Lattner0fd4a272009-02-08 19:53:29 +00001874
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001875<div class="doc_text">
1876
1877<h5>Overview:</h5>
Dan Gohmanff3ef322010-02-25 16:50:07 +00001878<p>The pointer type is used to specify memory locations.
1879 Pointers are commonly used to reference objects in memory.</p>
1880
1881<p>Pointer types may have an optional address space attribute defining the
1882 numbered address space where the pointed-to object resides. The default
1883 address space is number zero. The semantics of non-zero address
1884 spaces are target-specific.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001885
1886<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1887 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner0fd4a272009-02-08 19:53:29 +00001888
Chris Lattner7faa8832002-04-14 06:13:44 +00001889<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001890<pre>
1891 &lt;type&gt; *
1892</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001893
Chris Lattner7faa8832002-04-14 06:13:44 +00001894<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001895<table class="layout">
1896 <tr class="layout">
Dan Gohman2a08c532009-01-04 23:44:43 +00001897 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +00001898 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1899 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1900 </tr>
1901 <tr class="layout">
Dan Gohmanfe47aae2010-05-28 17:13:49 +00001902 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +00001903 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +00001904 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner23ff1f92007-12-19 05:04:11 +00001905 <tt>i32</tt>.</td>
1906 </tr>
1907 <tr class="layout">
1908 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1909 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1910 that resides in address space #5.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001911 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001912</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001913
Misha Brukman9d0919f2003-11-08 01:05:38 +00001914</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001915
Chris Lattnera58561b2004-08-12 19:12:28 +00001916<!-- _______________________________________________________________________ -->
Reid Spencer485bad12007-02-15 03:07:05 +00001917<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001918
Misha Brukman9d0919f2003-11-08 01:05:38 +00001919<div class="doc_text">
Chris Lattner69c11bb2005-04-25 17:34:15 +00001920
Chris Lattnera58561b2004-08-12 19:12:28 +00001921<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001922<p>A vector type is a simple derived type that represents a vector of elements.
1923 Vector types are used when multiple primitive data are operated in parallel
1924 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sandsd40d14e2009-11-27 13:38:03 +00001925 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001926 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001927
Chris Lattnera58561b2004-08-12 19:12:28 +00001928<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001929<pre>
1930 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1931</pre>
1932
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001933<p>The number of elements is a constant integer value; elementtype may be any
1934 integer or floating point type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001935
Chris Lattnera58561b2004-08-12 19:12:28 +00001936<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001937<table class="layout">
1938 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001939 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1940 <td class="left">Vector of 4 32-bit integer values.</td>
1941 </tr>
1942 <tr class="layout">
1943 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1944 <td class="left">Vector of 8 32-bit floating-point values.</td>
1945 </tr>
1946 <tr class="layout">
1947 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1948 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001949 </tr>
1950</table>
Dan Gohmand8791e52009-01-24 15:58:40 +00001951
Misha Brukman9d0919f2003-11-08 01:05:38 +00001952</div>
1953
Chris Lattner69c11bb2005-04-25 17:34:15 +00001954<!-- _______________________________________________________________________ -->
1955<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1956<div class="doc_text">
1957
1958<h5>Overview:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001959<p>Opaque types are used to represent unknown types in the system. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001960 corresponds (for example) to the C notion of a forward declared structure
1961 type. In LLVM, opaque types can eventually be resolved to any type (not just
1962 a structure type).</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001963
1964<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001965<pre>
1966 opaque
1967</pre>
1968
1969<h5>Examples:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001970<table class="layout">
1971 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001972 <td class="left"><tt>opaque</tt></td>
1973 <td class="left">An opaque type.</td>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001974 </tr>
1975</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001976
Chris Lattner69c11bb2005-04-25 17:34:15 +00001977</div>
1978
Chris Lattner242d61d2009-02-02 07:32:36 +00001979<!-- ======================================================================= -->
1980<div class="doc_subsection">
1981 <a name="t_uprefs">Type Up-references</a>
1982</div>
1983
1984<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001985
Chris Lattner242d61d2009-02-02 07:32:36 +00001986<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001987<p>An "up reference" allows you to refer to a lexically enclosing type without
1988 requiring it to have a name. For instance, a structure declaration may
1989 contain a pointer to any of the types it is lexically a member of. Example
1990 of up references (with their equivalent as named type declarations)
1991 include:</p>
Chris Lattner242d61d2009-02-02 07:32:36 +00001992
1993<pre>
Chris Lattner3060f5b2009-02-09 10:00:56 +00001994 { \2 * } %x = type { %x* }
Chris Lattner242d61d2009-02-02 07:32:36 +00001995 { \2 }* %y = type { %y }*
1996 \1* %z = type %z*
1997</pre>
1998
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001999<p>An up reference is needed by the asmprinter for printing out cyclic types
2000 when there is no declared name for a type in the cycle. Because the
2001 asmprinter does not want to print out an infinite type string, it needs a
2002 syntax to handle recursive types that have no names (all names are optional
2003 in llvm IR).</p>
Chris Lattner242d61d2009-02-02 07:32:36 +00002004
2005<h5>Syntax:</h5>
2006<pre>
2007 \&lt;level&gt;
2008</pre>
2009
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002010<p>The level is the count of the lexical type that is being referred to.</p>
Chris Lattner242d61d2009-02-02 07:32:36 +00002011
2012<h5>Examples:</h5>
Chris Lattner242d61d2009-02-02 07:32:36 +00002013<table class="layout">
2014 <tr class="layout">
2015 <td class="left"><tt>\1*</tt></td>
2016 <td class="left">Self-referential pointer.</td>
2017 </tr>
2018 <tr class="layout">
2019 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
2020 <td class="left">Recursive structure where the upref refers to the out-most
2021 structure.</td>
2022 </tr>
2023</table>
Chris Lattner242d61d2009-02-02 07:32:36 +00002024
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002025</div>
Chris Lattner69c11bb2005-04-25 17:34:15 +00002026
Chris Lattnerc3f59762004-12-09 17:30:23 +00002027<!-- *********************************************************************** -->
2028<div class="doc_section"> <a name="constants">Constants</a> </div>
2029<!-- *********************************************************************** -->
2030
2031<div class="doc_text">
2032
2033<p>LLVM has several different basic types of constants. This section describes
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002034 them all and their syntax.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002035
2036</div>
2037
2038<!-- ======================================================================= -->
Reid Spencercc16dc32004-12-09 18:02:53 +00002039<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002040
2041<div class="doc_text">
2042
2043<dl>
2044 <dt><b>Boolean constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002045 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewyckyec38da42009-09-27 00:45:11 +00002046 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002047
2048 <dt><b>Integer constants</b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002049 <dd>Standard integers (such as '4') are constants of
2050 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2051 with integer types.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002052
2053 <dt><b>Floating point constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002054 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002055 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2056 notation (see below). The assembler requires the exact decimal value of a
2057 floating-point constant. For example, the assembler accepts 1.25 but
2058 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2059 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002060
2061 <dt><b>Null pointer constants</b></dt>
John Criswell9e2485c2004-12-10 15:51:16 +00002062 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002063 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002064</dl>
2065
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002066<p>The one non-intuitive notation for constants is the hexadecimal form of
2067 floating point constants. For example, the form '<tt>double
2068 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2069 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2070 constants are required (and the only time that they are generated by the
2071 disassembler) is when a floating point constant must be emitted but it cannot
2072 be represented as a decimal floating point number in a reasonable number of
2073 digits. For example, NaN's, infinities, and other special values are
2074 represented in their IEEE hexadecimal format so that assembly and disassembly
2075 do not cause any bits to change in the constants.</p>
2076
Dale Johannesenbd5e5a82009-02-11 22:14:51 +00002077<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002078 represented using the 16-digit form shown above (which matches the IEEE754
2079 representation for double); float values must, however, be exactly
2080 representable as IEE754 single precision. Hexadecimal format is always used
2081 for long double, and there are three forms of long double. The 80-bit format
2082 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2083 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2084 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2085 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2086 currently supported target uses this format. Long doubles will only work if
2087 they match the long double format on your target. All hexadecimal formats
2088 are big-endian (sign bit at the left).</p>
2089
Chris Lattnerc3f59762004-12-09 17:30:23 +00002090</div>
2091
2092<!-- ======================================================================= -->
Chris Lattner70882792009-02-28 18:32:25 +00002093<div class="doc_subsection">
Bill Wendlingd9fe2982009-07-20 02:32:41 +00002094<a name="aggregateconstants"></a> <!-- old anchor -->
2095<a name="complexconstants">Complex Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002096</div>
2097
2098<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002099
Chris Lattner70882792009-02-28 18:32:25 +00002100<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002101 constants and smaller complex constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002102
2103<dl>
2104 <dt><b>Structure constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002105 <dd>Structure constants are represented with notation similar to structure
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002106 type definitions (a comma separated list of elements, surrounded by braces
2107 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2108 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2109 Structure constants must have <a href="#t_struct">structure type</a>, and
2110 the number and types of elements must match those specified by the
2111 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002112
Chris Lattnerfdfeb692010-02-12 20:49:41 +00002113 <dt><b>Union constants</b></dt>
2114 <dd>Union constants are represented with notation similar to a structure with
2115 a single element - that is, a single typed element surrounded
2116 by braces (<tt>{}</tt>)). For example: "<tt>{ i32 4 }</tt>". The
2117 <a href="#t_union">union type</a> can be initialized with a single-element
2118 struct as long as the type of the struct element matches the type of
2119 one of the union members.</dd>
2120
Chris Lattnerc3f59762004-12-09 17:30:23 +00002121 <dt><b>Array constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002122 <dd>Array constants are represented with notation similar to array type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002123 definitions (a comma separated list of elements, surrounded by square
2124 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2125 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2126 the number and types of elements must match those specified by the
2127 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002128
Reid Spencer485bad12007-02-15 03:07:05 +00002129 <dt><b>Vector constants</b></dt>
Reid Spencer485bad12007-02-15 03:07:05 +00002130 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002131 definitions (a comma separated list of elements, surrounded by
2132 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2133 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2134 have <a href="#t_vector">vector type</a>, and the number and types of
2135 elements must match those specified by the type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002136
2137 <dt><b>Zero initialization</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002138 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattnerfdfeb692010-02-12 20:49:41 +00002139 value to zero of <em>any</em> type, including scalar and
2140 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002141 This is often used to avoid having to print large zero initializers
2142 (e.g. for large arrays) and is always exactly equivalent to using explicit
2143 zero initializers.</dd>
Nick Lewycky21cc4462009-04-04 07:22:01 +00002144
2145 <dt><b>Metadata node</b></dt>
Nick Lewycky1e8c7a62009-05-30 16:08:30 +00002146 <dd>A metadata node is a structure-like constant with
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002147 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2148 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2149 be interpreted as part of the instruction stream, metadata is a place to
2150 attach additional information such as debug info.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002151</dl>
2152
2153</div>
2154
2155<!-- ======================================================================= -->
2156<div class="doc_subsection">
2157 <a name="globalconstants">Global Variable and Function Addresses</a>
2158</div>
2159
2160<div class="doc_text">
2161
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002162<p>The addresses of <a href="#globalvars">global variables</a>
2163 and <a href="#functionstructure">functions</a> are always implicitly valid
2164 (link-time) constants. These constants are explicitly referenced when
2165 the <a href="#identifiers">identifier for the global</a> is used and always
2166 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2167 legal LLVM file:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002168
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002169<pre class="doc_code">
Chris Lattnera18a4242007-06-06 18:28:13 +00002170@X = global i32 17
2171@Y = global i32 42
2172@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +00002173</pre>
2174
2175</div>
2176
2177<!-- ======================================================================= -->
Reid Spencer2dc45b82004-12-09 18:13:12 +00002178<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002179<div class="doc_text">
2180
Chris Lattner48a109c2009-09-07 22:52:39 +00002181<p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
Benjamin Kramer8040cd32009-10-12 14:46:08 +00002182 indicates that the user of the value may receive an unspecified bit-pattern.
Chris Lattner48a109c2009-09-07 22:52:39 +00002183 Undefined values may be of any type (other than label or void) and be used
2184 anywhere a constant is permitted.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002185
Chris Lattnerc608cb12009-09-11 01:49:31 +00002186<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner48a109c2009-09-07 22:52:39 +00002187 program is well defined no matter what value is used. This gives the
2188 compiler more freedom to optimize. Here are some examples of (potentially
2189 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002190
Chris Lattner48a109c2009-09-07 22:52:39 +00002191
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002192<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002193 %A = add %X, undef
2194 %B = sub %X, undef
2195 %C = xor %X, undef
2196Safe:
2197 %A = undef
2198 %B = undef
2199 %C = undef
2200</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002201
2202<p>This is safe because all of the output bits are affected by the undef bits.
2203Any output bit can have a zero or one depending on the input bits.</p>
2204
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002205<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002206 %A = or %X, undef
2207 %B = and %X, undef
2208Safe:
2209 %A = -1
2210 %B = 0
2211Unsafe:
2212 %A = undef
2213 %B = undef
2214</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002215
2216<p>These logical operations have bits that are not always affected by the input.
2217For example, if "%X" has a zero bit, then the output of the 'and' operation will
2218always be a zero, no matter what the corresponding bit from the undef is. As
Chris Lattnerc608cb12009-09-11 01:49:31 +00002219such, it is unsafe to optimize or assume that the result of the and is undef.
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002220However, it is safe to assume that all bits of the undef could be 0, and
2221optimize the and to 0. Likewise, it is safe to assume that all the bits of
2222the undef operand to the or could be set, allowing the or to be folded to
Chris Lattnerc608cb12009-09-11 01:49:31 +00002223-1.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +00002224
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002225<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002226 %A = select undef, %X, %Y
2227 %B = select undef, 42, %Y
2228 %C = select %X, %Y, undef
2229Safe:
2230 %A = %X (or %Y)
2231 %B = 42 (or %Y)
2232 %C = %Y
2233Unsafe:
2234 %A = undef
2235 %B = undef
2236 %C = undef
2237</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002238
2239<p>This set of examples show that undefined select (and conditional branch)
2240conditions can go "either way" but they have to come from one of the two
2241operands. In the %A example, if %X and %Y were both known to have a clear low
2242bit, then %A would have to have a cleared low bit. However, in the %C example,
2243the optimizer is allowed to assume that the undef operand could be the same as
2244%Y, allowing the whole select to be eliminated.</p>
2245
2246
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002247<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002248 %A = xor undef, undef
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002249
Chris Lattner48a109c2009-09-07 22:52:39 +00002250 %B = undef
2251 %C = xor %B, %B
2252
2253 %D = undef
2254 %E = icmp lt %D, 4
2255 %F = icmp gte %D, 4
2256
2257Safe:
2258 %A = undef
2259 %B = undef
2260 %C = undef
2261 %D = undef
2262 %E = undef
2263 %F = undef
2264</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002265
2266<p>This example points out that two undef operands are not necessarily the same.
2267This can be surprising to people (and also matches C semantics) where they
2268assume that "X^X" is always zero, even if X is undef. This isn't true for a
2269number of reasons, but the short answer is that an undef "variable" can
2270arbitrarily change its value over its "live range". This is true because the
2271"variable" doesn't actually <em>have a live range</em>. Instead, the value is
2272logically read from arbitrary registers that happen to be around when needed,
Benjamin Kramer8040cd32009-10-12 14:46:08 +00002273so the value is not necessarily consistent over time. In fact, %A and %C need
Chris Lattner349eb412009-09-08 15:13:16 +00002274to have the same semantics or the core LLVM "replace all uses with" concept
Chris Lattner48a109c2009-09-07 22:52:39 +00002275would not hold.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002276
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002277<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +00002278 %A = fdiv undef, %X
2279 %B = fdiv %X, undef
2280Safe:
2281 %A = undef
2282b: unreachable
2283</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002284
2285<p>These examples show the crucial difference between an <em>undefined
2286value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2287allowed to have an arbitrary bit-pattern. This means that the %A operation
2288can be constant folded to undef because the undef could be an SNaN, and fdiv is
2289not (currently) defined on SNaN's. However, in the second example, we can make
2290a more aggressive assumption: because the undef is allowed to be an arbitrary
2291value, we are allowed to assume that it could be zero. Since a divide by zero
Chris Lattner8e371aa2009-09-08 19:45:34 +00002292has <em>undefined behavior</em>, we are allowed to assume that the operation
Chris Lattner6e9057b2009-09-07 23:33:52 +00002293does not execute at all. This allows us to delete the divide and all code after
2294it: since the undefined operation "can't happen", the optimizer can assume that
2295it occurs in dead code.
2296</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002297
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002298<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +00002299a: store undef -> %X
2300b: store %X -> undef
2301Safe:
2302a: &lt;deleted&gt;
2303b: unreachable
2304</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002305
2306<p>These examples reiterate the fdiv example: a store "of" an undefined value
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002307can be assumed to not have any effect: we can assume that the value is
Chris Lattner6e9057b2009-09-07 23:33:52 +00002308overwritten with bits that happen to match what was already there. However, a
2309store "to" an undefined location could clobber arbitrary memory, therefore, it
2310has undefined behavior.</p>
2311
Chris Lattnerc3f59762004-12-09 17:30:23 +00002312</div>
2313
2314<!-- ======================================================================= -->
Dan Gohmanfff6c532010-04-22 23:14:21 +00002315<div class="doc_subsection"><a name="trapvalues">Trap Values</a></div>
2316<div class="doc_text">
2317
Dan Gohmanc68ce062010-04-26 20:21:21 +00002318<p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmanfff6c532010-04-22 23:14:21 +00002319 instead of representing an unspecified bit pattern, they represent the
2320 fact that an instruction or constant expression which cannot evoke side
2321 effects has nevertheless detected a condition which results in undefined
Dan Gohmanc68ce062010-04-26 20:21:21 +00002322 behavior.</p>
Dan Gohmanfff6c532010-04-22 23:14:21 +00002323
Dan Gohman34b3d992010-04-28 00:49:41 +00002324<p>There is currently no way of representing a trap value in the IR; they
Dan Gohman855abed2010-05-03 14:51:43 +00002325 only exist when produced by operations such as
Dan Gohman34b3d992010-04-28 00:49:41 +00002326 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002327
Dan Gohman34b3d992010-04-28 00:49:41 +00002328<p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002329
Dan Gohman34b3d992010-04-28 00:49:41 +00002330<ul>
2331<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2332 their operands.</li>
2333
2334<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2335 to their dynamic predecessor basic block.</li>
2336
2337<li>Function arguments depend on the corresponding actual argument values in
2338 the dynamic callers of their functions.</li>
2339
2340<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2341 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2342 control back to them.</li>
2343
Dan Gohmanb5328162010-05-03 14:55:22 +00002344<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2345 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2346 or exception-throwing call instructions that dynamically transfer control
2347 back to them.</li>
2348
Dan Gohman34b3d992010-04-28 00:49:41 +00002349<li>Non-volatile loads and stores depend on the most recent stores to all of the
2350 referenced memory addresses, following the order in the IR
2351 (including loads and stores implied by intrinsics such as
2352 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2353
Dan Gohman7c24ff12010-05-03 14:59:34 +00002354<!-- TODO: In the case of multiple threads, this only applies if the store
2355 "happens-before" the load or store. -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002356
Dan Gohman34b3d992010-04-28 00:49:41 +00002357<!-- TODO: floating-point exception state -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002358
Dan Gohman34b3d992010-04-28 00:49:41 +00002359<li>An instruction with externally visible side effects depends on the most
2360 recent preceding instruction with externally visible side effects, following
Dan Gohmanff70fe42010-07-06 15:26:33 +00002361 the order in the IR. (This includes
2362 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanfff6c532010-04-22 23:14:21 +00002363
Dan Gohmanb5328162010-05-03 14:55:22 +00002364<li>An instruction <i>control-depends</i> on a
2365 <a href="#terminators">terminator instruction</a>
2366 if the terminator instruction has multiple successors and the instruction
2367 is always executed when control transfers to one of the successors, and
2368 may not be executed when control is transfered to another.</li>
Dan Gohman34b3d992010-04-28 00:49:41 +00002369
2370<li>Dependence is transitive.</li>
2371
2372</ul>
Dan Gohman34b3d992010-04-28 00:49:41 +00002373
2374<p>Whenever a trap value is generated, all values which depend on it evaluate
2375 to trap. If they have side effects, the evoke their side effects as if each
2376 operand with a trap value were undef. If they have externally-visible side
2377 effects, the behavior is undefined.</p>
2378
2379<p>Here are some examples:</p>
Dan Gohmanc30f6e12010-04-26 20:54:53 +00002380
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002381<pre class="doc_code">
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002382entry:
2383 %trap = sub nuw i32 0, 1 ; Results in a trap value.
Dan Gohman34b3d992010-04-28 00:49:41 +00002384 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2385 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2386 store i32 0, i32* %trap_yet_again ; undefined behavior
2387
2388 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2389 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2390
2391 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2392
2393 %narrowaddr = bitcast i32* @g to i16*
2394 %wideaddr = bitcast i32* @g to i64*
2395 %trap3 = load 16* %narrowaddr ; Returns a trap value.
2396 %trap4 = load i64* %widaddr ; Returns a trap value.
2397
2398 %cmp = icmp i32 slt %trap, 0 ; Returns a trap value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002399 %br i1 %cmp, %true, %end ; Branch to either destination.
2400
2401true:
Dan Gohman34b3d992010-04-28 00:49:41 +00002402 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2403 ; it has undefined behavior.
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002404 br label %end
2405
2406end:
2407 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2408 ; Both edges into this PHI are
2409 ; control-dependent on %cmp, so this
Dan Gohman34b3d992010-04-28 00:49:41 +00002410 ; always results in a trap value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002411
2412 volatile store i32 0, i32* @g ; %end is control-equivalent to %entry
2413 ; so this is defined (ignoring earlier
2414 ; undefined behavior in this example).
