blob: 893876620a46a0f98d9a1927c06d3a2b70e383cf [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 Wendling55ae5152010-08-20 22:05:50 +000028 <li><a href="#linkage_linker_private_weak_def_auto">'<tt>linker_private_weak_def_auto</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +000029 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
30 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
31 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
32 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
33 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
35 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner5a2d8752009-10-10 18:26:06 +000036 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +000037 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
38 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
40 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendling3d10a5a2009-07-20 01:03:30 +000041 </ol>
42 </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +000043 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnere7886e42009-01-11 20:53:49 +000044 <li><a href="#namedtypes">Named Types</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000045 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +000046 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +000047 <li><a href="#aliasstructure">Aliases</a></li>
Devang Patelcd1fd252010-01-11 19:35:55 +000048 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
Reid Spencerca86e162006-12-31 07:07:53 +000049 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel2c9c3e72008-09-26 23:51:19 +000050 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +000051 <li><a href="#gc">Garbage Collector Names</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +000052 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencerde151942007-02-19 23:54:10 +000053 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman556ca272009-07-27 18:07:55 +000054 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +000055 <li><a href="#volatile">Volatile Memory Accesses</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +000056 </ol>
57 </li>
Chris Lattner00950542001-06-06 20:29:01 +000058 <li><a href="#typesystem">Type System</a>
59 <ol>
Chris Lattner4f69f462008-01-04 04:32:38 +000060 <li><a href="#t_classifications">Type Classifications</a></li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +000061 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner261efe92003-11-25 01:02:51 +000062 <ol>
Nick Lewyckyec38da42009-09-27 00:45:11 +000063 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner4f69f462008-01-04 04:32:38 +000064 <li><a href="#t_floating">Floating Point Types</a></li>
Dale Johannesen21fe99b2010-10-01 00:48:59 +000065 <li><a href="#t_x86mmx">X86mmx Type</a></li>
Chris Lattner4f69f462008-01-04 04:32:38 +000066 <li><a href="#t_void">Void Type</a></li>
67 <li><a href="#t_label">Label Type</a></li>
Nick Lewycky7a0370f2009-05-30 05:06:04 +000068 <li><a href="#t_metadata">Metadata Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000069 </ol>
70 </li>
Chris Lattner00950542001-06-06 20:29:01 +000071 <li><a href="#t_derived">Derived Types</a>
72 <ol>
Chris Lattnerfdfeb692010-02-12 20:49:41 +000073 <li><a href="#t_aggregate">Aggregate Types</a>
74 <ol>
75 <li><a href="#t_array">Array Type</a></li>
76 <li><a href="#t_struct">Structure Type</a></li>
77 <li><a href="#t_pstruct">Packed Structure Type</a></li>
Chris Lattnerfdfeb692010-02-12 20:49:41 +000078 <li><a href="#t_vector">Vector Type</a></li>
79 </ol>
80 </li>
Misha Brukman9d0919f2003-11-08 01:05:38 +000081 <li><a href="#t_function">Function Type</a></li>
82 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner69c11bb2005-04-25 17:34:15 +000083 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000084 </ol>
85 </li>
Chris Lattner242d61d2009-02-02 07:32:36 +000086 <li><a href="#t_uprefs">Type Up-references</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +000087 </ol>
88 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +000089 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +000090 <ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +000091 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner70882792009-02-28 18:32:25 +000092 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +000093 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
94 <li><a href="#undefvalues">Undefined Values</a></li>
Dan Gohmanfff6c532010-04-22 23:14:21 +000095 <li><a href="#trapvalues">Trap Values</a></li>
Chris Lattnerf9d078e2009-10-27 21:19:13 +000096 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +000097 <li><a href="#constantexprs">Constant Expressions</a></li>
Chris Lattnerc3f59762004-12-09 17:30:23 +000098 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +000099 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000100 <li><a href="#othervalues">Other Values</a>
101 <ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000102 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Devang Patelcd1fd252010-01-11 19:35:55 +0000103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000104 </ol>
105 </li>
Chris Lattner857755c2009-07-20 05:55:19 +0000106 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
107 <ol>
108 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner401e10c2009-07-20 06:14:25 +0000109 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
110 Global Variable</a></li>
Chris Lattner857755c2009-07-20 05:55:19 +0000111 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
112 Global Variable</a></li>
113 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
114 Global Variable</a></li>
115 </ol>
116 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000117 <li><a href="#instref">Instruction Reference</a>
118 <ol>
119 <li><a href="#terminators">Terminator Instructions</a>
120 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000121 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
122 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000123 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerab21db72009-10-28 00:19:10 +0000124 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000125 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000126 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000127 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000128 </ol>
129 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000130 <li><a href="#binaryops">Binary Operations</a>
131 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000132 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000133 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000134 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000135 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000136 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000137 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000138 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
139 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
140 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000141 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
142 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