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002415</pre>
Dan Gohmanfff6c532010-04-22 23:14:21 +00002416
Dan Gohmanfff6c532010-04-22 23:14:21 +00002417</div>
2418
2419<!-- ======================================================================= -->
Chris Lattnerf9d078e2009-10-27 21:19:13 +00002420<div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2421 Blocks</a></div>
Chris Lattnerc6f44362009-10-27 21:01:34 +00002422<div class="doc_text">
2423
Chris Lattnercdfc9402009-11-01 01:27:45 +00002424<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerc6f44362009-10-27 21:01:34 +00002425
2426<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner2dfdf2a2009-10-27 21:49:40 +00002427 basic block in the specified function, and always has an i8* type. Taking
Chris Lattnercdfc9402009-11-01 01:27:45 +00002428 the address of the entry block is illegal.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002429
Chris Lattnerc6f44362009-10-27 21:01:34 +00002430<p>This value only has defined behavior when used as an operand to the
Chris Lattnerab21db72009-10-28 00:19:10 +00002431 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
Chris Lattnerc6f44362009-10-27 21:01:34 +00002432 against null. Pointer equality tests between labels addresses is undefined
2433 behavior - though, again, comparison against null is ok, and no label is
Chris Lattnerf9d078e2009-10-27 21:19:13 +00002434 equal to the null pointer. This may also be passed around as an opaque
2435 pointer sized value as long as the bits are not inspected. This allows
Chris Lattner3fd77ce2009-10-27 21:44:20 +00002436 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
Chris Lattnerab21db72009-10-28 00:19:10 +00002437 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002438
Chris Lattnerf9d078e2009-10-27 21:19:13 +00002439<p>Finally, some targets may provide defined semantics when
Chris Lattnerc6f44362009-10-27 21:01:34 +00002440 using the value as the operand to an inline assembly, but that is target
2441 specific.
2442 </p>
2443
2444</div>
2445
2446
2447<!-- ======================================================================= -->
Chris Lattnerc3f59762004-12-09 17:30:23 +00002448<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2449</div>
2450
2451<div class="doc_text">
2452
2453<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002454 to be used as constants. Constant expressions may be of
2455 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2456 operation that does not have side effects (e.g. load and call are not
2457 supported). The following is the syntax for constant expressions:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002458
2459<dl>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002460 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002461 <dd>Truncate a constant to another type. The bit size of CST must be larger
2462 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002463
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002464 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002465 <dd>Zero extend a constant to another type. The bit size of CST must be
Duncan Sands28afd432010-07-13 12:06:14 +00002466 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002467
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002468 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002469 <dd>Sign extend a constant to another type. The bit size of CST must be
Duncan Sands28afd432010-07-13 12:06:14 +00002470 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002471
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002472 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002473 <dd>Truncate a floating point constant to another floating point type. The
2474 size of CST must be larger than the size of TYPE. Both types must be
2475 floating point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002476
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002477 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002478 <dd>Floating point extend a constant to another type. The size of CST must be
2479 smaller or equal to the size of TYPE. Both types must be floating
2480 point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002481
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002482 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002483 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002484 constant. TYPE must be a scalar or vector integer type. CST must be of
2485 scalar or vector floating point type. Both CST and TYPE must be scalars,
2486 or vectors of the same number of elements. If the value won't fit in the
2487 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002488
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002489 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002490 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002491 constant. TYPE must be a scalar or vector integer type. CST must be of
2492 scalar or vector floating point type. Both CST and TYPE must be scalars,
2493 or vectors of the same number of elements. If the value won't fit in the
2494 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002495
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002496 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002497 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002498 constant. TYPE must be a scalar or vector floating point type. CST must be
2499 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2500 vectors of the same number of elements. If the value won't fit in the
2501 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002502
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002503 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002504 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002505 constant. TYPE must be a scalar or vector floating point type. CST must be
2506 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2507 vectors of the same number of elements. If the value won't fit in the
2508 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002509
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002510 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5c0ef472006-11-11 23:08:07 +00002511 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002512 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2513 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2514 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +00002515
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002516 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002517 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2518 type. CST must be of integer type. The CST value is zero extended,
2519 truncated, or unchanged to make it fit in a pointer size. This one is
2520 <i>really</i> dangerous!</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +00002521
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002522 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner03bbad62009-02-28 18:27:03 +00002523 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2524 are the same as those for the <a href="#i_bitcast">bitcast
2525 instruction</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002526
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002527 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2528 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002529 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002530 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2531 instruction, the index list may have zero or more indexes, which are
2532 required to make sense for the type of "CSTPTR".</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002533
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002534 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002535 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer01c42592006-12-04 19:23:19 +00002536
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002537 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +00002538 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2539
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002540 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +00002541 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +00002542
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002543 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002544 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2545 constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +00002546
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002547 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002548 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2549 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +00002550
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002551 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002552 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2553 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +00002554
Nick Lewycky9e130ce2010-05-29 06:44:15 +00002555 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2556 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2557 constants. The index list is interpreted in a similar manner as indices in
2558 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2559 index value must be specified.</dd>
2560
2561 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2562 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2563 constants. The index list is interpreted in a similar manner as indices in
2564 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2565 index value must be specified.</dd>
2566
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002567 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002568 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2569 be any of the <a href="#binaryops">binary</a>
2570 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2571 on operands are the same as those for the corresponding instruction
2572 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002573</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002574
Chris Lattnerc3f59762004-12-09 17:30:23 +00002575</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +00002576
Chris Lattner00950542001-06-06 20:29:01 +00002577<!-- *********************************************************************** -->
Chris Lattnere87d6532006-01-25 23:47:57 +00002578<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2579<!-- *********************************************************************** -->
2580
2581<!-- ======================================================================= -->
2582<div class="doc_subsection">
2583<a name="inlineasm">Inline Assembler Expressions</a>
2584</div>
2585
2586<div class="doc_text">
2587
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002588<p>LLVM supports inline assembler expressions (as opposed
2589 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2590 a special value. This value represents the inline assembler as a string
2591 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen09fed252009-10-13 21:56:55 +00002592 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +00002593 expression has side effects, and a flag indicating whether the function
2594 containing the asm needs to align its stack conservatively. An example
2595 inline assembler expression is:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002596
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002597<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00002598i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +00002599</pre>
2600
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002601<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2602 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2603 have:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002604
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002605<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00002606%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattnere87d6532006-01-25 23:47:57 +00002607</pre>
2608
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002609<p>Inline asms with side effects not visible in the constraint list must be
2610 marked as having side effects. This is done through the use of the
2611 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002612
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002613<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00002614call void asm sideeffect "eieio", ""()
Chris Lattnere87d6532006-01-25 23:47:57 +00002615</pre>
2616
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +00002617<p>In some cases inline asms will contain code that will not work unless the
2618 stack is aligned in some way, such as calls or SSE instructions on x86,
2619 yet will not contain code that does that alignment within the asm.
2620 The compiler should make conservative assumptions about what the asm might
2621 contain and should generate its usual stack alignment code in the prologue
2622 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen09fed252009-10-13 21:56:55 +00002623
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002624<pre class="doc_code">
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +00002625call void asm alignstack "eieio", ""()
Dale Johannesen09fed252009-10-13 21:56:55 +00002626</pre>
Dale Johannesen09fed252009-10-13 21:56:55 +00002627
2628<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2629 first.</p>
2630
Chris Lattnere87d6532006-01-25 23:47:57 +00002631<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002632 documented here. Constraints on what can be done (e.g. duplication, moving,
2633 etc need to be documented). This is probably best done by reference to
2634 another document that covers inline asm from a holistic perspective.</p>
Chris Lattnercf9a4152010-04-07 05:38:05 +00002635</div>
2636
2637<div class="doc_subsubsection">
2638<a name="inlineasm_md">Inline Asm Metadata</a>
2639</div>
2640
2641<div class="doc_text">
2642
2643<p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2644 attached to it that contains a constant integer. If present, the code
2645 generator will use the integer as the location cookie value when report
2646 errors through the LLVMContext error reporting mechanisms. This allows a
Dan Gohman1c70c002010-04-28 00:36:01 +00002647 front-end to correlate backend errors that occur with inline asm back to the
Chris Lattnercf9a4152010-04-07 05:38:05 +00002648 source code that produced it. For example:</p>
2649
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002650<pre class="doc_code">
Chris Lattnercf9a4152010-04-07 05:38:05 +00002651call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2652...
2653!42 = !{ i32 1234567 }
2654</pre>
Chris Lattnercf9a4152010-04-07 05:38:05 +00002655
2656<p>It is up to the front-end to make sense of the magic numbers it places in the
2657 IR.</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002658
2659</div>
2660
Chris Lattnere6a5ddd2010-01-15 21:50:19 +00002661<!-- ======================================================================= -->
2662<div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2663 Strings</a>
2664</div>
2665
2666<div class="doc_text">
2667
2668<p>LLVM IR allows metadata to be attached to instructions in the program that
2669 can convey extra information about the code to the optimizers and code
2670 generator. One example application of metadata is source-level debug
2671 information. There are two metadata primitives: strings and nodes. All
2672 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2673 preceding exclamation point ('<tt>!</tt>').</p>
2674
2675<p>A metadata string is a string surrounded by double quotes. It can contain
2676 any character by escaping non-printable characters with "\xx" where "xx" is
2677 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2678
2679<p>Metadata nodes are represented with notation similar to structure constants
2680 (a comma separated list of elements, surrounded by braces and preceded by an
2681 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2682 10}</tt>". Metadata nodes can have any values as their operand.</p>
2683
2684<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2685 metadata nodes, which can be looked up in the module symbol table. For
2686 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2687
Devang Patele1d50cd2010-03-04 23:44:48 +00002688<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002689 function is using two metadata arguments.</p>
Devang Patele1d50cd2010-03-04 23:44:48 +00002690
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002691 <pre class="doc_code">
Devang Patele1d50cd2010-03-04 23:44:48 +00002692 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2693 </pre>
Devang Patele1d50cd2010-03-04 23:44:48 +00002694
2695<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002696 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.</p>
Devang Patele1d50cd2010-03-04 23:44:48 +00002697
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002698 <pre class="doc_code">
Devang Patele1d50cd2010-03-04 23:44:48 +00002699 %indvar.next = add i64 %indvar, 1, !dbg !21
2700 </pre>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +00002701</div>
2702
Chris Lattner857755c2009-07-20 05:55:19 +00002703
2704<!-- *********************************************************************** -->
2705<div class="doc_section">
2706 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2707</div>
2708<!-- *********************************************************************** -->
2709
2710<p>LLVM has a number of "magic" global variables that contain data that affect
2711code generation or other IR semantics. These are documented here. All globals
Chris Lattner401e10c2009-07-20 06:14:25 +00002712of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2713section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2714by LLVM.</p>
Chris Lattner857755c2009-07-20 05:55:19 +00002715
2716<!-- ======================================================================= -->
2717<div class="doc_subsection">
2718<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2719</div>
2720
2721<div class="doc_text">
2722
2723<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2724href="#linkage_appending">appending linkage</a>. This array contains a list of
2725pointers to global variables and functions which may optionally have a pointer
2726cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2727
2728<pre>
2729 @X = global i8 4
2730 @Y = global i32 123
2731
2732 @llvm.used = appending global [2 x i8*] [
2733 i8* @X,
2734 i8* bitcast (i32* @Y to i8*)
2735 ], section "llvm.metadata"
2736</pre>
2737
2738<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2739compiler, assembler, and linker are required to treat the symbol as if there is
2740a reference to the global that it cannot see. For example, if a variable has
2741internal linkage and no references other than that from the <tt>@llvm.used</tt>
2742list, it cannot be deleted. This is commonly used to represent references from
2743inline asms and other things the compiler cannot "see", and corresponds to
2744"attribute((used))" in GNU C.</p>
2745
2746<p>On some targets, the code generator must emit a directive to the assembler or
2747object file to prevent the assembler and linker from molesting the symbol.</p>
2748
2749</div>
2750
2751<!-- ======================================================================= -->
2752<div class="doc_subsection">
Chris Lattner401e10c2009-07-20 06:14:25 +00002753<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2754</div>
2755
2756<div class="doc_text">
2757
2758<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2759<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2760touching the symbol. On targets that support it, this allows an intelligent
2761linker to optimize references to the symbol without being impeded as it would be
2762by <tt>@llvm.used</tt>.</p>
2763
2764<p>This is a rare construct that should only be used in rare circumstances, and
2765should not be exposed to source languages.</p>
2766
2767</div>
2768
2769<!-- ======================================================================= -->
2770<div class="doc_subsection">
Chris Lattner857755c2009-07-20 05:55:19 +00002771<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2772</div>
2773
2774<div class="doc_text">
David Chisnalle31e9962010-04-30 19:23:49 +00002775<pre>
2776%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +00002777@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalle31e9962010-04-30 19:23:49 +00002778</pre>
2779<p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor functions and associated priorities. The functions referenced by this array will be called in ascending order of priority (i.e. lowest first) when the module is loaded. The order of functions with the same priority is not defined.
2780</p>
Chris Lattner857755c2009-07-20 05:55:19 +00002781
2782</div>
2783
2784<!-- ======================================================================= -->
2785<div class="doc_subsection">
2786<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2787</div>
2788
2789<div class="doc_text">
David Chisnalle31e9962010-04-30 19:23:49 +00002790<pre>
2791%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +00002792@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalle31e9962010-04-30 19:23:49 +00002793</pre>
Chris Lattner857755c2009-07-20 05:55:19 +00002794
David Chisnalle31e9962010-04-30 19:23:49 +00002795<p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions and associated priorities. The functions referenced by this array will be called in descending order of priority (i.e. highest first) when the module is loaded. The order of functions with the same priority is not defined.
2796</p>
Chris Lattner857755c2009-07-20 05:55:19 +00002797
2798</div>
2799
2800
Chris Lattnere87d6532006-01-25 23:47:57 +00002801<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00002802<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2803<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +00002804
Misha Brukman9d0919f2003-11-08 01:05:38 +00002805<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00002806
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002807<p>The LLVM instruction set consists of several different classifications of
2808 instructions: <a href="#terminators">terminator
2809 instructions</a>, <a href="#binaryops">binary instructions</a>,
2810 <a href="#bitwiseops">bitwise binary instructions</a>,
2811 <a href="#memoryops">memory instructions</a>, and
2812 <a href="#otherops">other instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002813
Misha Brukman9d0919f2003-11-08 01:05:38 +00002814</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002815
Chris Lattner00950542001-06-06 20:29:01 +00002816<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00002817<div class="doc_subsection"> <a name="terminators">Terminator
2818Instructions</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002819
Misha Brukman9d0919f2003-11-08 01:05:38 +00002820<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00002821
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002822<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2823 in a program ends with a "Terminator" instruction, which indicates which
2824 block should be executed after the current block is finished. These
2825 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2826 control flow, not values (the one exception being the
2827 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2828
Duncan Sands83821c82010-04-15 20:35:54 +00002829<p>There are seven different terminator instructions: the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002830 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2831 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2832 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
Bill Wendling21c346e2009-11-02 00:25:26 +00002833 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002834 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2835 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2836 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002837
Misha Brukman9d0919f2003-11-08 01:05:38 +00002838</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002839
Chris Lattner00950542001-06-06 20:29:01 +00002840<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00002841<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2842Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002843
Misha Brukman9d0919f2003-11-08 01:05:38 +00002844<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002845
Chris Lattner00950542001-06-06 20:29:01 +00002846<h5>Syntax:</h5>
Dan Gohmanb1e6b962008-10-04 19:00:07 +00002847<pre>
2848 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00002849 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +00002850</pre>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002851
Chris Lattner00950542001-06-06 20:29:01 +00002852<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002853<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2854 a value) from a function back to the caller.</p>
2855
2856<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2857 value and then causes control flow, and one that just causes control flow to
2858 occur.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002859
Chris Lattner00950542001-06-06 20:29:01 +00002860<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002861<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2862 return value. The type of the return value must be a
2863 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmanb1e6b962008-10-04 19:00:07 +00002864
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002865<p>A function is not <a href="#wellformed">well formed</a> if it it has a
2866 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2867 value or a return value with a type that does not match its type, or if it
2868 has a void return type and contains a '<tt>ret</tt>' instruction with a
2869 return value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002870
Chris Lattner00950542001-06-06 20:29:01 +00002871<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002872<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2873 the calling function's context. If the caller is a
2874 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2875 instruction after the call. If the caller was an
2876 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2877 the beginning of the "normal" destination block. If the instruction returns
2878 a value, that value shall set the call or invoke instruction's return
2879 value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002880
Chris Lattner00950542001-06-06 20:29:01 +00002881<h5>Example:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002882<pre>
2883 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00002884 ret void <i>; Return from a void function</i>
Bill Wendling0a4bbbf2009-02-28 22:12:54 +00002885 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Chris Lattner00950542001-06-06 20:29:01 +00002886</pre>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +00002887
Misha Brukman9d0919f2003-11-08 01:05:38 +00002888</div>
Chris Lattner00950542001-06-06 20:29:01 +00002889<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00002890<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002891
Misha Brukman9d0919f2003-11-08 01:05:38 +00002892<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002893
Chris Lattner00950542001-06-06 20:29:01 +00002894<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002895<pre>
2896 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 +00002897</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002898
Chris Lattner00950542001-06-06 20:29:01 +00002899<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002900<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2901 different basic block in the current function. There are two forms of this
2902 instruction, corresponding to a conditional branch and an unconditional
2903 branch.</p>
2904
Chris Lattner00950542001-06-06 20:29:01 +00002905<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002906<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2907 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2908 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2909 target.</p>
2910
Chris Lattner00950542001-06-06 20:29:01 +00002911<h5>Semantics:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +00002912<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002913 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2914 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2915 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2916
Chris Lattner00950542001-06-06 20:29:01 +00002917<h5>Example:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00002918<pre>
2919Test:
2920 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2921 br i1 %cond, label %IfEqual, label %IfUnequal
2922IfEqual:
2923 <a href="#i_ret">ret</a> i32 1
2924IfUnequal:
2925 <a href="#i_ret">ret</a> i32 0
2926</pre>
2927
Misha Brukman9d0919f2003-11-08 01:05:38 +00002928</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002929
Chris Lattner00950542001-06-06 20:29:01 +00002930<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002931<div class="doc_subsubsection">
2932 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2933</div>
2934
Misha Brukman9d0919f2003-11-08 01:05:38 +00002935<div class="doc_text">
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002936
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002937<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002938<pre>
2939 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2940</pre>
2941
Chris Lattner00950542001-06-06 20:29:01 +00002942<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002943<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002944 several different places. It is a generalization of the '<tt>br</tt>'
2945 instruction, allowing a branch to occur to one of many possible
2946 destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002947
Chris Lattner00950542001-06-06 20:29:01 +00002948<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002949<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002950 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2951 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2952 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002953
Chris Lattner00950542001-06-06 20:29:01 +00002954<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002955<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002956 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2957 is searched for the given value. If the value is found, control flow is
Benjamin Kramer8040cd32009-10-12 14:46:08 +00002958 transferred to the corresponding destination; otherwise, control flow is
2959 transferred to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +00002960
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002961<h5>Implementation:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002962<p>Depending on properties of the target machine and the particular
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002963 <tt>switch</tt> instruction, this instruction may be code generated in
2964 different ways. For example, it could be generated as a series of chained
2965 conditional branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002966
2967<h5>Example:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002968<pre>
2969 <i>; Emulate a conditional br instruction</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00002970 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman2a08c532009-01-04 23:44:43 +00002971 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002972
2973 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +00002974 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002975
2976 <i>; Implement a jump table:</i>
Dan Gohman2a08c532009-01-04 23:44:43 +00002977 switch i32 %val, label %otherwise [ i32 0, label %onzero
2978 i32 1, label %onone
2979 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +00002980</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002981
Misha Brukman9d0919f2003-11-08 01:05:38 +00002982</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00002983
Chris Lattnerf9be95f2009-10-27 19:13:16 +00002984
2985<!-- _______________________________________________________________________ -->
2986<div class="doc_subsubsection">
Chris Lattnerab21db72009-10-28 00:19:10 +00002987 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
Chris Lattnerf9be95f2009-10-27 19:13:16 +00002988</div>
2989
2990<div class="doc_text">
2991
2992<h5>Syntax:</h5>
2993<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +00002994 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +00002995</pre>
2996
2997<h5>Overview:</h5>
2998
Chris Lattnerab21db72009-10-28 00:19:10 +00002999<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003000 within the current function, whose address is specified by
Chris Lattnerc6f44362009-10-27 21:01:34 +00003001 "<tt>address</tt>". Address must be derived from a <a
3002 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003003
3004<h5>Arguments:</h5>
3005
3006<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3007 rest of the arguments indicate the full set of possible destinations that the
3008 address may point to. Blocks are allowed to occur multiple times in the
3009 destination list, though this isn't particularly useful.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003010
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003011<p>This destination list is required so that dataflow analysis has an accurate
3012 understanding of the CFG.</p>
3013
3014<h5>Semantics:</h5>
3015
3016<p>Control transfers to the block specified in the address argument. All
3017 possible destination blocks must be listed in the label list, otherwise this
3018 instruction has undefined behavior. This implies that jumps to labels
3019 defined in other functions have undefined behavior as well.</p>
3020
3021<h5>Implementation:</h5>
3022
3023<p>This is typically implemented with a jump through a register.</p>
3024
3025<h5>Example:</h5>
3026<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +00003027 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003028</pre>
3029
3030</div>
3031
3032
Chris Lattner00950542001-06-06 20:29:01 +00003033<!-- _______________________________________________________________________ -->
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003034<div class="doc_subsubsection">
3035 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3036</div>
3037
Misha Brukman9d0919f2003-11-08 01:05:38 +00003038<div class="doc_text">
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003039
Chris Lattner00950542001-06-06 20:29:01 +00003040<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003041<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +00003042 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Chris Lattner76b8a332006-05-14 18:23:06 +00003043 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003044</pre>
3045
Chris Lattner6536cfe2002-05-06 22:08:29 +00003046<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003047<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003048 function, with the possibility of control flow transfer to either the
3049 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3050 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3051 control flow will return to the "normal" label. If the callee (or any
3052 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3053 instruction, control is interrupted and continued at the dynamically nearest
3054 "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003055
Chris Lattner00950542001-06-06 20:29:01 +00003056<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003057<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003058
Chris Lattner00950542001-06-06 20:29:01 +00003059<ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003060 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3061 convention</a> the call should use. If none is specified, the call
3062 defaults to using C calling conventions.</li>
Devang Patelf642f472008-10-06 18:50:38 +00003063
3064 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003065 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3066 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelf642f472008-10-06 18:50:38 +00003067