143 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000144 </ol>
145 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000146 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
147 <ol>
Reid Spencer8e11bf82007-02-02 13:57:07 +0000148 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
149 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
150 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000151 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000152 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000153 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000154 </ol>
155 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000156 <li><a href="#vectorops">Vector Operations</a>
157 <ol>
158 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
159 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
160 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000161 </ol>
162 </li>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000163 <li><a href="#aggregateops">Aggregate Operations</a>
164 <ol>
165 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
166 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
167 </ol>
168 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000169 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000170 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000171 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000172 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
173 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
174 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000175 </ol>
176 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000177 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000178 <ol>
179 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
180 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
181 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
182 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
183 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000184 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
185 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
186 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
187 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000188 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
189 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000190 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000191 </ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000192 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000193 <li><a href="#otherops">Other Operations</a>
194 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000195 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
196 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000197 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000198 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000199 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000200 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000201 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000202 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000203 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000204 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000205 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000206 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000207 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
208 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000209 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
210 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
211 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000212 </ol>
213 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000214 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
215 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000216 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
217 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
218 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000219 </ol>
220 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000221 <li><a href="#int_codegen">Code Generator Intrinsics</a>
222 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000223 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
224 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
225 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
226 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
227 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
228 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohman31f1af12010-05-26 21:56:15 +0000229 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000230 </ol>
231 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000232 <li><a href="#int_libc">Standard C Library Intrinsics</a>
233 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000234 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
236 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
237 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
238 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman91c284c2007-10-15 20:30:11 +0000239 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
241 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000242 </ol>
243 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000244 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000245 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000246 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000247 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
248 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
249 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000250 </ol>
251 </li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000252 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
253 <ol>
Bill Wendlingda01af72009-02-08 04:04:40 +0000254 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
255 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
256 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
257 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
258 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendling41b485c2009-02-08 23:00:09 +0000259 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000260 </ol>
261 </li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000262 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
263 <ol>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000264 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
265 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000266 </ol>
267 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000268 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000269 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sandsf7331b32007-09-11 14:10:23 +0000270 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands36397f52007-07-27 12:58:54 +0000271 <ol>
272 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands36397f52007-07-27 12:58:54 +0000273 </ol>
274 </li>
Bill Wendling3c44f5b2008-11-18 22:10:53 +0000275 <li><a href="#int_atomics">Atomic intrinsics</a>
276 <ol>
277 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
278 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
279 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
280 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
281 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
282 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
283 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
284 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
285 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
286 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
287 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
288 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
289 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
290 </ol>
291 </li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000292 <li><a href="#int_memorymarkers">Memory Use Markers</a>
293 <ol>
294 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
295 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
296 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
297 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
298 </ol>
299 </li>
Reid Spencer20677642007-07-20 19:59:11 +0000300 <li><a href="#int_general">General intrinsics</a>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000301 <ol>
Reid Spencer20677642007-07-20 19:59:11 +0000302 <li><a href="#int_var_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000303 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000304 <li><a href="#int_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000305 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000306 <li><a href="#int_trap">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000307 '<tt>llvm.trap</tt>' Intrinsic</a></li>
308 <li><a href="#int_stackprotector">
309 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher0e671492009-11-30 08:03:53 +0000310 <li><a href="#int_objectsize">
311 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000312 </ol>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000313 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000314 </ol>
315 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000316</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000317
318<div class="doc_author">
319 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
320 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000321</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000322
Chris Lattner00950542001-06-06 20:29:01 +0000323<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000324<div class="doc_section"> <a name="abstract">Abstract </a></div>
325<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000326
Misha Brukman9d0919f2003-11-08 01:05:38 +0000327<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000328
329<p>This document is a reference manual for the LLVM assembly language. LLVM is
330 a Static Single Assignment (SSA) based representation that provides type
331 safety, low-level operations, flexibility, and the capability of representing
332 'all' high-level languages cleanly. It is the common code representation
333 used throughout all phases of the LLVM compilation strategy.</p>
334
Misha Brukman9d0919f2003-11-08 01:05:38 +0000335</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000336
Chris Lattner00950542001-06-06 20:29:01 +0000337<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000338<div class="doc_section"> <a name="introduction">Introduction</a> </div>
339<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000340
Misha Brukman9d0919f2003-11-08 01:05:38 +0000341<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000342
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000343<p>The LLVM code representation is designed to be used in three different forms:
344 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
345 for fast loading by a Just-In-Time compiler), and as a human readable
346 assembly language representation. This allows LLVM to provide a powerful
347 intermediate representation for efficient compiler transformations and
348 analysis, while providing a natural means to debug and visualize the
349 transformations. The three different forms of LLVM are all equivalent. This
350 document describes the human readable representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000351
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000352<p>The LLVM representation aims to be light-weight and low-level while being
353 expressive, typed, and extensible at the same time. It aims to be a
354 "universal IR" of sorts, by being at a low enough level that high-level ideas
355 may be cleanly mapped to it (similar to how microprocessors are "universal
356 IR's", allowing many source languages to be mapped to them). By providing
357 type information, LLVM can be used as the target of optimizations: for
358 example, through pointer analysis, it can be proven that a C automatic
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000359 variable is never accessed outside of the current function, allowing it to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000360 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000361
Misha Brukman9d0919f2003-11-08 01:05:38 +0000362</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000363
Chris Lattner00950542001-06-06 20:29:01 +0000364<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +0000365<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000366
Misha Brukman9d0919f2003-11-08 01:05:38 +0000367<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000368
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000369<p>It is important to note that this document describes 'well formed' LLVM
370 assembly language. There is a difference between what the parser accepts and
371 what is considered 'well formed'. For example, the following instruction is
372 syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000373
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000374<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000375%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000376</pre>
377
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000378<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
379 LLVM infrastructure provides a verification pass that may be used to verify
380 that an LLVM module is well formed. This pass is automatically run by the
381 parser after parsing input assembly and by the optimizer before it outputs
382 bitcode. The violations pointed out by the verifier pass indicate bugs in
383 transformation passes or input to the parser.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000384
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000385</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000386
Chris Lattnercc689392007-10-03 17:34:29 +0000387<!-- Describe the typesetting conventions here. -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000388
Chris Lattner00950542001-06-06 20:29:01 +0000389<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +0000390<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner00950542001-06-06 20:29:01 +0000391<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000392