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003068 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003069 function value being invoked. In most cases, this is a direct function
3070 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3071 off an arbitrary pointer to function value.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003072
3073 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003074 function to be invoked. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003075
3076 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +00003077 signature argument types and parameter attributes. All arguments must be
3078 of <a href="#t_firstclass">first class</a> type. If the function
3079 signature indicates the function accepts a variable number of arguments,
3080 the extra arguments can be specified.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003081
3082 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003083 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003084
3085 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003086 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003087
Devang Patel307e8ab2008-10-07 17:48:33 +00003088 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003089 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3090 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner00950542001-06-06 20:29:01 +00003091</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003092
Chris Lattner00950542001-06-06 20:29:01 +00003093<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003094<p>This instruction is designed to operate as a standard
3095 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3096 primary difference is that it establishes an association with a label, which
3097 is used by the runtime library to unwind the stack.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003098
3099<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003100 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3101 exception. Additionally, this is important for implementation of
3102 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003103
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003104<p>For the purposes of the SSA form, the definition of the value returned by the
3105 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3106 block to the "normal" label. If the callee unwinds then no return value is
3107 available.</p>
Dan Gohmanf96a4992009-05-22 21:47:08 +00003108
Chris Lattnerdf7a6802010-01-15 18:08:37 +00003109<p>Note that the code generator does not yet completely support unwind, and
3110that the invoke/unwind semantics are likely to change in future versions.</p>
3111
Chris Lattner00950542001-06-06 20:29:01 +00003112<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003113<pre>
Nick Lewyckyd703f652008-03-16 07:18:12 +00003114 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +00003115 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckyd703f652008-03-16 07:18:12 +00003116 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +00003117 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +00003118</pre>
Chris Lattner35eca582004-10-16 18:04:13 +00003119
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003120</div>
Chris Lattner35eca582004-10-16 18:04:13 +00003121
Chris Lattner27f71f22003-09-03 00:41:47 +00003122<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +00003123
Chris Lattner261efe92003-11-25 01:02:51 +00003124<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
3125Instruction</a> </div>
Chris Lattner35eca582004-10-16 18:04:13 +00003126
Misha Brukman9d0919f2003-11-08 01:05:38 +00003127<div class="doc_text">
Chris Lattner35eca582004-10-16 18:04:13 +00003128
Chris Lattner27f71f22003-09-03 00:41:47 +00003129<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003130<pre>
3131 unwind
3132</pre>
3133
Chris Lattner27f71f22003-09-03 00:41:47 +00003134<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003135<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003136 at the first callee in the dynamic call stack which used
3137 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3138 This is primarily used to implement exception handling.</p>
Chris Lattner35eca582004-10-16 18:04:13 +00003139
Chris Lattner27f71f22003-09-03 00:41:47 +00003140<h5>Semantics:</h5>
Chris Lattner72ed2002008-04-19 21:01:16 +00003141<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003142 immediately halt. The dynamic call stack is then searched for the
3143 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3144 Once found, execution continues at the "exceptional" destination block
3145 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3146 instruction in the dynamic call chain, undefined behavior results.</p>
3147
Chris Lattnerdf7a6802010-01-15 18:08:37 +00003148<p>Note that the code generator does not yet completely support unwind, and
3149that the invoke/unwind semantics are likely to change in future versions.</p>
3150
Misha Brukman9d0919f2003-11-08 01:05:38 +00003151</div>
Chris Lattner35eca582004-10-16 18:04:13 +00003152
3153<!-- _______________________________________________________________________ -->
3154
3155<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
3156Instruction</a> </div>
3157
3158<div class="doc_text">
3159
3160<h5>Syntax:</h5>
3161<pre>
3162 unreachable
3163</pre>
3164
3165<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003166<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003167 instruction is used to inform the optimizer that a particular portion of the
3168 code is not reachable. This can be used to indicate that the code after a
3169 no-return function cannot be reached, and other facts.</p>
Chris Lattner35eca582004-10-16 18:04:13 +00003170
3171<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003172<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003173
Chris Lattner35eca582004-10-16 18:04:13 +00003174</div>
3175
Chris Lattner00950542001-06-06 20:29:01 +00003176<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00003177<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003178
Misha Brukman9d0919f2003-11-08 01:05:38 +00003179<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003180
3181<p>Binary operators are used to do most of the computation in a program. They
3182 require two operands of the same type, execute an operation on them, and
3183 produce a single value. The operands might represent multiple data, as is
3184 the case with the <a href="#t_vector">vector</a> data type. The result value
3185 has the same type as its operands.</p>
3186
Misha Brukman9d0919f2003-11-08 01:05:38 +00003187<p>There are several different binary operators:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003188
Misha Brukman9d0919f2003-11-08 01:05:38 +00003189</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003190
Chris Lattner00950542001-06-06 20:29:01 +00003191<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003192<div class="doc_subsubsection">
3193 <a name="i_add">'<tt>add</tt>' Instruction</a>
3194</div>
3195
Misha Brukman9d0919f2003-11-08 01:05:38 +00003196<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003197
Chris Lattner00950542001-06-06 20:29:01 +00003198<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003199<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003200 &lt;result&gt; = add &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +00003201 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3202 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3203 &lt;result&gt; = add nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00003204</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003205
Chris Lattner00950542001-06-06 20:29:01 +00003206<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003207<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003208
Chris Lattner00950542001-06-06 20:29:01 +00003209<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003210<p>The two arguments to the '<tt>add</tt>' instruction must
3211 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3212 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003213
Chris Lattner00950542001-06-06 20:29:01 +00003214<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003215<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003216
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003217<p>If the sum has unsigned overflow, the result returned is the mathematical
3218 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003219
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003220<p>Because LLVM integers use a two's complement representation, this instruction
3221 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003222
Dan Gohman08d012e2009-07-22 22:44:56 +00003223<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3224 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3225 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +00003226 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3227 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003228
Chris Lattner00950542001-06-06 20:29:01 +00003229<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003230<pre>
3231 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00003232</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003233
Misha Brukman9d0919f2003-11-08 01:05:38 +00003234</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003235
Chris Lattner00950542001-06-06 20:29:01 +00003236<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003237<div class="doc_subsubsection">
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003238 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3239</div>
3240
3241<div class="doc_text">
3242
3243<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003244<pre>
3245 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3246</pre>
3247
3248<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003249<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3250
3251<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003252<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003253 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3254 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003255
3256<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003257<p>The value produced is the floating point sum of the two operands.</p>
3258
3259<h5>Example:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003260<pre>
3261 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3262</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003263
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003264</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003265
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003266<!-- _______________________________________________________________________ -->
3267<div class="doc_subsubsection">
Chris Lattner5568e942008-05-20 20:48:21 +00003268 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3269</div>
3270
Misha Brukman9d0919f2003-11-08 01:05:38 +00003271<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003272
Chris Lattner00950542001-06-06 20:29:01 +00003273<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003274<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +00003275 &lt;result&gt; = sub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +00003276 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3277 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3278 &lt;result&gt; = sub nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00003279</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003280
Chris Lattner00950542001-06-06 20:29:01 +00003281<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003282<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003283 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003284
3285<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003286 '<tt>neg</tt>' instruction present in most other intermediate
3287 representations.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003288
Chris Lattner00950542001-06-06 20:29:01 +00003289<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003290<p>The two arguments to the '<tt>sub</tt>' instruction must
3291 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3292 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003293
Chris Lattner00950542001-06-06 20:29:01 +00003294<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003295<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003296
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003297<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003298 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3299 result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003300
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003301<p>Because LLVM integers use a two's complement representation, this instruction
3302 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003303
Dan Gohman08d012e2009-07-22 22:44:56 +00003304<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3305 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3306 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +00003307 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3308 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003309
Chris Lattner00950542001-06-06 20:29:01 +00003310<h5>Example:</h5>
Bill Wendlingaac388b2007-05-29 09:42:13 +00003311<pre>
3312 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003313 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +00003314</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003315
Misha Brukman9d0919f2003-11-08 01:05:38 +00003316</div>
Chris Lattner5568e942008-05-20 20:48:21 +00003317
Chris Lattner00950542001-06-06 20:29:01 +00003318<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003319<div class="doc_subsubsection">
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003320 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3321</div>
3322
3323<div class="doc_text">
3324
3325<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003326<pre>
3327 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3328</pre>
3329
3330<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003331<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003332 operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003333
3334<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003335 '<tt>fneg</tt>' instruction present in most other intermediate
3336 representations.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003337
3338<h5>Arguments:</h5>
Bill Wendlingd9fe2982009-07-20 02:32:41 +00003339<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003340 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3341 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003342
3343<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003344<p>The value produced is the floating point difference of the two operands.</p>
3345
3346<h5>Example:</h5>
3347<pre>
3348 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3349 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3350</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003351
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003352</div>
3353
3354<!-- _______________________________________________________________________ -->
3355<div class="doc_subsubsection">
Chris Lattner5568e942008-05-20 20:48:21 +00003356 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3357</div>
3358
Misha Brukman9d0919f2003-11-08 01:05:38 +00003359<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003360
Chris Lattner00950542001-06-06 20:29:01 +00003361<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003362<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +00003363 &lt;result&gt; = mul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +00003364 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3365 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3366 &lt;result&gt; = mul nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00003367</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003368
Chris Lattner00950542001-06-06 20:29:01 +00003369<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003370<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003371
Chris Lattner00950542001-06-06 20:29:01 +00003372<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003373<p>The two arguments to the '<tt>mul</tt>' instruction must
3374 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3375 integer values. Both arguments must have identical types.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003376
Chris Lattner00950542001-06-06 20:29:01 +00003377<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003378<p>The value produced is the integer product of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003379
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003380<p>If the result of the multiplication has unsigned overflow, the result
3381 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3382 width of the result.</p>
3383
3384<p>Because LLVM integers use a two's complement representation, and the result
3385 is the same width as the operands, this instruction returns the correct
3386 result for both signed and unsigned integers. If a full product
3387 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3388 be sign-extended or zero-extended as appropriate to the width of the full
3389 product.</p>
3390
Dan Gohman08d012e2009-07-22 22:44:56 +00003391<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3392 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3393 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +00003394 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3395 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003396
Chris Lattner00950542001-06-06 20:29:01 +00003397<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003398<pre>
3399 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00003400</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003401
Misha Brukman9d0919f2003-11-08 01:05:38 +00003402</div>
Chris Lattner5568e942008-05-20 20:48:21 +00003403
Chris Lattner00950542001-06-06 20:29:01 +00003404<!-- _______________________________________________________________________ -->
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003405<div class="doc_subsubsection">
3406 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3407</div>
3408
3409<div class="doc_text">
3410
3411<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003412<pre>
3413 &lt;result&gt; = fmul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003414</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003415
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003416<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003417<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003418
3419<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003420<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003421 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3422 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003423
3424<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003425<p>The value produced is the floating point product of the two operands.</p>
3426
3427<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003428<pre>
3429 &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003430</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003431
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003432</div>
3433
3434<!-- _______________________________________________________________________ -->
Reid Spencer1628cec2006-10-26 06:15:43 +00003435<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3436</a></div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003437
Reid Spencer1628cec2006-10-26 06:15:43 +00003438<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003439
Reid Spencer1628cec2006-10-26 06:15:43 +00003440<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003441<pre>
3442 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003443</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003444
Reid Spencer1628cec2006-10-26 06:15:43 +00003445<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003446<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003447
Reid Spencer1628cec2006-10-26 06:15:43 +00003448<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003449<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003450 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3451 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003452
Reid Spencer1628cec2006-10-26 06:15:43 +00003453<h5>Semantics:</h5>
Chris Lattner5ec89832008-01-28 00:36:27 +00003454<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003455
Chris Lattner5ec89832008-01-28 00:36:27 +00003456<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003457 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3458
Chris Lattner5ec89832008-01-28 00:36:27 +00003459<p>Division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003460
Reid Spencer1628cec2006-10-26 06:15:43 +00003461<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003462<pre>
3463 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003464</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003465
Reid Spencer1628cec2006-10-26 06:15:43 +00003466</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003467
Reid Spencer1628cec2006-10-26 06:15:43 +00003468<!-- _______________________________________________________________________ -->
3469<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3470</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003471
Reid Spencer1628cec2006-10-26 06:15:43 +00003472<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003473
Reid Spencer1628cec2006-10-26 06:15:43 +00003474<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003475<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003476 &lt;result&gt; = sdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +00003477 &lt;result&gt; = sdiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003478</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003479
Reid Spencer1628cec2006-10-26 06:15:43 +00003480<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003481<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003482
Reid Spencer1628cec2006-10-26 06:15:43 +00003483<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003484<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003485 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3486 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003487
Reid Spencer1628cec2006-10-26 06:15:43 +00003488<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003489<p>The value produced is the signed integer quotient of the two operands rounded
3490 towards zero.</p>
3491
Chris Lattner5ec89832008-01-28 00:36:27 +00003492<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003493 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3494
Chris Lattner5ec89832008-01-28 00:36:27 +00003495<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003496 undefined behavior; this is a rare case, but can occur, for example, by doing
3497 a 32-bit division of -2147483648 by -1.</p>
3498
Dan Gohman9c5beed2009-07-22 00:04:19 +00003499<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman27ef9972010-04-23 15:23:32 +00003500 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
Dan Gohman38da9272010-07-11 00:08:34 +00003501 be rounded.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003502
Reid Spencer1628cec2006-10-26 06:15:43 +00003503<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003504<pre>
3505 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003506</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003507
Reid Spencer1628cec2006-10-26 06:15:43 +00003508</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003509
Reid Spencer1628cec2006-10-26 06:15:43 +00003510<!-- _______________________________________________________________________ -->
3511<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +00003512Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003513
Misha Brukman9d0919f2003-11-08 01:05:38 +00003514<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003515
Chris Lattner00950542001-06-06 20:29:01 +00003516<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003517<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +00003518 &lt;result&gt; = fdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +00003519</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003520
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003521<h5>Overview:</h5>
3522<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003523
Chris Lattner261efe92003-11-25 01:02:51 +00003524<h5>Arguments:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00003525<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003526 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3527 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003528
Chris Lattner261efe92003-11-25 01:02:51 +00003529<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +00003530<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003531
Chris Lattner261efe92003-11-25 01:02:51 +00003532<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003533<pre>
3534 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +00003535</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003536
Chris Lattner261efe92003-11-25 01:02:51 +00003537</div>
Chris Lattner5568e942008-05-20 20:48:21 +00003538
Chris Lattner261efe92003-11-25 01:02:51 +00003539<!-- _______________________________________________________________________ -->
Reid Spencer0a783f72006-11-02 01:53:59 +00003540<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3541</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003542
Reid Spencer0a783f72006-11-02 01:53:59 +00003543<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003544
Reid Spencer0a783f72006-11-02 01:53:59 +00003545<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003546<pre>
3547 &lt;result&gt; = urem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer0a783f72006-11-02 01:53:59 +00003548</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003549
Reid Spencer0a783f72006-11-02 01:53:59 +00003550<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003551<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3552 division of its two arguments.</p>
3553
Reid Spencer0a783f72006-11-02 01:53:59 +00003554<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003555<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003556 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3557 values. Both arguments must have identical types.</p>
3558
Reid Spencer0a783f72006-11-02 01:53:59 +00003559<h5>Semantics:</h5>
3560<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003561 This instruction always performs an unsigned division to get the
3562 remainder.</p>
3563
Chris Lattner5ec89832008-01-28 00:36:27 +00003564<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003565 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3566
Chris Lattner5ec89832008-01-28 00:36:27 +00003567<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003568
Reid Spencer0a783f72006-11-02 01:53:59 +00003569<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003570<pre>
3571 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +00003572</pre>
3573
3574</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003575
Reid Spencer0a783f72006-11-02 01:53:59 +00003576<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003577<div class="doc_subsubsection">
3578 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3579</div>
3580
Chris Lattner261efe92003-11-25 01:02:51 +00003581<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003582
Chris Lattner261efe92003-11-25 01:02:51 +00003583<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003584<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +00003585 &lt;result&gt; = srem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +00003586</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003587
Chris Lattner261efe92003-11-25 01:02:51 +00003588<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003589<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3590 division of its two operands. This instruction can also take
3591 <a href="#t_vector">vector</a> versions of the values in which case the
3592 elements must be integers.</p>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +00003593
Chris Lattner261efe92003-11-25 01:02:51 +00003594<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003595<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003596 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3597 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003598
Chris Lattner261efe92003-11-25 01:02:51 +00003599<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +00003600<p>This instruction returns the <i>remainder</i> of a division (where the result
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003601 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3602 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3603 a value. For more information about the difference,
3604 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3605 Math Forum</a>. For a table of how this is implemented in various languages,
3606 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3607 Wikipedia: modulo operation</a>.</p>
3608
Chris Lattner5ec89832008-01-28 00:36:27 +00003609<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003610 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3611
Chris Lattner5ec89832008-01-28 00:36:27 +00003612<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003613 Overflow also leads to undefined behavior; this is a rare case, but can
3614 occur, for example, by taking the remainder of a 32-bit division of
3615 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3616 lets srem be implemented using instructions that return both the result of
3617 the division and the remainder.)</p>
3618
Chris Lattner261efe92003-11-25 01:02:51 +00003619<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003620<pre>
3621 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +00003622</pre>
3623
3624</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003625
Reid Spencer0a783f72006-11-02 01:53:59 +00003626<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003627<div class="doc_subsubsection">
3628 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3629
Reid Spencer0a783f72006-11-02 01:53:59 +00003630<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003631
Reid Spencer0a783f72006-11-02 01:53:59 +00003632<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003633<pre>
3634 &lt;result&gt; = frem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer0a783f72006-11-02 01:53:59 +00003635</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003636
Reid Spencer0a783f72006-11-02 01:53:59 +00003637<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003638<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3639 its two operands.</p>
3640
Reid Spencer0a783f72006-11-02 01:53:59 +00003641<h5>Arguments:</h5>
3642<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003643 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3644 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003645
Reid Spencer0a783f72006-11-02 01:53:59 +00003646<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003647<p>This instruction returns the <i>remainder</i> of a division. The remainder
3648 has the same sign as the dividend.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003649
Reid Spencer0a783f72006-11-02 01:53:59 +00003650<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003651<pre>
3652 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +00003653</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003654
Misha Brukman9d0919f2003-11-08 01:05:38 +00003655</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +00003656
Reid Spencer8e11bf82007-02-02 13:57:07 +00003657<!-- ======================================================================= -->
3658<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3659Operations</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003660
Reid Spencer8e11bf82007-02-02 13:57:07 +00003661<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003662
3663<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3664 program. They are generally very efficient instructions and can commonly be
3665 strength reduced from other instructions. They require two operands of the
3666 same type, execute an operation on them, and produce a single value. The
3667 resulting value is the same type as its operands.</p>
3668
Reid Spencer8e11bf82007-02-02 13:57:07 +00003669</div>
3670
Reid Spencer569f2fa2007-01-31 21:39:12 +00003671<!-- _______________________________________________________________________ -->
3672<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3673Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003674
Reid Spencer569f2fa2007-01-31 21:39:12 +00003675<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003676
Reid Spencer569f2fa2007-01-31 21:39:12 +00003677<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003678<pre>
3679 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003680</pre>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003681
Reid Spencer569f2fa2007-01-31 21:39:12 +00003682<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003683<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3684 a specified number of bits.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003685
Reid Spencer569f2fa2007-01-31 21:39:12 +00003686<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003687<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3688 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3689 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003690
Reid Spencer569f2fa2007-01-31 21:39:12 +00003691<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003692<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3693 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3694 is (statically or dynamically) negative or equal to or larger than the number
3695 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3696 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3697 shift amount in <tt>op2</tt>.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003698
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003699<h5>Example:</h5>
3700<pre>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003701 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3702 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3703 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003704 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +00003705 &lt;result&gt; = shl &lt;2 x i32&gt; &lt; i32 1, i32 1&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 2, i32 4&gt;</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003706</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003707
Reid Spencer569f2fa2007-01-31 21:39:12 +00003708</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003709
Reid Spencer569f2fa2007-01-31 21:39:12 +00003710<!-- _______________________________________________________________________ -->
3711<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3712Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003713
Reid Spencer569f2fa2007-01-31 21:39:12 +00003714<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003715
Reid Spencer569f2fa2007-01-31 21:39:12 +00003716<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003717<pre>
3718 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003719</pre>
3720
3721<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003722<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3723 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003724
3725<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003726<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003727 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3728 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003729
3730<h5>Semantics:</h5>
3731<p>This instruction always performs a logical shift right operation. The most
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003732 significant bits of the result will be filled with zero bits after the shift.
3733 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3734 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3735 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3736 shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003737
3738<h5>Example:</h5>
3739<pre>
3740 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3741 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3742 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3743 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003744 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +00003745 &lt;result&gt; = lshr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0x7FFFFFFF, i32 1&gt;</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003746</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003747
Reid Spencer569f2fa2007-01-31 21:39:12 +00003748</div>
3749
Reid Spencer8e11bf82007-02-02 13:57:07 +00003750<!-- _______________________________________________________________________ -->
Reid Spencer569f2fa2007-01-31 21:39:12 +00003751<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3752Instruction</a> </div>
3753<div class="doc_text">
3754
3755<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003756<pre>
3757 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003758</pre>
3759
3760<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003761<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3762 operand shifted to the right a specified number of bits with sign
3763 extension.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003764
3765<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003766<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003767 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3768 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003769
3770<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003771<p>This instruction always performs an arithmetic shift right operation, The
3772 most significant bits of the result will be filled with the sign bit
3773 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3774 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3775 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3776 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003777
3778<h5>Example:</h5>
3779<pre>
3780 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3781 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3782 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3783 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003784 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +00003785 &lt;result&gt; = ashr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 3&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 -1, i32 0&gt;</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003786</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003787
Reid Spencer569f2fa2007-01-31 21:39:12 +00003788</div>
3789
Chris Lattner00950542001-06-06 20:29:01 +00003790<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00003791<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3792Instruction</a> </div>
Chris Lattner5568e942008-05-20 20:48:21 +00003793
Misha Brukman9d0919f2003-11-08 01:05:38 +00003794<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003795
Chris Lattner00950542001-06-06 20:29:01 +00003796<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003797<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +00003798 &lt;result&gt; = and &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00003799</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003800
Chris Lattner00950542001-06-06 20:29:01 +00003801<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003802<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3803 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003804
Chris Lattner00950542001-06-06 20:29:01 +00003805<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003806<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003807 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3808 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003809
Chris Lattner00950542001-06-06 20:29:01 +00003810<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003811<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003812
Misha Brukman9d0919f2003-11-08 01:05:38 +00003813<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00003814 <tbody>
3815 <tr>
3816 <td>In0</td>
3817 <td>In1</td>
3818 <td>Out</td>
3819 </tr>
3820 <tr>
3821 <td>0</td>
3822 <td>0</td>
3823 <td>0</td>
3824 </tr>
3825 <tr>
3826 <td>0</td>
3827 <td>1</td>
3828 <td>0</td>
3829 </tr>
3830 <tr>
3831 <td>1</td>
3832 <td>0</td>
3833 <td>0</td>
3834 </tr>
3835 <tr>
3836 <td>1</td>
3837 <td>1</td>
3838 <td>1</td>
3839 </tr>
3840 </tbody>
3841</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003842
Chris Lattner00950542001-06-06 20:29:01 +00003843<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003844<pre>
3845 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003846 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3847 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +00003848</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003849</div>
Chris Lattner00950542001-06-06 20:29:01 +00003850<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00003851<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattner5568e942008-05-20 20:48:21 +00003852
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003853<div class="doc_text">
3854
3855<h5>Syntax:</h5>
3856<pre>
3857 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3858</pre>
3859
3860<h5>Overview:</h5>
3861<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3862 two operands.</p>
3863
3864<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003865<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003866 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3867 values. Both arguments must have identical types.</p>
3868
Chris Lattner00950542001-06-06 20:29:01 +00003869<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003870<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003871
Chris Lattner261efe92003-11-25 01:02:51 +00003872<table border="1" cellspacing="0" cellpadding="4">
3873 <tbody>
3874 <tr>
3875 <td>In0</td>
3876 <td>In1</td>
3877 <td>Out</td>
3878 </tr>
3879 <tr>
3880 <td>0</td>
3881 <td>0</td>
3882 <td>0</td>
3883 </tr>
3884 <tr>
3885 <td>0</td>
3886 <td>1</td>
3887 <td>1</td>
3888 </tr>
3889 <tr>
3890 <td>1</td>
3891 <td>0</td>
3892 <td>1</td>
3893 </tr>
3894 <tr>
3895 <td>1</td>
3896 <td>1</td>
3897 <td>1</td>
3898 </tr>
3899 </tbody>
3900</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003901
Chris Lattner00950542001-06-06 20:29:01 +00003902<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003903<pre>
3904 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003905 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3906 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +00003907</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003908
Misha Brukman9d0919f2003-11-08 01:05:38 +00003909</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003910
Chris Lattner00950542001-06-06 20:29:01 +00003911<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00003912<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3913Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003914
Misha Brukman9d0919f2003-11-08 01:05:38 +00003915<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003916
Chris Lattner00950542001-06-06 20:29:01 +00003917<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003918<pre>
3919 &lt;result&gt; = xor &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00003920</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003921
Chris Lattner00950542001-06-06 20:29:01 +00003922<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003923<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3924 its two operands. The <tt>xor</tt> is used to implement the "one's
3925 complement" operation, which is the "~" operator in C.</p>
3926
Chris Lattner00950542001-06-06 20:29:01 +00003927<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003928<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003929 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3930 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003931
Chris Lattner00950542001-06-06 20:29:01 +00003932<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003933<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003934
Chris Lattner261efe92003-11-25 01:02:51 +00003935<table border="1" cellspacing="0" cellpadding="4">
3936 <tbody>
3937 <tr>
3938 <td>In0</td>
3939 <td>In1</td>
3940 <td>Out</td>
3941 </tr>
3942 <tr>
3943 <td>0</td>
3944 <td>0</td>
3945 <td>0</td>
3946 </tr>
3947 <tr>
3948 <td>0</td>
3949 <td>1</td>
3950 <td>1</td>
3951 </tr>
3952 <tr>
3953 <td>1</td>
3954 <td>0</td>
3955 <td>1</td>
3956 </tr>
3957 <tr>
3958 <td>1</td>
3959 <td>1</td>
3960 <td>0</td>
3961 </tr>
3962 </tbody>
3963</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003964
Chris Lattner00950542001-06-06 20:29:01 +00003965<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003966<pre>
3967 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003968 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3969 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3970 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +00003971</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003972
Misha Brukman9d0919f2003-11-08 01:05:38 +00003973</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00003974
Chris Lattner00950542001-06-06 20:29:01 +00003975<!-- ======================================================================= -->
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003976<div class="doc_subsection">
Chris Lattner3df241e2006-04-08 23:07:04 +00003977 <a name="vectorops">Vector Operations</a>
3978</div>
3979
3980<div class="doc_text">
3981
3982<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003983 target-independent manner. These instructions cover the element-access and
3984 vector-specific operations needed to process vectors effectively. While LLVM
3985 does directly support these vector operations, many sophisticated algorithms
3986 will want to use target-specific intrinsics to take full advantage of a
3987 specific target.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00003988
3989</div>
3990
3991<!-- _______________________________________________________________________ -->
3992<div class="doc_subsubsection">
3993 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3994</div>
3995
3996<div class="doc_text">
3997
3998<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00003999<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004000 &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 +00004001</pre>
4002
4003<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004004<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4005 from a vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004006
4007
4008<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004009<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4010 of <a href="#t_vector">vector</a> type. The second operand is an index
4011 indicating the position from which to extract the element. The index may be
4012 a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004013
4014<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004015<p>The result is a scalar of the same type as the element type of
4016 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4017 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4018 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004019
4020<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004021<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +00004022 &lt;result&gt; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattner3df241e2006-04-08 23:07:04 +00004023</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +00004024
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004025</div>
Chris Lattner3df241e2006-04-08 23:07:04 +00004026
4027<!-- _______________________________________________________________________ -->
4028<div class="doc_subsubsection">
4029 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4030</div>
4031
4032<div class="doc_text">
4033
4034<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004035<pre>
Dan Gohmanf3480b92008-05-12 23:38:42 +00004036 &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 +00004037</pre>
4038
4039<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004040<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4041 vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004042
4043<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004044<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4045 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4046 whose type must equal the element type of the first operand. The third
4047 operand is an index indicating the position at which to insert the value.
4048 The index may be a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004049
4050<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004051<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4052 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4053 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4054 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004055
4056<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004057<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +00004058 &lt;result&gt; = 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 +00004059</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004060
Chris Lattner3df241e2006-04-08 23:07:04 +00004061</div>
4062
4063<!-- _______________________________________________________________________ -->
4064<div class="doc_subsubsection">
4065 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4066</div>
4067
4068<div class="doc_text">
4069
4070<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004071<pre>
Mon P Wangaeb06d22008-11-10 04:46:22 +00004072 &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;m x i32&gt; &lt;mask&gt; <i>; yields &lt;m x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +00004073</pre>
4074
4075<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004076<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4077 from two input vectors, returning a vector with the same element type as the
4078 input and length that is the same as the shuffle mask.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004079
4080<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004081<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4082 with types that match each other. The third argument is a shuffle mask whose
4083 element type is always 'i32'. The result of the instruction is a vector
4084 whose length is the same as the shuffle mask and whose element type is the
4085 same as the element type of the first two operands.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004086
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004087<p>The shuffle mask operand is required to be a constant vector with either
4088 constant integer or undef values.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004089
4090<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004091<p>The elements of the two input vectors are numbered from left to right across
4092 both of the vectors. The shuffle mask operand specifies, for each element of
4093 the result vector, which element of the two input vectors the result element
4094 gets. The element selector may be undef (meaning "don't care") and the
4095 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004096
4097<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004098<pre>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004099 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen6f1cc772007-04-22 01:17:39 +00004100 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004101 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Reid Spencerca86e162006-12-31 07:07:53 +00004102 &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.
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004103 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wangaeb06d22008-11-10 04:46:22 +00004104 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004105 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wangaeb06d22008-11-10 04:46:22 +00004106 &lt;8 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 &gt; <i>; yields &lt;8 x i32&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +00004107</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +00004108
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004109</div>
Tanya Lattner09474292006-04-14 19:24:33 +00004110
Chris Lattner3df241e2006-04-08 23:07:04 +00004111<!-- ======================================================================= -->
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004112<div class="doc_subsection">
Dan Gohmana334d5f2008-05-12 23:51:09 +00004113 <a name="aggregateops">Aggregate Operations</a>
4114</div>
4115
4116<div class="doc_text">
4117
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004118<p>LLVM supports several instructions for working with
4119 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004120
4121</div>
4122
4123<!-- _______________________________________________________________________ -->
4124<div class="doc_subsubsection">
4125 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4126</div>
4127
4128<div class="doc_text">
4129
4130<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004131<pre>
4132 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4133</pre>
4134
4135<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004136<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4137 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004138
4139<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004140<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004141 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4142 <a href="#t_array">array</a> type. The operands are constant indices to
4143 specify which value to extract in a similar manner as indices in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004144 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004145
4146<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004147<p>The result is the value at the position in the aggregate specified by the
4148 index operands.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004149
4150<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004151<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +00004152 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004153</pre>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004154
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004155</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004156
4157<!-- _______________________________________________________________________ -->
4158<div class="doc_subsubsection">
4159 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4160</div>
4161
4162<div class="doc_text">
4163
4164<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004165<pre>
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +00004166 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, &lt;idx&gt; <i>; yields &lt;aggregate type&gt;</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004167</pre>
4168
4169<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004170<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4171 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004172
4173<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004174<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004175 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4176 <a href="#t_array">array</a> type. The second operand is a first-class
4177 value to insert. The following operands are constant indices indicating
4178 the position at which to insert the value in a similar manner as indices in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004179 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
4180 value to insert must have the same type as the value identified by the
4181 indices.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004182
4183<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004184<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4185 that of <tt>val</tt> except that the value at the position specified by the
4186 indices is that of <tt>elt</tt>.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004187
4188<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004189<pre>
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +00004190 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4191 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004192</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004193
Dan Gohmana334d5f2008-05-12 23:51:09 +00004194</div>
4195
4196
4197<!-- ======================================================================= -->
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004198<div class="doc_subsection">
Chris Lattner884a9702006-08-15 00:45:58 +00004199 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004200</div>
4201
Misha Brukman9d0919f2003-11-08 01:05:38 +00004202<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00004203
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004204<p>A key design point of an SSA-based representation is how it represents
4205 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez2fee2942009-10-26 23:44:29 +00004206 very simple. This section describes how to read, write, and allocate
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004207 memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004208
Misha Brukman9d0919f2003-11-08 01:05:38 +00004209</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004210
Chris Lattner00950542001-06-06 20:29:01 +00004211<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +00004212<div class="doc_subsubsection">
Chris Lattner2cbdc452005-11-06 08:02:57 +00004213 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4214</div>
4215
Misha Brukman9d0919f2003-11-08 01:05:38 +00004216<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00004217
Chris Lattner00950542001-06-06 20:29:01 +00004218<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004219<pre>
Dan Gohmanf75a7d32010-05-28 01:14:11 +00004220 &lt;result&gt; = alloca &lt;type&gt;[, &lt;ty&gt; &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +00004221</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004222
Chris Lattner00950542001-06-06 20:29:01 +00004223<h5>Overview:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00004224<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004225 currently executing function, to be automatically released when this function
4226 returns to its caller. The object is always allocated in the generic address
4227 space (address space zero).</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004228
Chris Lattner00950542001-06-06 20:29:01 +00004229<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004230<p>The '<tt>alloca</tt>' instruction
4231 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4232 runtime stack, returning a pointer of the appropriate type to the program.
4233 If "NumElements" is specified, it is the number of elements allocated,
4234 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4235 specified, the value result of the allocation is guaranteed to be aligned to
4236 at least that boundary. If not specified, or if zero, the target can choose
4237 to align the allocation on any convenient boundary compatible with the
4238 type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004239
Misha Brukman9d0919f2003-11-08 01:05:38 +00004240<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004241
Chris Lattner00950542001-06-06 20:29:01 +00004242<h5>Semantics:</h5>
Bill Wendling871eb0a2009-05-08 20:49:29 +00004243<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004244 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4245 memory is automatically released when the function returns. The
4246 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4247 variables that must have an address available. When the function returns
4248 (either with the <tt><a href="#i_ret">ret</a></tt>
4249 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4250 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004251
Chris Lattner00950542001-06-06 20:29:01 +00004252<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004253<pre>
Dan Gohman81e21672009-01-04 23:49:44 +00004254 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4255 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4256 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4257 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +00004258</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004259
Misha Brukman9d0919f2003-11-08 01:05:38 +00004260</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004261
Chris Lattner00950542001-06-06 20:29:01 +00004262<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00004263<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
4264Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004265
Misha Brukman9d0919f2003-11-08 01:05:38 +00004266<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004267
Chris Lattner2b7d3202002-05-06 03:03:22 +00004268<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004269<pre>
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004270 &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4271 &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4272 !&lt;index&gt; = !{ i32 1 }
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004273</pre>
4274
Chris Lattner2b7d3202002-05-06 03:03:22 +00004275<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004276<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004277
Chris Lattner2b7d3202002-05-06 03:03:22 +00004278<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004279<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4280 from which to load. The pointer must point to
4281 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4282 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00004283 number or order of execution of this <tt>load</tt> with other <a
4284 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004285
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004286<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004287 operation (that is, the alignment of the memory address). A value of 0 or an
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004288 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004289 alignment for the target. It is the responsibility of the code emitter to
4290 ensure that the alignment information is correct. Overestimating the
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004291 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004292 produce less efficient code. An alignment of 1 is always safe.</p>
4293
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004294<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4295 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004296 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004297 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4298 and code generator that this load is not expected to be reused in the cache.
4299 The code generator may select special instructions to save cache bandwidth,
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004300 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +00004301
Chris Lattner2b7d3202002-05-06 03:03:22 +00004302<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004303<p>The location of memory pointed to is loaded. If the value being loaded is of
4304 scalar type then the number of bytes read does not exceed the minimum number
4305 of bytes needed to hold all bits of the type. For example, loading an
4306 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4307 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4308 is undefined if the value was not originally written using a store of the
4309 same type.</p>
4310
Chris Lattner2b7d3202002-05-06 03:03:22 +00004311<h5>Examples:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004312<pre>
4313 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4314 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerca86e162006-12-31 07:07:53 +00004315 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00004316</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004317
Misha Brukman9d0919f2003-11-08 01:05:38 +00004318</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004319
Chris Lattner2b7d3202002-05-06 03:03:22 +00004320<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00004321<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4322Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004323
Reid Spencer035ab572006-11-09 21:18:01 +00004324<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004325
Chris Lattner2b7d3202002-05-06 03:03:22 +00004326<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004327<pre>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004328 store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;] <i>; yields {void}</i>
4329 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;] <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00004330</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004331
Chris Lattner2b7d3202002-05-06 03:03:22 +00004332<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004333<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004334
Chris Lattner2b7d3202002-05-06 03:03:22 +00004335<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004336<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4337 and an address at which to store it. The type of the
4338 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4339 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00004340 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
4341 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4342 order of execution of this <tt>store</tt> with other <a
4343 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004344
4345<p>The optional constant "align" argument specifies the alignment of the
4346 operation (that is, the alignment of the memory address). A value of 0 or an
4347 omitted "align" argument means that the operation has the preferential
4348 alignment for the target. It is the responsibility of the code emitter to
4349 ensure that the alignment information is correct. Overestimating the
4350 alignment results in an undefined behavior. Underestimating the alignment may
4351 produce less efficient code. An alignment of 1 is always safe.</p>
4352
David Greene8939b0d2010-02-16 20:50:18 +00004353<p>The optional !nontemporal metadata must reference a single metatadata
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004354 name &lt;index&gt; corresponding to a metadata node with one i32 entry of
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004355 value 1. The existence of the !nontemporal metatadata on the
David Greene8939b0d2010-02-16 20:50:18 +00004356 instruction tells the optimizer and code generator that this load is
4357 not expected to be reused in the cache. The code generator may
4358 select special instructions to save cache bandwidth, such as the
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004359 MOVNT instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +00004360
4361
Chris Lattner261efe92003-11-25 01:02:51 +00004362<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004363<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4364 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4365 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4366 does not exceed the minimum number of bytes needed to hold all bits of the
4367 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4368 writing a value of a type like <tt>i20</tt> with a size that is not an
4369 integral number of bytes, it is unspecified what happens to the extra bits
4370 that do not belong to the type, but they will typically be overwritten.</p>
4371
Chris Lattner2b7d3202002-05-06 03:03:22 +00004372<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004373<pre>
4374 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8c6c72d2007-10-22 05:10:05 +00004375 store i32 3, i32* %ptr <i>; yields {void}</i>
4376 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00004377</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004378
Reid Spencer47ce1792006-11-09 21:15:49 +00004379</div>
4380
Chris Lattner2b7d3202002-05-06 03:03:22 +00004381<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004382<div class="doc_subsubsection">
4383 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4384</div>
4385
Misha Brukman9d0919f2003-11-08 01:05:38 +00004386<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004387
Chris Lattner7faa8832002-04-14 06:13:44 +00004388<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004389<pre>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004390 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmandd8004d2009-07-27 21:53:46 +00004391 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004392</pre>
4393
Chris Lattner7faa8832002-04-14 06:13:44 +00004394<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004395<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004396 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4397 It performs address calculation only and does not access memory.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004398
Chris Lattner7faa8832002-04-14 06:13:44 +00004399<h5>Arguments:</h5>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004400<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnerc8eef442009-07-29 06:44:13 +00004401 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004402 elements of the aggregate object are indexed. The interpretation of each
4403 index is dependent on the type being indexed into. The first index always
4404 indexes the pointer value given as the first argument, the second index
4405 indexes a value of the type pointed to (not necessarily the value directly
4406 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004407 indexed into must be a pointer value, subsequent types can be arrays,
4408 vectors, structs and unions. Note that subsequent types being indexed into
4409 can never be pointers, since that would require loading the pointer before
4410 continuing calculation.</p>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004411
4412<p>The type of each index argument depends on the type it is indexing into.