Misha Brukman9d0919f2003-11-08 01:05:38 +0000393<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +0000394
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000395<p>LLVM identifiers come in two basic types: global and local. Global
396 identifiers (functions, global variables) begin with the <tt>'@'</tt>
397 character. Local identifiers (register names, types) begin with
398 the <tt>'%'</tt> character. Additionally, there are three different formats
399 for identifiers, for different purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000400
Chris Lattner00950542001-06-06 20:29:01 +0000401<ol>
Reid Spencer2c452282007-08-07 14:34:28 +0000402 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000403 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
404 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
405 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
406 other characters in their names can be surrounded with quotes. Special
407 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
408 ASCII code for the character in hexadecimal. In this way, any character
409 can be used in a name value, even quotes themselves.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000410
Reid Spencer2c452282007-08-07 14:34:28 +0000411 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000412 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000413
Reid Spencercc16dc32004-12-09 18:02:53 +0000414 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000415 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000416</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000417
Reid Spencer2c452282007-08-07 14:34:28 +0000418<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000419 don't need to worry about name clashes with reserved words, and the set of
420 reserved words may be expanded in the future without penalty. Additionally,
421 unnamed identifiers allow a compiler to quickly come up with a temporary
422 variable without having to avoid symbol table conflicts.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000423
Chris Lattner261efe92003-11-25 01:02:51 +0000424<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000425 languages. There are keywords for different opcodes
426 ('<tt><a href="#i_add">add</a></tt>',
427 '<tt><a href="#i_bitcast">bitcast</a></tt>',
428 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
429 ('<tt><a href="#t_void">void</a></tt>',
430 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
431 reserved words cannot conflict with variable names, because none of them
432 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000433
434<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000435 '<tt>%X</tt>' by 8:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000436
Misha Brukman9d0919f2003-11-08 01:05:38 +0000437<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000438
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000439<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000440%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000441</pre>
442
Misha Brukman9d0919f2003-11-08 01:05:38 +0000443<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000444
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000445<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000446%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000447</pre>
448
Misha Brukman9d0919f2003-11-08 01:05:38 +0000449<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000450
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000451<pre class="doc_code">
Gabor Greifec58f752009-10-28 13:05:07 +0000452%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
453%1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000454%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000455</pre>
456
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000457<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
458 lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000459
Chris Lattner00950542001-06-06 20:29:01 +0000460<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000461 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000462 line.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000463
464 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000465 assigned to a named value.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000466
Misha Brukman9d0919f2003-11-08 01:05:38 +0000467 <li>Unnamed temporaries are numbered sequentially</li>
468</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000469
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000470<p>It also shows a convention that we follow in this document. When
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000471 demonstrating instructions, we will follow an instruction with a comment that
472 defines the type and name of value produced. Comments are shown in italic
473 text.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000474
Misha Brukman9d0919f2003-11-08 01:05:38 +0000475</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000476
477<!-- *********************************************************************** -->
478<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
479<!-- *********************************************************************** -->
480
481<!-- ======================================================================= -->
482<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
483</div>
484
485<div class="doc_text">
486
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000487<p>LLVM programs are composed of "Module"s, each of which is a translation unit
488 of the input programs. Each module consists of functions, global variables,
489 and symbol table entries. Modules may be combined together with the LLVM
490 linker, which merges function (and global variable) definitions, resolves
491 forward declarations, and merges symbol table entries. Here is an example of
492 the "hello world" module:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000493
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000494<pre class="doc_code">
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000495<i>; Declare the string constant as a global constant.</i>&nbsp;
Nick Lewyckydb9cd762011-01-29 01:09:53 +0000496<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a>&nbsp;<a href="#globalvars">constant</a>&nbsp;<a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000497
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000498<i>; External declaration of the puts function</i>&nbsp;
499<a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000500
501<i>; Definition of main function</i>
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000502define i32 @main() { <i>; i32()* </i>&nbsp;
503 <i>; Convert [13 x i8]* to i8 *...</i>&nbsp;
504 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000505
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000506 <i>; Call puts function to write out the string to stdout.</i>&nbsp;
507 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>&nbsp;
508 <a href="#i_ret">ret</a> i32 0&nbsp;
509}
Devang Patelcd1fd252010-01-11 19:35:55 +0000510
511<i>; Named metadata</i>
512!1 = metadata !{i32 41}