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004413 When indexing into a (optionally packed) structure or union, only <tt>i32</tt>
4414 integer <b>constants</b> are allowed. When indexing into an array, pointer
4415 or vector, integers of any width are allowed, and they are not required to be
Chris Lattnerc8eef442009-07-29 06:44:13 +00004416 constant.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004417
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004418<p>For example, let's consider a C code fragment and how it gets compiled to
4419 LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004420
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004421<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004422struct RT {
4423 char A;
Chris Lattnercabc8462007-05-29 15:43:56 +00004424 int B[10][20];
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004425 char C;
4426};
4427struct ST {
Chris Lattnercabc8462007-05-29 15:43:56 +00004428 int X;
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004429 double Y;
4430 struct RT Z;
4431};
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004432
Chris Lattnercabc8462007-05-29 15:43:56 +00004433int *foo(struct ST *s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004434 return &amp;s[1].Z.B[5][13];
4435}
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004436</pre>
4437
Misha Brukman9d0919f2003-11-08 01:05:38 +00004438<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004439
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004440<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +00004441%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4442%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004443
Dan Gohman4df605b2009-07-25 02:23:48 +00004444define i32* @foo(%ST* %s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004445entry:
4446 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4447 ret i32* %reg
4448}
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004449</pre>
4450
Chris Lattner7faa8832002-04-14 06:13:44 +00004451<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004452<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004453 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4454 }</tt>' type, a structure. The second index indexes into the third element
4455 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4456 i8 }</tt>' type, another structure. The third index indexes into the second
4457 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4458 array. The two dimensions of the array are subscripted into, yielding an
4459 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4460 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004461
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004462<p>Note that it is perfectly legal to index partially through a structure,
4463 returning a pointer to an inner element. Because of this, the LLVM code for
4464 the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004465
4466<pre>
Dan Gohman4df605b2009-07-25 02:23:48 +00004467 define i32* @foo(%ST* %s) {
Reid Spencerca86e162006-12-31 07:07:53 +00004468 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00004469 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4470 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
Reid Spencerca86e162006-12-31 07:07:53 +00004471 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4472 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4473 ret i32* %t5
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004474 }
Chris Lattner6536cfe2002-05-06 22:08:29 +00004475</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +00004476
Dan Gohmandd8004d2009-07-27 21:53:46 +00004477<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman27ef9972010-04-23 15:23:32 +00004478 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
4479 base pointer is not an <i>in bounds</i> address of an allocated object,
4480 or if any of the addresses that would be formed by successive addition of
4481 the offsets implied by the indices to the base address with infinitely
4482 precise arithmetic are not an <i>in bounds</i> address of that allocated
4483 object. The <i>in bounds</i> addresses for an allocated object are all
4484 the addresses that point into the object, plus the address one byte past
4485 the end.</p>
Dan Gohmandd8004d2009-07-27 21:53:46 +00004486
4487<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4488 the base address with silently-wrapping two's complement arithmetic, and
4489 the result value of the <tt>getelementptr</tt> may be outside the object
4490 pointed to by the base pointer. The result value may not necessarily be
4491 used to access memory though, even if it happens to point into allocated
4492 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4493 section for more information.</p>
4494
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004495<p>The getelementptr instruction is often confusing. For some more insight into
4496 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner884a9702006-08-15 00:45:58 +00004497
Chris Lattner7faa8832002-04-14 06:13:44 +00004498<h5>Example:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004499<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004500 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004501 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4502 <i>; yields i8*:vptr</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +00004503 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004504 <i>; yields i8*:eptr</i>
4505 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta9f805c22009-04-25 07:27:44 +00004506 <i>; yields i32*:iptr</i>
Sanjiv Gupta16ffa802009-04-24 16:38:13 +00004507 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004508</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004509
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004510</div>
Reid Spencer47ce1792006-11-09 21:15:49 +00004511
Chris Lattner00950542001-06-06 20:29:01 +00004512<!-- ======================================================================= -->
Reid Spencer2fd21e62006-11-08 01:18:52 +00004513<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004514</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004515
Misha Brukman9d0919f2003-11-08 01:05:38 +00004516<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004517
Reid Spencer2fd21e62006-11-08 01:18:52 +00004518<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004519 which all take a single operand and a type. They perform various bit
4520 conversions on the operand.</p>
4521
Misha Brukman9d0919f2003-11-08 01:05:38 +00004522</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004523
Chris Lattner6536cfe2002-05-06 22:08:29 +00004524<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +00004525<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004526 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4527</div>
4528<div class="doc_text">
4529
4530<h5>Syntax:</h5>
4531<pre>
4532 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4533</pre>
4534
4535<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004536<p>The '<tt>trunc</tt>' instruction truncates its operand to the
4537 type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004538
4539<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004540<p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4541 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4542 size and type of the result, which must be
4543 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4544 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4545 allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004546
4547<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004548<p>The '<tt>trunc</tt>' instruction truncates the high order bits
4549 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4550 source size must be larger than the destination size, <tt>trunc</tt> cannot
4551 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004552
4553<h5>Example:</h5>
4554<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004555 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00004556 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004557 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004558</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004559
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004560</div>
4561
4562<!-- _______________________________________________________________________ -->
4563<div class="doc_subsubsection">
4564 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4565</div>
4566<div class="doc_text">
4567
4568<h5>Syntax:</h5>
4569<pre>
4570 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4571</pre>
4572
4573<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004574<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004575 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004576
4577
4578<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004579<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004580 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4581 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004582 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004583 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004584
4585<h5>Semantics:</h5>
4586<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004587 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004588
Reid Spencerb5929522007-01-12 15:46:11 +00004589<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004590
4591<h5>Example:</h5>
4592<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004593 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00004594 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004595</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004596
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004597</div>
4598
4599<!-- _______________________________________________________________________ -->
4600<div class="doc_subsubsection">
4601 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4602</div>
4603<div class="doc_text">
4604
4605<h5>Syntax:</h5>
4606<pre>
4607 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4608</pre>
4609
4610<h5>Overview:</h5>
4611<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4612
4613<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004614<p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004615 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4616 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004617 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004618 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004619
4620<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004621<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4622 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4623 of the type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004624
Reid Spencerc78f3372007-01-12 03:35:51 +00004625<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004626
4627<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004628<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004629 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00004630 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004631</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004632
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004633</div>
4634
4635<!-- _______________________________________________________________________ -->
4636<div class="doc_subsubsection">
Reid Spencer3fa91b02006-11-09 21:48:10 +00004637 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4638</div>
4639
4640<div class="doc_text">
4641
4642<h5>Syntax:</h5>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004643<pre>
4644 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4645</pre>
4646
4647<h5>Overview:</h5>
4648<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004649 <tt>ty2</tt>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004650
4651<h5>Arguments:</h5>
4652<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004653 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4654 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004655 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004656 <i>no-op cast</i>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004657
4658<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004659<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004660 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004661 <a href="#t_floating">floating point</a> type. If the value cannot fit
4662 within the destination type, <tt>ty2</tt>, then the results are
4663 undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004664
4665<h5>Example:</h5>
4666<pre>
4667 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4668 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4669</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004670
Reid Spencer3fa91b02006-11-09 21:48:10 +00004671</div>
4672
4673<!-- _______________________________________________________________________ -->
4674<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004675 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4676</div>
4677<div class="doc_text">
4678
4679<h5>Syntax:</h5>
4680<pre>
4681 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4682</pre>
4683
4684<h5>Overview:</h5>
4685<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004686 floating point value.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004687
4688<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004689<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004690 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4691 a <a href="#t_floating">floating point</a> type to cast it to. The source
4692 type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004693
4694<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +00004695<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004696 <a href="#t_floating">floating point</a> type to a larger
4697 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4698 used to make a <i>no-op cast</i> because it always changes bits. Use
4699 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004700
4701<h5>Example:</h5>
4702<pre>
4703 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4704 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4705</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004706
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004707</div>
4708
4709<!-- _______________________________________________________________________ -->
4710<div class="doc_subsubsection">
Reid Spencer24d6da52007-01-21 00:29:26 +00004711 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004712</div>
4713<div class="doc_text">
4714
4715<h5>Syntax:</h5>
4716<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +00004717 &lt;result&gt; = fptoui &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004718</pre>
4719
4720<h5>Overview:</h5>
Reid Spencer1539a1c2007-07-31 14:40:14 +00004721<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004722 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004723
4724<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004725<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4726 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4727 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4728 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4729 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004730
4731<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004732<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004733 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4734 towards zero) unsigned integer value. If the value cannot fit
4735 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004736
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004737<h5>Example:</h5>
4738<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +00004739 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner88519042007-09-22 03:17:52 +00004740 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004741 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004742</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004743
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004744</div>
4745
4746<!-- _______________________________________________________________________ -->
4747<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +00004748 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004749</div>
4750<div class="doc_text">
4751
4752<h5>Syntax:</h5>
4753<pre>
Reid Spencerd4448792006-11-09 23:03:26 +00004754 &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 +00004755</pre>
4756
4757<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004758<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004759 <a href="#t_floating">floating point</a> <tt>value</tt> to
4760 type <tt>ty2</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004761
Chris Lattner6536cfe2002-05-06 22:08:29 +00004762<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004763<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4764 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4765 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4766 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4767 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004768
Chris Lattner6536cfe2002-05-06 22:08:29 +00004769<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004770<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004771 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4772 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4773 the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004774
Chris Lattner33ba0d92001-07-09 00:26:23 +00004775<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004776<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +00004777 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner88519042007-09-22 03:17:52 +00004778 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004779 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004780</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004781
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004782</div>
4783
4784<!-- _______________________________________________________________________ -->
4785<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +00004786 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004787</div>
4788<div class="doc_text">
4789
4790<h5>Syntax:</h5>
4791<pre>
Reid Spencerd4448792006-11-09 23:03:26 +00004792 &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 +00004793</pre>
4794
4795<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +00004796<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004797 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004798
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004799<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +00004800<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004801 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4802 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4803 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4804 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004805
4806<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +00004807<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004808 integer quantity and converts it to the corresponding floating point
4809 value. If the value cannot fit in the floating point value, the results are
4810 undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004811
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004812<h5>Example:</h5>
4813<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004814 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +00004815 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004816</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004817
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004818</div>
4819
4820<!-- _______________________________________________________________________ -->
4821<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +00004822 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004823</div>
4824<div class="doc_text">
4825
4826<h5>Syntax:</h5>
4827<pre>
Reid Spencerd4448792006-11-09 23:03:26 +00004828 &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 +00004829</pre>
4830
4831<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004832<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4833 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004834
4835<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +00004836<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004837 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4838 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4839 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4840 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004841
4842<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004843<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4844 quantity and converts it to the corresponding floating point value. If the
4845 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004846
4847<h5>Example:</h5>
4848<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004849 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +00004850 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004851</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004852
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004853</div>
4854
4855<!-- _______________________________________________________________________ -->
4856<div class="doc_subsubsection">
Reid Spencer72679252006-11-11 21:00:47 +00004857 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4858</div>
4859<div class="doc_text">
4860
4861<h5>Syntax:</h5>
4862<pre>
4863 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4864</pre>
4865
4866<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004867<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4868 the integer type <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004869
4870<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004871<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4872 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4873 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004874
4875<h5>Semantics:</h5>
4876<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004877 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4878 truncating or zero extending that value to the size of the integer type. If
4879 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4880 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4881 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4882 change.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004883
4884<h5>Example:</h5>
4885<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +00004886 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4887 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +00004888</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004889
Reid Spencer72679252006-11-11 21:00:47 +00004890</div>
4891
4892<!-- _______________________________________________________________________ -->
4893<div class="doc_subsubsection">
4894 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4895</div>
4896<div class="doc_text">
4897
4898<h5>Syntax:</h5>
4899<pre>
4900 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4901</pre>
4902
4903<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004904<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4905 pointer type, <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004906
4907<h5>Arguments:</h5>
Duncan Sands8036ca42007-03-30 12:22:09 +00004908<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004909 value to cast, and a type to cast it to, which must be a
4910 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004911
4912<h5>Semantics:</h5>
4913<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004914 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4915 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4916 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4917 than the size of a pointer then a zero extension is done. If they are the
4918 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencer72679252006-11-11 21:00:47 +00004919
4920<h5>Example:</h5>
4921<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +00004922 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004923 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4924 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +00004925</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004926
Reid Spencer72679252006-11-11 21:00:47 +00004927</div>
4928
4929<!-- _______________________________________________________________________ -->
4930<div class="doc_subsubsection">
Reid Spencer5c0ef472006-11-11 23:08:07 +00004931 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004932</div>
4933<div class="doc_text">
4934
4935<h5>Syntax:</h5>
4936<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +00004937 &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 +00004938</pre>
4939
4940<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +00004941<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004942 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004943
4944<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004945<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4946 non-aggregate first class value, and a type to cast it to, which must also be
4947 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4948 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4949 identical. If the source type is a pointer, the destination type must also be
4950 a pointer. This instruction supports bitwise conversion of vectors to
4951 integers and to vectors of other types (as long as they have the same
4952 size).</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004953
4954<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +00004955<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004956 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4957 this conversion. The conversion is done as if the <tt>value</tt> had been
4958 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4959 be converted to other pointer types with this instruction. To convert
4960 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4961 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004962
4963<h5>Example:</h5>
4964<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +00004965 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerca86e162006-12-31 07:07:53 +00004966 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004967 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +00004968</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004969
Misha Brukman9d0919f2003-11-08 01:05:38 +00004970</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004971
Reid Spencer2fd21e62006-11-08 01:18:52 +00004972<!-- ======================================================================= -->
4973<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004974
Reid Spencer2fd21e62006-11-08 01:18:52 +00004975<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004976
4977<p>The instructions in this category are the "miscellaneous" instructions, which
4978 defy better classification.</p>
4979
Reid Spencer2fd21e62006-11-08 01:18:52 +00004980</div>
Reid Spencerf3a70a62006-11-18 21:50:54 +00004981
4982<!-- _______________________________________________________________________ -->
4983<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4984</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004985
Reid Spencerf3a70a62006-11-18 21:50:54 +00004986<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004987
Reid Spencerf3a70a62006-11-18 21:50:54 +00004988<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004989<pre>
4990 &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +00004991</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004992
Reid Spencerf3a70a62006-11-18 21:50:54 +00004993<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004994<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4995 boolean values based on comparison of its two integer, integer vector, or
4996 pointer operands.</p>
4997
Reid Spencerf3a70a62006-11-18 21:50:54 +00004998<h5>Arguments:</h5>
4999<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005000 the condition code indicating the kind of comparison to perform. It is not a
5001 value, just a keyword. The possible condition code are:</p>
5002
Reid Spencerf3a70a62006-11-18 21:50:54 +00005003<ol>
5004 <li><tt>eq</tt>: equal</li>
5005 <li><tt>ne</tt>: not equal </li>
5006 <li><tt>ugt</tt>: unsigned greater than</li>
5007 <li><tt>uge</tt>: unsigned greater or equal</li>
5008 <li><tt>ult</tt>: unsigned less than</li>
5009 <li><tt>ule</tt>: unsigned less or equal</li>
5010 <li><tt>sgt</tt>: signed greater than</li>
5011 <li><tt>sge</tt>: signed greater or equal</li>
5012 <li><tt>slt</tt>: signed less than</li>
5013 <li><tt>sle</tt>: signed less or equal</li>
5014</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005015
Chris Lattner3b19d652007-01-15 01:54:13 +00005016<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005017 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5018 typed. They must also be identical types.</p>
5019
Reid Spencerf3a70a62006-11-18 21:50:54 +00005020<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005021<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5022 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewyckyec38da42009-09-27 00:45:11 +00005023 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005024 result, as follows:</p>
5025
Reid Spencerf3a70a62006-11-18 21:50:54 +00005026<ol>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005027 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005028 <tt>false</tt> otherwise. No sign interpretation is necessary or
5029 performed.</li>
5030
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005031 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005032 <tt>false</tt> otherwise. No sign interpretation is necessary or
5033 performed.</li>
5034
Reid Spencerf3a70a62006-11-18 21:50:54 +00005035 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005036 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5037
Reid Spencerf3a70a62006-11-18 21:50:54 +00005038 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005039 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5040 to <tt>op2</tt>.</li>
5041
Reid Spencerf3a70a62006-11-18 21:50:54 +00005042 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005043 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5044
Reid Spencerf3a70a62006-11-18 21:50:54 +00005045 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005046 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5047
Reid Spencerf3a70a62006-11-18 21:50:54 +00005048 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005049 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5050
Reid Spencerf3a70a62006-11-18 21:50:54 +00005051 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005052 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5053 to <tt>op2</tt>.</li>
5054
Reid Spencerf3a70a62006-11-18 21:50:54 +00005055 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005056 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5057
Reid Spencerf3a70a62006-11-18 21:50:54 +00005058 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005059 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005060</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005061
Reid Spencerf3a70a62006-11-18 21:50:54 +00005062<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005063 values are compared as if they were integers.</p>
5064
5065<p>If the operands are integer vectors, then they are compared element by
5066 element. The result is an <tt>i1</tt> vector with the same number of elements
5067 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005068
5069<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005070<pre>
5071 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerca86e162006-12-31 07:07:53 +00005072 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
5073 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
5074 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5075 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5076 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005077</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +00005078
5079<p>Note that the code generator does not yet support vector types with
5080 the <tt>icmp</tt> instruction.</p>
5081
Reid Spencerf3a70a62006-11-18 21:50:54 +00005082</div>
5083
5084<!-- _______________________________________________________________________ -->
5085<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5086</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005087
Reid Spencerf3a70a62006-11-18 21:50:54 +00005088<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005089
Reid Spencerf3a70a62006-11-18 21:50:54 +00005090<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005091<pre>
5092 &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005093</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005094
Reid Spencerf3a70a62006-11-18 21:50:54 +00005095<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005096<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5097 values based on comparison of its operands.</p>
5098
5099<p>If the operands are floating point scalars, then the result type is a boolean
Nick Lewyckyec38da42009-09-27 00:45:11 +00005100(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005101
5102<p>If the operands are floating point vectors, then the result type is a vector
5103 of boolean with the same number of elements as the operands being
5104 compared.</p>
5105
Reid Spencerf3a70a62006-11-18 21:50:54 +00005106<h5>Arguments:</h5>
5107<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005108 the condition code indicating the kind of comparison to perform. It is not a
5109 value, just a keyword. The possible condition code are:</p>
5110
Reid Spencerf3a70a62006-11-18 21:50:54 +00005111<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +00005112 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005113 <li><tt>oeq</tt>: ordered and equal</li>
5114 <li><tt>ogt</tt>: ordered and greater than </li>
5115 <li><tt>oge</tt>: ordered and greater than or equal</li>
5116 <li><tt>olt</tt>: ordered and less than </li>
5117 <li><tt>ole</tt>: ordered and less than or equal</li>
5118 <li><tt>one</tt>: ordered and not equal</li>
5119 <li><tt>ord</tt>: ordered (no nans)</li>
5120 <li><tt>ueq</tt>: unordered or equal</li>
5121 <li><tt>ugt</tt>: unordered or greater than </li>
5122 <li><tt>uge</tt>: unordered or greater than or equal</li>
5123 <li><tt>ult</tt>: unordered or less than </li>
5124 <li><tt>ule</tt>: unordered or less than or equal</li>
5125 <li><tt>une</tt>: unordered or not equal</li>
5126 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +00005127 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005128</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005129
Jeff Cohenb627eab2007-04-29 01:07:00 +00005130<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005131 <i>unordered</i> means that either operand may be a QNAN.</p>
5132
5133<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5134 a <a href="#t_floating">floating point</a> type or
5135 a <a href="#t_vector">vector</a> of floating point type. They must have
5136 identical types.</p>
5137
Reid Spencerf3a70a62006-11-18 21:50:54 +00005138<h5>Semantics:</h5>
Gabor Greiffb224a22008-08-07 21:46:00 +00005139<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005140 according to the condition code given as <tt>cond</tt>. If the operands are
5141 vectors, then the vectors are compared element by element. Each comparison
Nick Lewyckyec38da42009-09-27 00:45:11 +00005142 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005143 follows:</p>
5144
Reid Spencerf3a70a62006-11-18 21:50:54 +00005145<ol>
5146 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005147
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005148 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005149 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5150
Reid Spencerb7f26282006-11-19 03:00:14 +00005151 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00005152 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005153
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005154 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005155 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5156
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005157 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005158 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5159
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005160 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005161 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5162
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005163 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005164 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5165
Reid Spencerb7f26282006-11-19 03:00:14 +00005166 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005167
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005168 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005169 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5170
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005171 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005172 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5173
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005174 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005175 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5176
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005177 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005178 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5179
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005180 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005181 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5182
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005183 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005184 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5185
Reid Spencerb7f26282006-11-19 03:00:14 +00005186 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005187
Reid Spencerf3a70a62006-11-18 21:50:54 +00005188 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5189</ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005190
5191<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005192<pre>
5193 &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanf72fb672008-09-09 01:02:47 +00005194 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5195 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5196 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005197</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +00005198
5199<p>Note that the code generator does not yet support vector types with
5200 the <tt>fcmp</tt> instruction.</p>
5201
Reid Spencerf3a70a62006-11-18 21:50:54 +00005202</div>
5203
Reid Spencer2fd21e62006-11-08 01:18:52 +00005204<!-- _______________________________________________________________________ -->
Nate Begemanac80ade2008-05-12 19:01:56 +00005205<div class="doc_subsubsection">
Chris Lattner5568e942008-05-20 20:48:21 +00005206 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5207</div>
5208
Reid Spencer2fd21e62006-11-08 01:18:52 +00005209<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00005210
Reid Spencer2fd21e62006-11-08 01:18:52 +00005211<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005212<pre>
5213 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
5214</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00005215
Reid Spencer2fd21e62006-11-08 01:18:52 +00005216<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005217<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
5218 SSA graph representing the function.</p>
5219
Reid Spencer2fd21e62006-11-08 01:18:52 +00005220<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005221<p>The type of the incoming values is specified with the first type field. After
5222 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5223 one pair for each predecessor basic block of the current block. Only values
5224 of <a href="#t_firstclass">first class</a> type may be used as the value
5225 arguments to the PHI node. Only labels may be used as the label
5226 arguments.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00005227
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005228<p>There must be no non-phi instructions between the start of a basic block and
5229 the PHI instructions: i.e. PHI instructions must be first in a basic
5230 block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00005231
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005232<p>For the purposes of the SSA form, the use of each incoming value is deemed to
5233 occur on the edge from the corresponding predecessor block to the current
5234 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5235 value on the same edge).</p>
Jay Foadd2449092009-06-03 10:20:10 +00005236
Reid Spencer2fd21e62006-11-08 01:18:52 +00005237<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005238<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005239 specified by the pair corresponding to the predecessor basic block that
5240 executed just prior to the current block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00005241
Reid Spencer2fd21e62006-11-08 01:18:52 +00005242<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00005243<pre>
5244Loop: ; Infinite loop that counts from 0 on up...
5245 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5246 %nextindvar = add i32 %indvar, 1
5247 br label %Loop
5248</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005249
Reid Spencer2fd21e62006-11-08 01:18:52 +00005250</div>
5251
Chris Lattnercc37aae2004-03-12 05:50:16 +00005252<!-- _______________________________________________________________________ -->
5253<div class="doc_subsubsection">
5254 <a name="i_select">'<tt>select</tt>' Instruction</a>
5255</div>
5256
5257<div class="doc_text">
5258
5259<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005260<pre>
Dan Gohmanf72fb672008-09-09 01:02:47 +00005261 &lt;result&gt; = select <i>selty</i> &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
5262
Dan Gohman0e451ce2008-10-14 16:51:45 +00005263 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnercc37aae2004-03-12 05:50:16 +00005264</pre>
5265
5266<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005267<p>The '<tt>select</tt>' instruction is used to choose one value based on a
5268 condition, without branching.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005269
5270
5271<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005272<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5273 values indicating the condition, and two values of the
5274 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5275 vectors and the condition is a scalar, then entire vectors are selected, not
5276 individual elements.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005277
5278<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005279<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5280 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005281
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005282<p>If the condition is a vector of i1, then the value arguments must be vectors
5283 of the same size, and the selection is done element by element.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005284
5285<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005286<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +00005287 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005288</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +00005289
5290<p>Note that the code generator does not yet support conditions
5291 with vector type.</p>
5292
Chris Lattnercc37aae2004-03-12 05:50:16 +00005293</div>
5294
Robert Bocchino05ccd702006-01-15 20:48:27 +00005295<!-- _______________________________________________________________________ -->
5296<div class="doc_subsubsection">
Chris Lattner2bff5242005-05-06 05:47:36 +00005297 <a name="i_call">'<tt>call</tt>' Instruction</a>
5298</div>
5299
Misha Brukman9d0919f2003-11-08 01:05:38 +00005300<div class="doc_text">
Chris Lattner2bff5242005-05-06 05:47:36 +00005301
Chris Lattner00950542001-06-06 20:29:01 +00005302<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00005303<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +00005304 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Chris Lattner2bff5242005-05-06 05:47:36 +00005305</pre>
5306
Chris Lattner00950542001-06-06 20:29:01 +00005307<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00005308<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00005309
Chris Lattner00950542001-06-06 20:29:01 +00005310<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00005311<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00005312
Chris Lattner6536cfe2002-05-06 22:08:29 +00005313<ol>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005314 <li>The optional "tail" marker indicates that the callee function does not
5315 access any allocas or varargs in the caller. Note that calls may be
5316 marked "tail" even if they do not occur before
5317 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5318 present, the function call is eligible for tail call optimization,
5319 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Chengdc444e92010-03-08 21:05:02 +00005320 optimized into a jump</a>. The code generator may optimize calls marked
5321 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5322 sibling call optimization</a> when the caller and callee have
5323 matching signatures, or 2) forced tail call optimization when the
5324 following extra requirements are met:
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005325 <ul>
5326 <li>Caller and callee both have the calling
5327 convention <tt>fastcc</tt>.</li>
5328 <li>The call is in tail position (ret immediately follows call and ret
5329 uses value of call or is void).</li>
5330 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohmanfbbee8d2010-03-02 01:08:11 +00005331 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005332 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5333 constraints are met.</a></li>
5334 </ul>
5335 </li>
Devang Patelf642f472008-10-06 18:50:38 +00005336
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005337 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5338 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005339 defaults to using C calling conventions. The calling convention of the
5340 call must match the calling convention of the target function, or else the
5341 behavior is undefined.</li>
Devang Patelf642f472008-10-06 18:50:38 +00005342
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005343 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5344 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5345 '<tt>inreg</tt>' attributes are valid here.</li>
5346
5347 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5348 type of the return value. Functions that return no value are marked
5349 <tt><a href="#t_void">void</a></tt>.</li>
5350
5351 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5352 being invoked. The argument types must match the types implied by this
5353 signature. This type can be omitted if the function is not varargs and if
5354 the function type does not return a pointer to a function.</li>
5355
5356 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5357 be invoked. In most cases, this is a direct function invocation, but
5358 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5359 to function value.</li>
5360
5361 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +00005362 signature argument types and parameter attributes. All arguments must be
5363 of <a href="#t_firstclass">first class</a> type. If the function
5364 signature indicates the function accepts a variable number of arguments,
5365 the extra arguments can be specified.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005366
5367 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5368 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5369 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner6536cfe2002-05-06 22:08:29 +00005370</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +00005371
Chris Lattner00950542001-06-06 20:29:01 +00005372<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005373<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5374 a specified function, with its incoming arguments bound to the specified
5375 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5376 function, control flow continues with the instruction after the function
5377 call, and the return value of the function is bound to the result
5378 argument.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00005379
Chris Lattner00950542001-06-06 20:29:01 +00005380<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00005381<pre>
Nick Lewyckydb7e3c92007-09-08 13:57:50 +00005382 %retval = call i32 @test(i32 %argc)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005383 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattner772fccf2008-03-21 17:24:17 +00005384 %X = tail call i32 @foo() <i>; yields i32</i>
5385 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5386 call void %foo(i8 97 signext)
Devang Patelc3fc6df2008-03-10 20:49:15 +00005387
5388 %struct.A = type { i32, i8 }
Devang Patelf642f472008-10-06 18:50:38 +00005389 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmanb1e6b962008-10-04 19:00:07 +00005390 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5391 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner85a350f2008-10-08 06:26:11 +00005392 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmancb73d192008-10-07 10:03:45 +00005393 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattner2bff5242005-05-06 05:47:36 +00005394</pre>
5395
Dale Johannesen07de8d12009-09-24 18:38:21 +00005396<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen9f8380b2009-09-25 17:04:42 +00005397standard C99 library as being the C99 library functions, and may perform
5398optimizations or generate code for them under that assumption. This is
5399something we'd like to change in the future to provide better support for
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00005400freestanding environments and non-C-based languages.</p>
Dale Johannesen07de8d12009-09-24 18:38:21 +00005401
Misha Brukman9d0919f2003-11-08 01:05:38 +00005402</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005403
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005404<!-- _______________________________________________________________________ -->
Chris Lattnere19d7a72004-09-27 21:51:25 +00005405<div class="doc_subsubsection">
Chris Lattnerfb6977d2006-01-13 23:26:01 +00005406 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005407</div>
5408
Misha Brukman9d0919f2003-11-08 01:05:38 +00005409<div class="doc_text">
Chris Lattnere19d7a72004-09-27 21:51:25 +00005410
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005411<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005412<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005413 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattnere19d7a72004-09-27 21:51:25 +00005414</pre>
5415
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005416<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005417<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005418 the "variable argument" area of a function call. It is used to implement the
5419 <tt>va_arg</tt> macro in C.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005420
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005421<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005422<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5423 argument. It returns a value of the specified argument type and increments
5424 the <tt>va_list</tt> to point to the next argument. The actual type
5425 of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005426
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005427<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005428<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5429 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5430 to the next argument. For more information, see the variable argument
5431 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005432
5433<p>It is legal for this instruction to be called in a function which does not
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005434 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5435 function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005436
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005437<p><tt>va_arg</tt> is an LLVM instruction instead of
5438 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5439 argument.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005440
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005441<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005442<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5443
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005444<p>Note that the code generator does not yet fully support va_arg on many
5445 targets. Also, it does not currently support va_arg with aggregate types on
5446 any target.</p>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +00005447
Misha Brukman9d0919f2003-11-08 01:05:38 +00005448</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005449
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005450<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00005451<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5452<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005453
Misha Brukman9d0919f2003-11-08 01:05:38 +00005454<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +00005455
5456<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005457 well known names and semantics and are required to follow certain
5458 restrictions. Overall, these intrinsics represent an extension mechanism for
5459 the LLVM language that does not require changing all of the transformations
5460 in LLVM when adding to the language (or the bitcode reader/writer, the
5461 parser, etc...).</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005462
John Criswellfc6b8952005-05-16 16:17:45 +00005463<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005464 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5465 begin with this prefix. Intrinsic functions must always be external
5466 functions: you cannot define the body of intrinsic functions. Intrinsic
5467 functions may only be used in call or invoke instructions: it is illegal to
5468 take the address of an intrinsic function. Additionally, because intrinsic
5469 functions are part of the LLVM language, it is required if any are added that
5470 they be documented here.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005471
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005472<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5473 family of functions that perform the same operation but on different data
5474 types. Because LLVM can represent over 8 million different integer types,
5475 overloading is used commonly to allow an intrinsic function to operate on any
5476 integer type. One or more of the argument types or the result type can be
5477 overloaded to accept any integer type. Argument types may also be defined as
5478 exactly matching a previous argument's type or the result type. This allows
5479 an intrinsic function which accepts multiple arguments, but needs all of them
5480 to be of the same type, to only be overloaded with respect to a single
5481 argument or the result.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005482
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005483<p>Overloaded intrinsics will have the names of its overloaded argument types
5484 encoded into its function name, each preceded by a period. Only those types
5485 which are overloaded result in a name suffix. Arguments whose type is matched
5486 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5487 can take an integer of any width and returns an integer of exactly the same
5488 integer width. This leads to a family of functions such as
5489 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5490 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5491 suffix is required. Because the argument's type is matched against the return
5492 type, it does not require its own name suffix.</p>
Reid Spencer409e28f2007-04-01 08:04:23 +00005493
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005494<p>To learn how to add an intrinsic function, please see the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005495 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005496
Misha Brukman9d0919f2003-11-08 01:05:38 +00005497</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005498
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005499<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005500<div class="doc_subsection">
5501 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5502</div>
5503
Misha Brukman9d0919f2003-11-08 01:05:38 +00005504<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00005505
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005506<p>Variable argument support is defined in LLVM with
5507 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5508 intrinsic functions. These functions are related to the similarly named
5509 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005510
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005511<p>All of these functions operate on arguments that use a target-specific value
5512 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5513 not define what this type is, so all transformations should be prepared to
5514 handle these functions regardless of the type used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005515
Chris Lattner374ab302006-05-15 17:26:46 +00005516<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005517 instruction and the variable argument handling intrinsic functions are
5518 used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005519
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00005520<pre class="doc_code">
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005521define i32 @test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +00005522 ; Initialize variable argument processing
Jeff Cohenb627eab2007-04-29 01:07:00 +00005523 %ap = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +00005524 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005525 call void @llvm.va_start(i8* %ap2)
Chris Lattner33aec9e2004-02-12 17:01:32 +00005526
5527 ; Read a single integer argument
Jeff Cohenb627eab2007-04-29 01:07:00 +00005528 %tmp = va_arg i8** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +00005529
5530 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohenb627eab2007-04-29 01:07:00 +00005531 %aq = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +00005532 %aq2 = bitcast i8** %aq to i8*
Jeff Cohenb627eab2007-04-29 01:07:00 +00005533 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005534 call void @llvm.va_end(i8* %aq2)
Chris Lattner33aec9e2004-02-12 17:01:32 +00005535
5536 ; Stop processing of arguments.
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005537 call void @llvm.va_end(i8* %ap2)
Reid Spencerca86e162006-12-31 07:07:53 +00005538 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +00005539}
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005540
5541declare void @llvm.va_start(i8*)
5542declare void @llvm.va_copy(i8*, i8*)
5543declare void @llvm.va_end(i8*)
Chris Lattner33aec9e2004-02-12 17:01:32 +00005544</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +00005545
Bill Wendling2f7a8b02007-05-29 09:04:49 +00005546</div>
5547
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005548<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005549<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005550 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00005551</div>
5552
5553
Misha Brukman9d0919f2003-11-08 01:05:38 +00005554<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005555
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005556<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005557<pre>
5558 declare void %llvm.va_start(i8* &lt;arglist&gt;)
5559</pre>
5560
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005561<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005562<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
5563 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005564
5565<h5>Arguments:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +00005566<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005567
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005568<h5>Semantics:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +00005569<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005570 macro available in C. In a target-dependent way, it initializes
5571 the <tt>va_list</tt> element to which the argument points, so that the next
5572 call to <tt>va_arg</tt> will produce the first variable argument passed to
5573 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5574 need to know the last argument of the function as the compiler can figure
5575 that out.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005576
Misha Brukman9d0919f2003-11-08 01:05:38 +00005577</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005578
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005579<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005580<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005581 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00005582</div>
5583
Misha Brukman9d0919f2003-11-08 01:05:38 +00005584<div class="doc_text">
Chris Lattnerb75137d2007-01-08 07:55:15 +00005585
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005586<h5>Syntax:</h5>
5587<pre>
5588 declare void @llvm.va_end(i8* &lt;arglist&gt;)
5589</pre>
5590
5591<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005592<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005593 which has been initialized previously
5594 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5595 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +00005596
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005597<h5>Arguments:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005598<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +00005599
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005600<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00005601<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005602 macro available in C. In a target-dependent way, it destroys
5603 the <tt>va_list</tt> element to which the argument points. Calls
5604 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5605 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5606 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +00005607
Misha Brukman9d0919f2003-11-08 01:05:38 +00005608</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005609
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005610<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005611<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005612 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00005613</div>
5614
Misha Brukman9d0919f2003-11-08 01:05:38 +00005615<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00005616
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005617<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005618<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +00005619 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +00005620</pre>
5621
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005622<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005623<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005624 from the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005625
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005626<h5>Arguments:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005627<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005628 The second argument is a pointer to a <tt>va_list</tt> element to copy
5629 from.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005630
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005631<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005632<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005633 macro available in C. In a target-dependent way, it copies the
5634 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5635 element. This intrinsic is necessary because
5636 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5637 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005638
Misha Brukman9d0919f2003-11-08 01:05:38 +00005639</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005640
Chris Lattner33aec9e2004-02-12 17:01:32 +00005641<!-- ======================================================================= -->
5642<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +00005643 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5644</div>
5645
5646<div class="doc_text">
5647
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005648<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattnerd3eda892008-08-05 18:29:16 +00005649Collection</a> (GC) requires the implementation and generation of these
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005650intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5651roots on the stack</a>, as well as garbage collector implementations that
5652require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5653barriers. Front-ends for type-safe garbage collected languages should generate
5654these intrinsics to make use of the LLVM garbage collectors. For more details,
5655see <a href="GarbageCollection.html">Accurate Garbage Collection with
5656LLVM</a>.</p>
Christopher Lamb303dae92007-12-17 01:00:21 +00005657
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005658<p>The garbage collection intrinsics only operate on objects in the generic
5659 address space (address space zero).</p>
Christopher Lamb303dae92007-12-17 01:00:21 +00005660
Chris Lattnerd7923912004-05-23 21:06:01 +00005661</div>
5662
5663<!-- _______________________________________________________________________ -->
5664<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005665 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +00005666</div>
5667
5668<div class="doc_text">
5669
5670<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005671<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005672 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +00005673</pre>
5674
5675<h5>Overview:</h5>
John Criswell9e2485c2004-12-10 15:51:16 +00005676<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005677 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005678
5679<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005680<p>The first argument specifies the address of a stack object that contains the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005681 root pointer. The second pointer (which must be either a constant or a
5682 global value address) contains the meta-data to be associated with the
5683 root.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005684
5685<h5>Semantics:</h5>
Chris Lattner05d67092008-04-24 05:59:56 +00005686<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005687 location. At compile-time, the code generator generates information to allow
5688 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5689 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5690 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005691
5692</div>
5693
Chris Lattnerd7923912004-05-23 21:06:01 +00005694<!-- _______________________________________________________________________ -->
5695<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005696 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +00005697</div>
5698
5699<div class="doc_text">
5700
5701<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005702<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005703 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +00005704</pre>
5705
5706<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005707<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005708 locations, allowing garbage collector implementations that require read
5709 barriers.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005710
5711<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +00005712<p>The second argument is the address to read from, which should be an address
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005713 allocated from the garbage collector. The first object is a pointer to the
5714 start of the referenced object, if needed by the language runtime (otherwise
5715 null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005716
5717<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005718<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005719 instruction, but may be replaced with substantially more complex code by the
5720 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5721 may only be used in a function which <a href="#gc">specifies a GC
5722 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005723
5724</div>
5725
Chris Lattnerd7923912004-05-23 21:06:01 +00005726<!-- _______________________________________________________________________ -->
5727<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005728 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +00005729</div>
5730
5731<div class="doc_text">
5732
5733<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005734<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005735 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +00005736</pre>
5737
5738<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005739<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005740 locations, allowing garbage collector implementations that require write
5741 barriers (such as generational or reference counting collectors).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005742
5743<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +00005744<p>The first argument is the reference to store, the second is the start of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005745 object to store it to, and the third is the address of the field of Obj to
5746 store to. If the runtime does not require a pointer to the object, Obj may
5747 be null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005748
5749<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005750<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005751 instruction, but may be replaced with substantially more complex code by the
5752 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5753 may only be used in a function which <a href="#gc">specifies a GC
5754 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005755
5756</div>
5757
Chris Lattnerd7923912004-05-23 21:06:01 +00005758<!-- ======================================================================= -->
5759<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +00005760 <a name="int_codegen">Code Generator Intrinsics</a>
5761</div>
5762
5763<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005764
5765<p>These intrinsics are provided by LLVM to expose special features that may
5766 only be implemented with code generator support.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005767
5768</div>
5769
5770<!-- _______________________________________________________________________ -->
5771<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005772 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
Chris Lattner10610642004-02-14 04:08:35 +00005773</div>
5774
5775<div class="doc_text">
5776
5777<h5>Syntax:</h5>
5778<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +00005779 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00005780</pre>
5781
5782<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005783<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5784 target-specific value indicating the return address of the current function
5785 or one of its callers.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005786
5787<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005788<p>The argument to this intrinsic indicates which function to return the address
5789 for. Zero indicates the calling function, one indicates its caller, etc.
5790 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005791
5792<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005793<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5794 indicating the return address of the specified call frame, or zero if it
5795 cannot be identified. The value returned by this intrinsic is likely to be
5796 incorrect or 0 for arguments other than zero, so it should only be used for
5797 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005798
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005799<p>Note that calling this intrinsic does not prevent function inlining or other
5800 aggressive transformations, so the value returned may not be that of the
5801 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005802
Chris Lattner10610642004-02-14 04:08:35 +00005803</div>
5804
Chris Lattner10610642004-02-14 04:08:35 +00005805<!-- _______________________________________________________________________ -->
5806<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005807 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
Chris Lattner10610642004-02-14 04:08:35 +00005808</div>
5809
5810<div class="doc_text">
5811
5812<h5>Syntax:</h5>
5813<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005814 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00005815</pre>
5816
5817<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005818<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5819 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005820
5821<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005822<p>The argument to this intrinsic indicates which function to return the frame
5823 pointer for. Zero indicates the calling function, one indicates its caller,
5824 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005825
5826<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005827<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5828 indicating the frame address of the specified call frame, or zero if it
5829 cannot be identified. The value returned by this intrinsic is likely to be
5830 incorrect or 0 for arguments other than zero, so it should only be used for
5831 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005832
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005833<p>Note that calling this intrinsic does not prevent function inlining or other
5834 aggressive transformations, so the value returned may not be that of the
5835 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005836
Chris Lattner10610642004-02-14 04:08:35 +00005837</div>
5838
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005839<!-- _______________________________________________________________________ -->
5840<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005841 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
Chris Lattner57e1f392006-01-13 02:03:13 +00005842</div>
5843
5844<div class="doc_text">
5845
5846<h5>Syntax:</h5>
5847<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005848 declare i8* @llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +00005849</pre>
5850
5851<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005852<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5853 of the function stack, for use
5854 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5855 useful for implementing language features like scoped automatic variable
5856 sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005857
5858<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005859<p>This intrinsic returns a opaque pointer value that can be passed
5860 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5861 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5862 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5863 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5864 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5865 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005866
5867</div>
5868
5869<!-- _______________________________________________________________________ -->
5870<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005871 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
Chris Lattner57e1f392006-01-13 02:03:13 +00005872</div>
5873
5874<div class="doc_text">
5875
5876<h5>Syntax:</h5>
5877<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005878 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +00005879</pre>
5880
5881<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005882<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5883 the function stack to the state it was in when the
5884 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5885 executed. This is useful for implementing language features like scoped
5886 automatic variable sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005887
5888<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005889<p>See the description
5890 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005891
5892</div>
5893
Chris Lattner57e1f392006-01-13 02:03:13 +00005894<!-- _______________________________________________________________________ -->
5895<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005896 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005897</div>
5898
5899<div class="doc_text">
5900
5901<h5>Syntax:</h5>
5902<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005903 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005904</pre>
5905
5906<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005907<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5908 insert a prefetch instruction if supported; otherwise, it is a noop.
5909 Prefetches have no effect on the behavior of the program but can change its
5910 performance characteristics.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005911
5912<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005913<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5914 specifier determining if the fetch should be for a read (0) or write (1),
5915 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5916 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5917 and <tt>locality</tt> arguments must be constant integers.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005918
5919<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005920<p>This intrinsic does not modify the behavior of the program. In particular,
5921 prefetches cannot trap and do not produce a value. On targets that support
5922 this intrinsic, the prefetch can provide hints to the processor cache for
5923 better performance.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005924
5925</div>
5926
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005927<!-- _______________________________________________________________________ -->
5928<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005929 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005930</div>
5931
5932<div class="doc_text">
5933
5934<h5>Syntax:</h5>
5935<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005936 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005937</pre>
5938
5939<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005940<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5941 Counter (PC) in a region of code to simulators and other tools. The method
5942 is target specific, but it is expected that the marker will use exported
5943 symbols to transmit the PC of the marker. The marker makes no guarantees
5944 that it will remain with any specific instruction after optimizations. It is
5945 possible that the presence of a marker will inhibit optimizations. The
5946 intended use is to be inserted after optimizations to allow correlations of
5947 simulation runs.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005948
5949<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005950<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005951
5952<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005953<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00005954 not support this intrinsic may ignore it.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005955
5956</div>
5957
Andrew Lenharth51b8d542005-11-11 16:47:30 +00005958<!-- _______________________________________________________________________ -->
5959<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005960 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
Andrew Lenharth51b8d542005-11-11 16:47:30 +00005961</div>
5962
5963<div class="doc_text">
5964
5965<h5>Syntax:</h5>
5966<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00005967 declare i64 @llvm.readcyclecounter()
Andrew Lenharth51b8d542005-11-11 16:47:30 +00005968</pre>
5969
5970<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005971<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5972 counter register (or similar low latency, high accuracy clocks) on those
5973 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5974 should map to RPCC. As the backing counters overflow quickly (on the order
5975 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +00005976
5977<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005978<p>When directly supported, reading the cycle counter should not modify any
5979 memory. Implementations are allowed to either return a application specific
5980 value or a system wide value. On backends without support, this is lowered
5981 to a constant 0.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +00005982
5983</div>
5984
Chris Lattner10610642004-02-14 04:08:35 +00005985<!-- ======================================================================= -->
5986<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +00005987 <a name="int_libc">Standard C Library Intrinsics</a>
5988</div>
5989
5990<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005991
5992<p>LLVM provides intrinsics for a few important standard C library functions.
5993 These intrinsics allow source-language front-ends to pass information about
5994 the alignment of the pointer arguments to the code generator, providing
5995 opportunity for more efficient code generation.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005996
5997</div>
5998
5999<!-- _______________________________________________________________________ -->
6000<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006001 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006002</div>
6003
6004<div class="doc_text">
6005
6006<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006007<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
Mon P Wange88909b2010-04-07 06:35:53 +00006008 integer bit width and for different address spaces. Not all targets support
6009 all bit widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006010
Chris Lattner33aec9e2004-02-12 17:01:32 +00006011<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006012 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006013 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006014 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006015 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +00006016</pre>
6017
6018<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006019<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6020 source location to the destination location.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006021
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006022<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +00006023 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6024 and the pointers can be in specified address spaces.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006025
6026<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006027
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006028<p>The first argument is a pointer to the destination, the second is a pointer
6029 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +00006030 number of bytes to copy, the fourth argument is the alignment of the
6031 source and destination locations, and the fifth is a boolean indicating a
6032 volatile access.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006033
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006034<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006035 then the caller guarantees that both the source and destination pointers are
6036 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +00006037
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00006038<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6039 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6040 The detailed access behavior is not very cleanly specified and it is unwise
6041 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +00006042
Chris Lattner33aec9e2004-02-12 17:01:32 +00006043<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006044
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006045<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6046 source location to the destination location, which are not allowed to
6047 overlap. It copies "len" bytes of memory over. If the argument is known to
6048 be aligned to some boundary, this can be specified as the fourth argument,
6049 otherwise it should be set to 0 or 1.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006050
Chris Lattner33aec9e2004-02-12 17:01:32 +00006051</div>
6052
Chris Lattner0eb51b42004-02-12 18:10:10 +00006053<!-- _______________________________________________________________________ -->
6054<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006055 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006056</div>
6057
6058<div class="doc_text">
6059
6060<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +00006061<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Mon P Wange88909b2010-04-07 06:35:53 +00006062 width and for different address space. Not all targets support all bit
6063 widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006064
Chris Lattner0eb51b42004-02-12 18:10:10 +00006065<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006066 declare void @llvm.memmove.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006067 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006068 declare void @llvm.memmove.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006069 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +00006070</pre>
6071
6072<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006073<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6074 source location to the destination location. It is similar to the
6075 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6076 overlap.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006077
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006078<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +00006079 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6080 and the pointers can be in specified address spaces.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006081
6082<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006083
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006084<p>The first argument is a pointer to the destination, the second is a pointer
6085 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +00006086 number of bytes to copy, the fourth argument is the alignment of the
6087 source and destination locations, and the fifth is a boolean indicating a
6088 volatile access.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006089
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006090<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006091 then the caller guarantees that the source and destination pointers are
6092 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +00006093
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00006094<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6095 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6096 The detailed access behavior is not very cleanly specified and it is unwise
6097 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +00006098
Chris Lattner0eb51b42004-02-12 18:10:10 +00006099<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006100
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006101<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6102 source location to the destination location, which may overlap. It copies
6103 "len" bytes of memory over. If the argument is known to be aligned to some
6104 boundary, this can be specified as the fourth argument, otherwise it should
6105 be set to 0 or 1.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006106
Chris Lattner0eb51b42004-02-12 18:10:10 +00006107</div>
6108
Chris Lattner10610642004-02-14 04:08:35 +00006109<!-- _______________________________________________________________________ -->
6110<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006111 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner10610642004-02-14 04:08:35 +00006112</div>
6113
6114<div class="doc_text">
6115
6116<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +00006117<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
Mon P Wange88909b2010-04-07 06:35:53 +00006118 width and for different address spaces. Not all targets support all bit
6119 widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006120
Chris Lattner10610642004-02-14 04:08:35 +00006121<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006122 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +00006123 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006124 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +00006125 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00006126</pre>
6127
6128<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006129<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6130 particular byte value.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006131
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006132<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +00006133 intrinsic does not return a value, takes extra alignment/volatile arguments,
6134 and the destination can be in an arbitrary address space.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006135
6136<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006137<p>The first argument is a pointer to the destination to fill, the second is the
6138 byte value to fill it with, the third argument is an integer argument
6139 specifying the number of bytes to fill, and the fourth argument is the known
6140 alignment of destination location.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006141
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006142<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006143 then the caller guarantees that the destination pointer is aligned to that
6144 boundary.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006145
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00006146<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6147 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6148 The detailed access behavior is not very cleanly specified and it is unwise
6149 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +00006150
Chris Lattner10610642004-02-14 04:08:35 +00006151<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006152<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6153 at the destination location. If the argument is known to be aligned to some
6154 boundary, this can be specified as the fourth argument, otherwise it should
6155 be set to 0 or 1.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006156
Chris Lattner10610642004-02-14 04:08:35 +00006157</div>
6158
Chris Lattner32006282004-06-11 02:28:03 +00006159<!-- _______________________________________________________________________ -->
6160<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006161 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattnera4d74142005-07-21 01:29:16 +00006162</div>
6163
6164<div class="doc_text">
6165
6166<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006167<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6168 floating point or vector of floating point type. Not all targets support all
6169 types however.</p>
6170
Chris Lattnera4d74142005-07-21 01:29:16 +00006171<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +00006172 declare float @llvm.sqrt.f32(float %Val)
6173 declare double @llvm.sqrt.f64(double %Val)
6174 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6175 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6176 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +00006177</pre>
6178
6179<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006180<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6181 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6182 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6183 behavior for negative numbers other than -0.0 (which allows for better
6184 optimization, because there is no need to worry about errno being
6185 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +00006186
6187<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006188<p>The argument and return value are floating point numbers of the same
6189 type.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +00006190
6191<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006192<p>This function returns the sqrt of the specified operand if it is a
6193 nonnegative floating point number.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +00006194
Chris Lattnera4d74142005-07-21 01:29:16 +00006195</div>
6196
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006197<!-- _______________________________________________________________________ -->
6198<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006199 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006200</div>
6201
6202<div class="doc_text">
6203
6204<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006205<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6206 floating point or vector of floating point type. Not all targets support all
6207 types however.</p>
6208
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006209<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +00006210 declare float @llvm.powi.f32(float %Val, i32 %power)
6211 declare double @llvm.powi.f64(double %Val, i32 %power)
6212 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6213 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6214 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006215</pre>
6216
6217<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006218<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6219 specified (positive or negative) power. The order of evaluation of
6220 multiplications is not defined. When a vector of floating point type is
6221 used, the second argument remains a scalar integer value.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006222
6223<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006224<p>The second argument is an integer power, and the first is a value to raise to
6225 that power.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006226
6227<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006228<p>This function returns the first value raised to the second power with an
6229 unspecified sequence of rounding operations.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006230
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006231</div>
6232
Dan Gohman91c284c2007-10-15 20:30:11 +00006233<!-- _______________________________________________________________________ -->
6234<div class="doc_subsubsection">
6235 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6236</div>
6237
6238<div class="doc_text">
6239
6240<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006241<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6242 floating point or vector of floating point type. Not all targets support all
6243 types however.</p>
6244
Dan Gohman91c284c2007-10-15 20:30:11 +00006245<pre>
6246 declare float @llvm.sin.f32(float %Val)
6247 declare double @llvm.sin.f64(double %Val)
6248 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6249 declare fp128 @llvm.sin.f128(fp128 %Val)
6250 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6251</pre>
6252
6253<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006254<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006255
6256<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006257<p>The argument and return value are floating point numbers of the same
6258 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006259
6260<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006261<p>This function returns the sine of the specified operand, returning the same
6262 values as the libm <tt>sin</tt> functions would, and handles error conditions
6263 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006264
Dan Gohman91c284c2007-10-15 20:30:11 +00006265</div>
6266
6267<!-- _______________________________________________________________________ -->
6268<div class="doc_subsubsection">
6269 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6270</div>
6271
6272<div class="doc_text">
6273
6274<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006275<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6276 floating point or vector of floating point type. Not all targets support all
6277 types however.</p>
6278
Dan Gohman91c284c2007-10-15 20:30:11 +00006279<pre>
6280 declare float @llvm.cos.f32(float %Val)
6281 declare double @llvm.cos.f64(double %Val)
6282 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6283 declare fp128 @llvm.cos.f128(fp128 %Val)
6284 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6285</pre>
6286
6287<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006288<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006289
6290<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006291<p>The argument and return value are floating point numbers of the same
6292 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006293
6294<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006295<p>This function returns the cosine of the specified operand, returning the same
6296 values as the libm <tt>cos</tt> functions would, and handles error conditions
6297 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006298
Dan Gohman91c284c2007-10-15 20:30:11 +00006299</div>
6300
6301<!-- _______________________________________________________________________ -->
6302<div class="doc_subsubsection">
6303 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6304</div>
6305
6306<div class="doc_text">
6307
6308<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006309<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6310 floating point or vector of floating point type. Not all targets support all
6311 types however.</p>
6312
Dan Gohman91c284c2007-10-15 20:30:11 +00006313<pre>
6314 declare float @llvm.pow.f32(float %Val, float %Power)
6315 declare double @llvm.pow.f64(double %Val, double %Power)
6316 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6317 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6318 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6319</pre>
6320
6321<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006322<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6323 specified (positive or negative) power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006324
6325<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006326<p>The second argument is a floating point power, and the first is a value to
6327 raise to that power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006328
6329<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006330<p>This function returns the first value raised to the second power, returning
6331 the same values as the libm <tt>pow</tt> functions would, and handles error
6332 conditions in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006333
Dan Gohman91c284c2007-10-15 20:30:11 +00006334</div>
6335
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006336<!-- ======================================================================= -->
6337<div class="doc_subsection">
Nate Begeman7e36c472006-01-13 23:26:38 +00006338 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006339</div>
6340
6341<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006342
6343<p>LLVM provides intrinsics for a few important bit manipulation operations.
6344 These allow efficient code generation for some algorithms.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006345
6346</div>
6347
6348<!-- _______________________________________________________________________ -->
6349<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006350 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
Nate Begeman7e36c472006-01-13 23:26:38 +00006351</div>
6352
6353<div class="doc_text">
6354
6355<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +00006356<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006357 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6358
Nate Begeman7e36c472006-01-13 23:26:38 +00006359<pre>
Chandler Carruth69940402007-08-04 01:51:18 +00006360 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
6361 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
6362 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +00006363</pre>
6364
6365<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006366<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6367 values with an even number of bytes (positive multiple of 16 bits). These
6368 are useful for performing operations on data that is not in the target's
6369 native byte order.</p>
Nate Begeman7e36c472006-01-13 23:26:38 +00006370
6371<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006372<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6373 and low byte of the input i16 swapped. Similarly,
6374 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6375 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6376 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6377 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6378 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6379 more, respectively).</p>
Nate Begeman7e36c472006-01-13 23:26:38 +00006380
6381</div>
6382
6383<!-- _______________________________________________________________________ -->
6384<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +00006385 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006386</div>
6387
6388<div class="doc_text">
6389
6390<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +00006391<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006392 width. Not all targets support all bit widths however.</p>
6393
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006394<pre>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006395 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006396 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +00006397 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006398 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
6399 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006400</pre>
6401
6402<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006403<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6404 in a value.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006405
6406<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006407<p>The only argument is the value to be counted. The argument may be of any
6408 integer type. The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006409
6410<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006411<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006412
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006413</div>
6414
6415<!-- _______________________________________________________________________ -->
6416<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +00006417 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006418</div>
6419
6420<div class="doc_text">
6421
6422<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006423<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6424 integer bit width. Not all targets support all bit widths however.</p>
6425
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006426<pre>
Chandler Carruth69940402007-08-04 01:51:18 +00006427 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
6428 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +00006429 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006430 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
6431 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006432</pre>
6433
6434<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006435<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6436 leading zeros in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006437
6438<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006439<p>The only argument is the value to be counted. The argument may be of any
6440 integer type. The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006441
6442<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006443<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6444 zeros in a variable. If the src == 0 then the result is the size in bits of
6445 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006446
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006447</div>
Chris Lattner32006282004-06-11 02:28:03 +00006448
Chris Lattnereff29ab2005-05-15 19:39:26 +00006449<!-- _______________________________________________________________________ -->
6450<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +00006451 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006452</div>
6453
6454<div class="doc_text">
6455
6456<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006457<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6458 integer bit width. Not all targets support all bit widths however.</p>
6459
Chris Lattnereff29ab2005-05-15 19:39:26 +00006460<pre>
Chandler Carruth69940402007-08-04 01:51:18 +00006461 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
6462 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +00006463 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006464 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
6465 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +00006466</pre>
6467
6468<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006469<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6470 trailing zeros.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006471
6472<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006473<p>The only argument is the value to be counted. The argument may be of any
6474 integer type. The return type must match the argument type.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006475
6476<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006477<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6478 zeros in a variable. If the src == 0 then the result is the size in bits of
6479 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006480
Chris Lattnereff29ab2005-05-15 19:39:26 +00006481</div>
6482
Bill Wendlingda01af72009-02-08 04:04:40 +00006483<!-- ======================================================================= -->
6484<div class="doc_subsection">
6485 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6486</div>
6487
6488<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006489
6490<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingda01af72009-02-08 04:04:40 +00006491
6492</div>
6493
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006494<!-- _______________________________________________________________________ -->
6495<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006496 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006497</div>
6498
6499<div class="doc_text">
6500
6501<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006502<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006503 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006504
6505<pre>
6506 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6507 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6508 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6509</pre>
6510
6511<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006512<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006513 a signed addition of the two arguments, and indicate whether an overflow
6514 occurred during the signed summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006515
6516<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006517<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006518 be of integer types of any bit width, but they must have the same bit
6519 width. The second element of the result structure must be of
6520 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6521 undergo signed addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006522
6523<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006524<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006525 a signed addition of the two variables. They return a structure &mdash; the
6526 first element of which is the signed summation, and the second element of
6527 which is a bit specifying if the signed summation resulted in an
6528 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006529
6530<h5>Examples:</h5>
6531<pre>
6532 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6533 %sum = extractvalue {i32, i1} %res, 0
6534 %obit = extractvalue {i32, i1} %res, 1
6535 br i1 %obit, label %overflow, label %normal
6536</pre>
6537
6538</div>
6539
6540<!-- _______________________________________________________________________ -->
6541<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006542 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006543</div>
6544
6545<div class="doc_text">
6546
6547<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006548<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006549 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006550
6551<pre>
6552 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6553 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6554 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6555</pre>
6556
6557<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006558<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006559 an unsigned addition of the two arguments, and indicate whether a carry
6560 occurred during the unsigned summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006561
6562<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006563<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006564 be of integer types of any bit width, but they must have the same bit
6565 width. The second element of the result structure must be of
6566 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6567 undergo unsigned addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006568
6569<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006570<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006571 an unsigned addition of the two arguments. They return a structure &mdash;
6572 the first element of which is the sum, and the second element of which is a
6573 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006574
6575<h5>Examples:</h5>
6576<pre>
6577 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6578 %sum = extractvalue {i32, i1} %res, 0
6579 %obit = extractvalue {i32, i1} %res, 1
6580 br i1 %obit, label %carry, label %normal
6581</pre>
6582
6583</div>
6584
6585<!-- _______________________________________________________________________ -->
6586<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006587 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006588</div>
6589
6590<div class="doc_text">
6591
6592<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006593<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006594 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006595
6596<pre>
6597 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6598 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6599 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6600</pre>
6601
6602<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006603<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006604 a signed subtraction of the two arguments, and indicate whether an overflow
6605 occurred during the signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006606
6607<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006608<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006609 be of integer types of any bit width, but they must have the same bit
6610 width. The second element of the result structure must be of
6611 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6612 undergo signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006613
6614<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006615<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006616 a signed subtraction of the two arguments. They return a structure &mdash;
6617 the first element of which is the subtraction, and the second element of
6618 which is a bit specifying if the signed subtraction resulted in an
6619 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006620
6621<h5>Examples:</h5>
6622<pre>
6623 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6624 %sum = extractvalue {i32, i1} %res, 0
6625 %obit = extractvalue {i32, i1} %res, 1
6626 br i1 %obit, label %overflow, label %normal
6627</pre>
6628
6629</div>
6630
6631<!-- _______________________________________________________________________ -->
6632<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006633 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006634</div>
6635
6636<div class="doc_text">
6637
6638<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006639<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006640 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006641
6642<pre>
6643 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6644 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6645 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6646</pre>
6647
6648<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006649<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006650 an unsigned subtraction of the two arguments, and indicate whether an
6651 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006652
6653<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006654<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006655 be of integer types of any bit width, but they must have the same bit
6656 width. The second element of the result structure must be of
6657 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6658 undergo unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006659
6660<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006661<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006662 an unsigned subtraction of the two arguments. They return a structure &mdash;
6663 the first element of which is the subtraction, and the second element of
6664 which is a bit specifying if the unsigned subtraction resulted in an
6665 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006666
6667<h5>Examples:</h5>
6668<pre>
6669 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6670 %sum = extractvalue {i32, i1} %res, 0
6671 %obit = extractvalue {i32, i1} %res, 1
6672 br i1 %obit, label %overflow, label %normal
6673</pre>
6674
6675</div>
6676
6677<!-- _______________________________________________________________________ -->
6678<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006679 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006680</div>
6681
6682<div class="doc_text">
6683
6684<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006685<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006686 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006687
6688<pre>
6689 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6690 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6691 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6692</pre>
6693
6694<h5>Overview:</h5>
6695
6696<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006697 a signed multiplication of the two arguments, and indicate whether an
6698 overflow occurred during the signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006699
6700<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006701<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006702 be of integer types of any bit width, but they must have the same bit
6703 width. The second element of the result structure must be of
6704 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6705 undergo signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006706
6707<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006708<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006709 a signed multiplication of the two arguments. They return a structure &mdash;
6710 the first element of which is the multiplication, and the second element of
6711 which is a bit specifying if the signed multiplication resulted in an
6712 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006713
6714<h5>Examples:</h5>
6715<pre>
6716 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6717 %sum = extractvalue {i32, i1} %res, 0
6718 %obit = extractvalue {i32, i1} %res, 1
6719 br i1 %obit, label %overflow, label %normal
6720</pre>
6721
Reid Spencerf86037f2007-04-11 23:23:49 +00006722</div>
6723
Bill Wendling41b485c2009-02-08 23:00:09 +00006724<!-- _______________________________________________________________________ -->
6725<div class="doc_subsubsection">
6726 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6727</div>
6728
6729<div class="doc_text">
6730
6731<h5>Syntax:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006732<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006733 on any integer bit width.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006734
6735<pre>
6736 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6737 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6738 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6739</pre>
6740
6741<h5>Overview:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006742<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006743 a unsigned multiplication of the two arguments, and indicate whether an
6744 overflow occurred during the unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006745
6746<h5>Arguments:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006747<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006748 be of integer types of any bit width, but they must have the same bit
6749 width. The second element of the result structure must be of
6750 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6751 undergo unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006752
6753<h5>Semantics:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006754<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006755 an unsigned multiplication of the two arguments. They return a structure
6756 &mdash; the first element of which is the multiplication, and the second
6757 element of which is a bit specifying if the unsigned multiplication resulted
6758 in an overflow.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006759
6760<h5>Examples:</h5>
6761<pre>
6762 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6763 %sum = extractvalue {i32, i1} %res, 0
6764 %obit = extractvalue {i32, i1} %res, 1
6765 br i1 %obit, label %overflow, label %normal
6766</pre>
6767
6768</div>
6769
Chris Lattner8ff75902004-01-06 05:31:32 +00006770<!-- ======================================================================= -->
6771<div class="doc_subsection">
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006772 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
6773</div>
6774
6775<div class="doc_text">
6776
Chris Lattner0cec9c82010-03-15 04:12:21 +00006777<p>Half precision floating point is a storage-only format. This means that it is
6778 a dense encoding (in memory) but does not support computation in the
6779 format.</p>
Chris Lattner82c3dc62010-03-14 23:03:31 +00006780
Chris Lattner0cec9c82010-03-15 04:12:21 +00006781<p>This means that code must first load the half-precision floating point
Chris Lattner82c3dc62010-03-14 23:03:31 +00006782 value as an i16, then convert it to float with <a
6783 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
6784 Computation can then be performed on the float value (including extending to
Chris Lattner0cec9c82010-03-15 04:12:21 +00006785 double etc). To store the value back to memory, it is first converted to
6786 float if needed, then converted to i16 with
Chris Lattner82c3dc62010-03-14 23:03:31 +00006787 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
6788 storing as an i16 value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006789</div>
6790
6791<!-- _______________________________________________________________________ -->
6792<div class="doc_subsubsection">
Chris Lattner82c3dc62010-03-14 23:03:31 +00006793 <a name="int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006794</div>
6795
6796<div class="doc_text">
6797
6798<h5>Syntax:</h5>
6799<pre>
6800 declare i16 @llvm.convert.to.fp16(f32 %a)
6801</pre>
6802
6803<h5>Overview:</h5>
6804<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6805 a conversion from single precision floating point format to half precision
6806 floating point format.</p>
6807
6808<h5>Arguments:</h5>
6809<p>The intrinsic function contains single argument - the value to be
6810 converted.</p>
6811
6812<h5>Semantics:</h5>
6813<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6814 a conversion from single precision floating point format to half precision
Chris Lattner0cec9c82010-03-15 04:12:21 +00006815 floating point format. The return value is an <tt>i16</tt> which
Chris Lattner82c3dc62010-03-14 23:03:31 +00006816 contains the converted number.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006817
6818<h5>Examples:</h5>
6819<pre>
6820 %res = call i16 @llvm.convert.to.fp16(f32 %a)
6821 store i16 %res, i16* @x, align 2
6822</pre>
6823
6824</div>
6825
6826<!-- _______________________________________________________________________ -->
6827<div class="doc_subsubsection">
Chris Lattner82c3dc62010-03-14 23:03:31 +00006828 <a name="int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006829</div>
6830
6831<div class="doc_text">
6832
6833<h5>Syntax:</h5>
6834<pre>
6835 declare f32 @llvm.convert.from.fp16(i16 %a)
6836</pre>
6837
6838<h5>Overview:</h5>
6839<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
6840 a conversion from half precision floating point format to single precision
6841 floating point format.</p>
6842
6843<h5>Arguments:</h5>
6844<p>The intrinsic function contains single argument - the value to be
6845 converted.</p>
6846
6847<h5>Semantics:</h5>
6848<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner0cec9c82010-03-15 04:12:21 +00006849 conversion from half single precision floating point format to single
Chris Lattner82c3dc62010-03-14 23:03:31 +00006850 precision floating point format. The input half-float value is represented by
6851 an <tt>i16</tt> value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006852
6853<h5>Examples:</h5>
6854<pre>
6855 %a = load i16* @x, align 2
6856 %res = call f32 @llvm.convert.from.fp16(i16 %a)
6857</pre>
6858
6859</div>
6860
6861<!-- ======================================================================= -->
6862<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +00006863 <a name="int_debugger">Debugger Intrinsics</a>
6864</div>
6865
6866<div class="doc_text">
Chris Lattner8ff75902004-01-06 05:31:32 +00006867
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006868<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6869 prefix), are described in
6870 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6871 Level Debugging</a> document.</p>
6872
6873</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00006874
Jim Laskeydd4ef1b2007-03-14 19:31:19 +00006875<!-- ======================================================================= -->
6876<div class="doc_subsection">
6877 <a name="int_eh">Exception Handling Intrinsics</a>
6878</div>
6879
6880<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006881
6882<p>The LLVM exception handling intrinsics (which all start with
6883 <tt>llvm.eh.</tt> prefix), are described in
6884 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6885 Handling</a> document.</p>
6886
Jim Laskeydd4ef1b2007-03-14 19:31:19 +00006887</div>
6888
Tanya Lattner6d806e92007-06-15 20:50:54 +00006889<!-- ======================================================================= -->
6890<div class="doc_subsection">
Duncan Sandsf7331b32007-09-11 14:10:23 +00006891 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands36397f52007-07-27 12:58:54 +00006892</div>
6893
6894<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006895
6896<p>This intrinsic makes it possible to excise one parameter, marked with
Dan Gohmanff235352010-07-02 23:18:08 +00006897 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
6898 The result is a callable
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006899 function pointer lacking the nest parameter - the caller does not need to
6900 provide a value for it. Instead, the value to use is stored in advance in a
6901 "trampoline", a block of memory usually allocated on the stack, which also
6902 contains code to splice the nest value into the argument list. This is used
6903 to implement the GCC nested function address extension.</p>
6904
6905<p>For example, if the function is
6906 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6907 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6908 follows:</p>
6909
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00006910<pre class="doc_code">
Duncan Sandsf7331b32007-09-11 14:10:23 +00006911 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6912 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00006913 %p = call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval)
Duncan Sandsf7331b32007-09-11 14:10:23 +00006914 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands36397f52007-07-27 12:58:54 +00006915</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006916
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00006917<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
6918 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006919
Duncan Sands36397f52007-07-27 12:58:54 +00006920</div>
6921
6922<!-- _______________________________________________________________________ -->
6923<div class="doc_subsubsection">
6924 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6925</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006926
Duncan Sands36397f52007-07-27 12:58:54 +00006927<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006928
Duncan Sands36397f52007-07-27 12:58:54 +00006929<h5>Syntax:</h5>
6930<pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006931 declare i8* @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands36397f52007-07-27 12:58:54 +00006932</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006933
Duncan Sands36397f52007-07-27 12:58:54 +00006934<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006935<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6936 function pointer suitable for executing it.</p>
6937
Duncan Sands36397f52007-07-27 12:58:54 +00006938<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006939<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6940 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6941 sufficiently aligned block of memory; this memory is written to by the
6942 intrinsic. Note that the size and the alignment are target-specific - LLVM
6943 currently provides no portable way of determining them, so a front-end that
6944 generates this intrinsic needs to have some target-specific knowledge.
6945 The <tt>func</tt> argument must hold a function bitcast to
6946 an <tt>i8*</tt>.</p>
6947
Duncan Sands36397f52007-07-27 12:58:54 +00006948<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006949<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6950 dependent code, turning it into a function. A pointer to this function is
6951 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6952 function pointer type</a> before being called. The new function's signature
6953 is the same as that of <tt>func</tt> with any arguments marked with
6954 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6955 is allowed, and it must be of pointer type. Calling the new function is
6956 equivalent to calling <tt>func</tt> with the same argument list, but
6957 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6958 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6959 by <tt>tramp</tt> is modified, then the effect of any later call to the
6960 returned function pointer is undefined.</p>
6961
Duncan Sands36397f52007-07-27 12:58:54 +00006962</div>
6963
6964<!-- ======================================================================= -->
6965<div class="doc_subsection">
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00006966 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6967</div>
6968
6969<div class="doc_text">
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00006970
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006971<p>These intrinsic functions expand the "universal IR" of LLVM to represent
6972 hardware constructs for atomic operations and memory synchronization. This
6973 provides an interface to the hardware, not an interface to the programmer. It
6974 is aimed at a low enough level to allow any programming models or APIs
6975 (Application Programming Interfaces) which need atomic behaviors to map
6976 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6977 hardware provides a "universal IR" for source languages, it also provides a
6978 starting point for developing a "universal" atomic operation and
6979 synchronization IR.</p>
6980
6981<p>These do <em>not</em> form an API such as high-level threading libraries,
6982 software transaction memory systems, atomic primitives, and intrinsic
6983 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6984 application libraries. The hardware interface provided by LLVM should allow
6985 a clean implementation of all of these APIs and parallel programming models.
6986 No one model or paradigm should be selected above others unless the hardware
6987 itself ubiquitously does so.</p>
6988
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00006989</div>
6990
6991<!-- _______________________________________________________________________ -->
6992<div class="doc_subsubsection">
6993 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6994</div>
6995<div class="doc_text">
6996<h5>Syntax:</h5>
6997<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00006998 declare void @llvm.memory.barrier(i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;, i1 &lt;device&gt;)
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00006999</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007000
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007001<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007002<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7003 specific pairs of memory access types.</p>
7004
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007005<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007006<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7007 The first four arguments enables a specific barrier as listed below. The
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00007008 fifth argument specifies that the barrier applies to io or device or uncached
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007009 memory.</p>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007010
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007011<ul>
7012 <li><tt>ll</tt>: load-load barrier</li>
7013 <li><tt>ls</tt>: load-store barrier</li>
7014 <li><tt>sl</tt>: store-load barrier</li>
7015 <li><tt>ss</tt>: store-store barrier</li>
7016 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7017</ul>
7018
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007019<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007020<p>This intrinsic causes the system to enforce some ordering constraints upon
7021 the loads and stores of the program. This barrier does not
7022 indicate <em>when</em> any events will occur, it only enforces
7023 an <em>order</em> in which they occur. For any of the specified pairs of load
7024 and store operations (f.ex. load-load, or store-load), all of the first
7025 operations preceding the barrier will complete before any of the second
7026 operations succeeding the barrier begin. Specifically the semantics for each
7027 pairing is as follows:</p>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007028
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007029<ul>
7030 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7031 after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007032 <li><tt>ls</tt>: All loads before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007033 store after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007034 <li><tt>ss</tt>: All stores before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007035 store after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007036 <li><tt>sl</tt>: All stores before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007037 load after the barrier begins.</li>
7038</ul>
7039
7040<p>These semantics are applied with a logical "and" behavior when more than one
7041 is enabled in a single memory barrier intrinsic.</p>
7042
7043<p>Backends may implement stronger barriers than those requested when they do
7044 not support as fine grained a barrier as requested. Some architectures do
7045 not need all types of barriers and on such architectures, these become
7046 noops.</p>
7047
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007048<h5>Example:</h5>
7049<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007050%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7051%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007052 store i32 4, %ptr
7053
7054%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007055 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false)
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007056 <i>; guarantee the above finishes</i>
7057 store i32 8, %ptr <i>; before this begins</i>
7058</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007059
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007060</div>
7061
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007062<!-- _______________________________________________________________________ -->
7063<div class="doc_subsubsection">
Mon P Wang28873102008-06-25 08:15:39 +00007064 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007065</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007066
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007067<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007068
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007069<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007070<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7071 any integer bit width and for different address spaces. Not all targets
7072 support all bit widths however.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007073
7074<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007075 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt;)
7076 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt;)
7077 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt;)
7078 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007079</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007080
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007081<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007082<p>This loads a value in memory and compares it to a given value. If they are
7083 equal, it stores a new value into the memory.</p>
7084
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007085<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007086<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7087 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7088 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7089 this integer type. While any bit width integer may be used, targets may only
7090 lower representations they support in hardware.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007091
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007092<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007093<p>This entire intrinsic must be executed atomically. It first loads the value
7094 in memory pointed to by <tt>ptr</tt> and compares it with the
7095 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7096 memory. The loaded value is yielded in all cases. This provides the
7097 equivalent of an atomic compare-and-swap operation within the SSA
7098 framework.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007099
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007100<h5>Examples:</h5>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007101<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007102%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7103%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007104 store i32 4, %ptr
7105
7106%val1 = add i32 4, 4
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007107%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007108 <i>; yields {i32}:result1 = 4</i>
7109%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7110%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7111
7112%val2 = add i32 1, 1
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007113%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007114 <i>; yields {i32}:result2 = 8</i>
7115%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7116
7117%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7118</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007119
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007120</div>
7121
7122<!-- _______________________________________________________________________ -->
7123<div class="doc_subsubsection">
7124 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7125</div>
7126<div class="doc_text">
7127<h5>Syntax:</h5>
7128
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007129<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7130 integer bit width. Not all targets support all bit widths however.</p>
7131
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007132<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007133 declare i8 @llvm.atomic.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;val&gt;)
7134 declare i16 @llvm.atomic.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;val&gt;)
7135 declare i32 @llvm.atomic.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;val&gt;)
7136 declare i64 @llvm.atomic.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;val&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007137</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007138
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007139<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007140<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7141 the value from memory. It then stores the value in <tt>val</tt> in the memory
7142 at <tt>ptr</tt>.</p>
7143
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007144<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007145<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7146 the <tt>val</tt> argument and the result must be integers of the same bit
7147 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7148 integer type. The targets may only lower integer representations they
7149 support.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007150
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007151<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007152<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7153 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7154 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007155
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007156<h5>Examples:</h5>
7157<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007158%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7159%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007160 store i32 4, %ptr
7161
7162%val1 = add i32 4, 4
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007163%result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007164 <i>; yields {i32}:result1 = 4</i>
7165%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7166%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7167
7168%val2 = add i32 1, 1
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007169%result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007170 <i>; yields {i32}:result2 = 8</i>
7171
7172%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7173%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7174</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007175
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007176</div>
7177
7178<!-- _______________________________________________________________________ -->
7179<div class="doc_subsubsection">
Mon P Wang28873102008-06-25 08:15:39 +00007180 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007181
7182</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007183
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007184<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007185
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007186<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007187<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7188 any integer bit width. Not all targets support all bit widths however.</p>
7189
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007190<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007191 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7192 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7193 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7194 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007195</pre>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007196
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007197<h5>Overview:</h5>
7198<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7199 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7200
7201<h5>Arguments:</h5>
7202<p>The intrinsic takes two arguments, the first a pointer to an integer value
7203 and the second an integer value. The result is also an integer value. These
7204 integer types can have any bit width, but they must all have the same bit
7205 width. The targets may only lower integer representations they support.</p>
7206
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007207<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007208<p>This intrinsic does a series of operations atomically. It first loads the
7209 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7210 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007211
7212<h5>Examples:</h5>
7213<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007214%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7215%ptr = bitcast i8* %mallocP to i32*
7216 store i32 4, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007217%result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007218 <i>; yields {i32}:result1 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007219%result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007220 <i>; yields {i32}:result2 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007221%result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007222 <i>; yields {i32}:result3 = 10</i>
Mon P Wang28873102008-06-25 08:15:39 +00007223%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007224</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007225
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007226</div>
7227
Mon P Wang28873102008-06-25 08:15:39 +00007228<!-- _______________________________________________________________________ -->
7229<div class="doc_subsubsection">
7230 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
7231
7232</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007233
Mon P Wang28873102008-06-25 08:15:39 +00007234<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007235
Mon P Wang28873102008-06-25 08:15:39 +00007236<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007237<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
7238 any integer bit width and for different address spaces. Not all targets
7239 support all bit widths however.</p>
7240
Mon P Wang28873102008-06-25 08:15:39 +00007241<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007242 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7243 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7244 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7245 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007246</pre>
Mon P Wang28873102008-06-25 08:15:39 +00007247
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007248<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007249<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007250 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7251
7252<h5>Arguments:</h5>
7253<p>The intrinsic takes two arguments, the first a pointer to an integer value
7254 and the second an integer value. The result is also an integer value. These
7255 integer types can have any bit width, but they must all have the same bit
7256 width. The targets may only lower integer representations they support.</p>
7257
Mon P Wang28873102008-06-25 08:15:39 +00007258<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007259<p>This intrinsic does a series of operations atomically. It first loads the
7260 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
7261 result to <tt>ptr</tt>. It yields the original value stored
7262 at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007263
7264<h5>Examples:</h5>
7265<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007266%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7267%ptr = bitcast i8* %mallocP to i32*
7268 store i32 8, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007269%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
Mon P Wang28873102008-06-25 08:15:39 +00007270 <i>; yields {i32}:result1 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007271%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
Mon P Wang28873102008-06-25 08:15:39 +00007272 <i>; yields {i32}:result2 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007273%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
Mon P Wang28873102008-06-25 08:15:39 +00007274 <i>; yields {i32}:result3 = 2</i>
7275%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
7276</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007277
Mon P Wang28873102008-06-25 08:15:39 +00007278</div>
7279
7280<!-- _______________________________________________________________________ -->
7281<div class="doc_subsubsection">
7282 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
7283 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
7284 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
7285 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
Mon P Wang28873102008-06-25 08:15:39 +00007286</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007287
Mon P Wang28873102008-06-25 08:15:39 +00007288<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007289
Mon P Wang28873102008-06-25 08:15:39 +00007290<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007291<p>These are overloaded intrinsics. You can
7292 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
7293 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
7294 bit width and for different address spaces. Not all targets support all bit
7295 widths however.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007296
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007297<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007298 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7299 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7300 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7301 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007302</pre>
7303
7304<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007305 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7306 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7307 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7308 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007309</pre>
7310
7311<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007312 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7313 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7314 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7315 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007316</pre>
7317
7318<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007319 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7320 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7321 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7322 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007323</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007324
Mon P Wang28873102008-06-25 08:15:39 +00007325<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007326<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
7327 the value stored in memory at <tt>ptr</tt>. It yields the original value
7328 at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007329
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007330<h5>Arguments:</h5>
7331<p>These intrinsics take two arguments, the first a pointer to an integer value
7332 and the second an integer value. The result is also an integer value. These
7333 integer types can have any bit width, but they must all have the same bit
7334 width. The targets may only lower integer representations they support.</p>
7335
Mon P Wang28873102008-06-25 08:15:39 +00007336<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007337<p>These intrinsics does a series of operations atomically. They first load the
7338 value stored at <tt>ptr</tt>. They then do the bitwise
7339 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
7340 original value stored at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007341
7342<h5>Examples:</h5>
7343<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007344%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7345%ptr = bitcast i8* %mallocP to i32*
7346 store i32 0x0F0F, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007347%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang28873102008-06-25 08:15:39 +00007348 <i>; yields {i32}:result0 = 0x0F0F</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007349%result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang28873102008-06-25 08:15:39 +00007350 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007351%result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang28873102008-06-25 08:15:39 +00007352 <i>; yields {i32}:result2 = 0xF0</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007353%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang28873102008-06-25 08:15:39 +00007354 <i>; yields {i32}:result3 = FF</i>
7355%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
7356</pre>
Mon P Wang28873102008-06-25 08:15:39 +00007357
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007358</div>
Mon P Wang28873102008-06-25 08:15:39 +00007359
7360<!-- _______________________________________________________________________ -->
7361<div class="doc_subsubsection">
7362 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
7363 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
7364 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
7365 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
Mon P Wang28873102008-06-25 08:15:39 +00007366</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007367
Mon P Wang28873102008-06-25 08:15:39 +00007368<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007369
Mon P Wang28873102008-06-25 08:15:39 +00007370<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007371<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
7372 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
7373 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
7374 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007375
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007376<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007377 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7378 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7379 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7380 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007381</pre>
7382
7383<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007384 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7385 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7386 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7387 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007388</pre>
7389
7390<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007391 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7392 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7393 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7394 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007395</pre>
7396
7397<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007398 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7399 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7400 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7401 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007402</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007403
Mon P Wang28873102008-06-25 08:15:39 +00007404<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007405<p>These intrinsics takes the signed or unsigned minimum or maximum of
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007406 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7407 original value at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007408
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007409<h5>Arguments:</h5>
7410<p>These intrinsics take two arguments, the first a pointer to an integer value
7411 and the second an integer value. The result is also an integer value. These
7412 integer types can have any bit width, but they must all have the same bit
7413 width. The targets may only lower integer representations they support.</p>
7414
Mon P Wang28873102008-06-25 08:15:39 +00007415<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007416<p>These intrinsics does a series of operations atomically. They first load the
7417 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7418 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7419 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007420
7421<h5>Examples:</h5>
7422<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007423%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7424%ptr = bitcast i8* %mallocP to i32*
7425 store i32 7, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007426%result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
Mon P Wang28873102008-06-25 08:15:39 +00007427 <i>; yields {i32}:result0 = 7</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007428%result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
Mon P Wang28873102008-06-25 08:15:39 +00007429 <i>; yields {i32}:result1 = -2</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007430%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
Mon P Wang28873102008-06-25 08:15:39 +00007431 <i>; yields {i32}:result2 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007432%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
Mon P Wang28873102008-06-25 08:15:39 +00007433 <i>; yields {i32}:result3 = 8</i>
7434%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7435</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007436
Mon P Wang28873102008-06-25 08:15:39 +00007437</div>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007438
Nick Lewyckycc271862009-10-13 07:03:23 +00007439
7440<!-- ======================================================================= -->
7441<div class="doc_subsection">
7442 <a name="int_memorymarkers">Memory Use Markers</a>
7443</div>
7444
7445<div class="doc_text">
7446
7447<p>This class of intrinsics exists to information about the lifetime of memory
7448 objects and ranges where variables are immutable.</p>
7449
7450</div>
7451
7452<!-- _______________________________________________________________________ -->
7453<div class="doc_subsubsection">
7454 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7455</div>
7456
7457<div class="doc_text">
7458
7459<h5>Syntax:</h5>
7460<pre>
7461 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7462</pre>
7463
7464<h5>Overview:</h5>
7465<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7466 object's lifetime.</p>
7467
7468<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +00007469<p>The first argument is a constant integer representing the size of the
7470 object, or -1 if it is variable sized. The second argument is a pointer to
7471 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007472
7473<h5>Semantics:</h5>
7474<p>This intrinsic indicates that before this point in the code, the value of the
7475 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
Nick Lewycky8d336592009-10-27 16:56:58 +00007476 never be used and has an undefined value. A load from the pointer that
7477 precedes this intrinsic can be replaced with
Nick Lewyckycc271862009-10-13 07:03:23 +00007478 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7479
7480</div>
7481
7482<!-- _______________________________________________________________________ -->
7483<div class="doc_subsubsection">
7484 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7485</div>
7486
7487<div class="doc_text">
7488
7489<h5>Syntax:</h5>
7490<pre>
7491 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7492</pre>
7493
7494<h5>Overview:</h5>
7495<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7496 object's lifetime.</p>
7497
7498<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +00007499<p>The first argument is a constant integer representing the size of the
7500 object, or -1 if it is variable sized. The second argument is a pointer to
7501 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007502
7503<h5>Semantics:</h5>
7504<p>This intrinsic indicates that after this point in the code, the value of the
7505 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7506 never be used and has an undefined value. Any stores into the memory object
7507 following this intrinsic may be removed as dead.
7508
7509</div>
7510
7511<!-- _______________________________________________________________________ -->
7512<div class="doc_subsubsection">
7513 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7514</div>
7515
7516<div class="doc_text">
7517
7518<h5>Syntax:</h5>
7519<pre>
7520 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;) readonly
7521</pre>
7522
7523<h5>Overview:</h5>
7524<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7525 a memory object will not change.</p>
7526
7527<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +00007528<p>The first argument is a constant integer representing the size of the
7529 object, or -1 if it is variable sized. The second argument is a pointer to
7530 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007531
7532<h5>Semantics:</h5>
7533<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7534 the return value, the referenced memory location is constant and
7535 unchanging.</p>
7536
7537</div>
7538
7539<!-- _______________________________________________________________________ -->
7540<div class="doc_subsubsection">
7541 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7542</div>
7543
7544<div class="doc_text">
7545
7546<h5>Syntax:</h5>
7547<pre>
7548 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7549</pre>
7550
7551<h5>Overview:</h5>
7552<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7553 a memory object are mutable.</p>
7554
7555<h5>Arguments:</h5>
7556<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky321333e2009-10-13 07:57:33 +00007557 The second argument is a constant integer representing the size of the
7558 object, or -1 if it is variable sized and the third argument is a pointer
7559 to the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007560
7561<h5>Semantics:</h5>
7562<p>This intrinsic indicates that the memory is mutable again.</p>
7563
7564</div>
7565
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007566<!-- ======================================================================= -->
7567<div class="doc_subsection">
Tanya Lattner6d806e92007-06-15 20:50:54 +00007568 <a name="int_general">General Intrinsics</a>
7569</div>
7570
7571<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007572
7573<p>This class of intrinsics is designed to be generic and has no specific
7574 purpose.</p>
7575
Tanya Lattner6d806e92007-06-15 20:50:54 +00007576</div>
7577
7578<!-- _______________________________________________________________________ -->
7579<div class="doc_subsubsection">
7580 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7581</div>
7582
7583<div class="doc_text">
7584
7585<h5>Syntax:</h5>
7586<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007587 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 +00007588</pre>
7589
7590<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007591<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +00007592
7593<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007594<p>The first argument is a pointer to a value, the second is a pointer to a
7595 global string, the third is a pointer to a global string which is the source
7596 file name, and the last argument is the line number.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +00007597
7598<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007599<p>This intrinsic allows annotation of local variables with arbitrary strings.
7600 This can be useful for special purpose optimizations that want to look for
7601 these annotations. These have no other defined use, they are ignored by code
7602 generation and optimization.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +00007603
Tanya Lattner6d806e92007-06-15 20:50:54 +00007604</div>
7605
Tanya Lattnerb6367882007-09-21 22:59:12 +00007606<!-- _______________________________________________________________________ -->
7607<div class="doc_subsubsection">
Tanya Lattnere1a8da02007-09-21 23:57:59 +00007608 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007609</div>
7610
7611<div class="doc_text">
7612
7613<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007614<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7615 any integer bit width.</p>
7616
Tanya Lattnerb6367882007-09-21 22:59:12 +00007617<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007618 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7619 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7620 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7621 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7622 declare i256 @llvm.annotation.i256(i256 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Tanya Lattnerb6367882007-09-21 22:59:12 +00007623</pre>
7624
7625<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007626<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007627
7628<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007629<p>The first argument is an integer value (result of some expression), the
7630 second is a pointer to a global string, the third is a pointer to a global
7631 string which is the source file name, and the last argument is the line
7632 number. It returns the value of the first argument.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007633
7634<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007635<p>This intrinsic allows annotations to be put on arbitrary expressions with
7636 arbitrary strings. This can be useful for special purpose optimizations that
7637 want to look for these annotations. These have no other defined use, they
7638 are ignored by code generation and optimization.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007639
Tanya Lattnerb6367882007-09-21 22:59:12 +00007640</div>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +00007641
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007642<!-- _______________________________________________________________________ -->
7643<div class="doc_subsubsection">
7644 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7645</div>
7646
7647<div class="doc_text">
7648
7649<h5>Syntax:</h5>
7650<pre>
7651 declare void @llvm.trap()
7652</pre>
7653
7654<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007655<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007656
7657<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007658<p>None.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007659
7660<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007661<p>This intrinsics is lowered to the target dependent trap instruction. If the
7662 target does not have a trap instruction, this intrinsic will be lowered to
7663 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007664
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007665</div>
7666
Bill Wendling69e4adb2008-11-19 05:56:17 +00007667<!-- _______________________________________________________________________ -->
7668<div class="doc_subsubsection">
Misha Brukmandccb0252008-11-22 23:55:29 +00007669 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendling69e4adb2008-11-19 05:56:17 +00007670</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007671
Bill Wendling69e4adb2008-11-19 05:56:17 +00007672<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007673
Bill Wendling69e4adb2008-11-19 05:56:17 +00007674<h5>Syntax:</h5>
7675<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007676 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling69e4adb2008-11-19 05:56:17 +00007677</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007678
Bill Wendling69e4adb2008-11-19 05:56:17 +00007679<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007680<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7681 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7682 ensure that it is placed on the stack before local variables.</p>
7683
Bill Wendling69e4adb2008-11-19 05:56:17 +00007684<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007685<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7686 arguments. The first argument is the value loaded from the stack
7687 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7688 that has enough space to hold the value of the guard.</p>
7689
Bill Wendling69e4adb2008-11-19 05:56:17 +00007690<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007691<p>This intrinsic causes the prologue/epilogue inserter to force the position of
7692 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7693 stack. This is to ensure that if a local variable on the stack is
7694 overwritten, it will destroy the value of the guard. When the function exits,
7695 the guard on the stack is checked against the original guard. If they're
7696 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7697 function.</p>
7698
Bill Wendling69e4adb2008-11-19 05:56:17 +00007699</div>
7700
Eric Christopher0e671492009-11-30 08:03:53 +00007701<!-- _______________________________________________________________________ -->
7702<div class="doc_subsubsection">
7703 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7704</div>
7705
7706<div class="doc_text">
7707
7708<h5>Syntax:</h5>
7709<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007710 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;type&gt;)
7711 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;type&gt;)
Eric Christopher0e671492009-11-30 08:03:53 +00007712</pre>
7713
7714<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007715<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
Eric Christopherd003c5b2010-01-08 21:42:39 +00007716 to the optimizers to discover at compile time either a) when an
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007717 operation like memcpy will either overflow a buffer that corresponds to
7718 an object, or b) to determine that a runtime check for overflow isn't
7719 necessary. An object in this context means an allocation of a
Eric Christopher8295a0a2009-12-23 00:29:49 +00007720 specific class, structure, array, or other object.</p>
Eric Christopher0e671492009-11-30 08:03:53 +00007721
7722<h5>Arguments:</h5>
7723<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher8295a0a2009-12-23 00:29:49 +00007724 argument is a pointer to or into the <tt>object</tt>. The second argument
7725 is a boolean 0 or 1. This argument determines whether you want the
7726 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7727 1, variables are not allowed.</p>
7728
Eric Christopher0e671492009-11-30 08:03:53 +00007729<h5>Semantics:</h5>
7730<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007731 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7732 (depending on the <tt>type</tt> argument if the size cannot be determined
7733 at compile time.</p>
Eric Christopher0e671492009-11-30 08:03:53 +00007734
7735</div>
7736
Chris Lattner00950542001-06-06 20:29:01 +00007737<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +00007738<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007739<address>
7740 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +00007741 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007742 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +00007743 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007744
7745 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +00007746 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007747 Last modified: $Date$
7748</address>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +00007749
Misha Brukman9d0919f2003-11-08 01:05:38 +00007750</body>
7751</html>