513!foo = !{!1, null}
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000514</pre>
Chris Lattnerfa730212004-12-09 16:11:40 +0000515
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000516<p>This example is made up of a <a href="#globalvars">global variable</a> named
Devang Patelcd1fd252010-01-11 19:35:55 +0000517 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000518 a <a href="#functionstructure">function definition</a> for
Devang Patelcd1fd252010-01-11 19:35:55 +0000519 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
520 "<tt>foo"</tt>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000521
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000522<p>In general, a module is made up of a list of global values, where both
523 functions and global variables are global values. Global values are
524 represented by a pointer to a memory location (in this case, a pointer to an
525 array of char, and a pointer to a function), and have one of the
526 following <a href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000527
Chris Lattnere5d947b2004-12-09 16:36:40 +0000528</div>
529
530<!-- ======================================================================= -->
531<div class="doc_subsection">
532 <a name="linkage">Linkage Types</a>
533</div>
534
535<div class="doc_text">
536
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000537<p>All Global Variables and Functions have one of the following types of
538 linkage:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000539
540<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000541 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000542 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
543 by objects in the current module. In particular, linking code into a
544 module with an private global value may cause the private to be renamed as
545 necessary to avoid collisions. Because the symbol is private to the
546 module, all references can be updated. This doesn't show up in any symbol
547 table in the object file.</dd>
Rafael Espindolabb46f522009-01-15 20:18:42 +0000548
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000549 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000550 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
551 assembler and evaluated by the linker. Unlike normal strong symbols, they
552 are removed by the linker from the final linked image (executable or
553 dynamic library).</dd>
554
555 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
556 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
557 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
558 linker. The symbols are removed by the linker from the final linked image
559 (executable or dynamic library).</dd>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000560
Bill Wendling55ae5152010-08-20 22:05:50 +0000561 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
562 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
563 of the object is not taken. For instance, functions that had an inline
564 definition, but the compiler decided not to inline it. Note,
565 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
566 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
567 visibility. The symbols are removed by the linker from the final linked
568 image (executable or dynamic library).</dd>
569
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000570 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling07d31772010-06-29 22:34:52 +0000571 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000572 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
573 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000574
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000575 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000576 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000577 into the object file corresponding to the LLVM module. They exist to
578 allow inlining and other optimizations to take place given knowledge of
579 the definition of the global, which is known to be somewhere outside the
580 module. Globals with <tt>available_externally</tt> linkage are allowed to
581 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
582 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000583
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000584 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattner4887bd82007-01-14 06:51:48 +0000585 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner873187c2010-01-09 19:15:14 +0000586 the same name when linkage occurs. This can be used to implement
587 some forms of inline functions, templates, or other code which must be
588 generated in each translation unit that uses it, but where the body may
589 be overridden with a more definitive definition later. Unreferenced
590 <tt>linkonce</tt> globals are allowed to be discarded. Note that
591 <tt>linkonce</tt> linkage does not actually allow the optimizer to
592 inline the body of this function into callers because it doesn't know if
593 this definition of the function is the definitive definition within the
594 program or whether it will be overridden by a stronger definition.
595 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
596 linkage.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000597
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000598 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000599 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
600 <tt>linkonce</tt> linkage, except that unreferenced globals with
601 <tt>weak</tt> linkage may not be discarded. This is used for globals that
602 are declared "weak" in C source code.</dd>
603
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000604 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000605 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
606 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
607 global scope.
608 Symbols with "<tt>common</tt>" linkage are merged in the same way as
609 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000610 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000611 must have a zero initializer, and may not be marked '<a
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000612 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
613 have common linkage.</dd>
Chris Lattner26d054d2009-08-05 05:21:07 +0000614
Chris Lattnere5d947b2004-12-09 16:36:40 +0000615
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000616 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000617 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000618 pointer to array type. When two global variables with appending linkage
619 are linked together, the two global arrays are appended together. This is
620 the LLVM, typesafe, equivalent of having the system linker append together
621 "sections" with identical names when .o files are linked.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000622
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000623 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000624 <dd>The semantics of this linkage follow the ELF object file model: the symbol
625 is weak until linked, if not linked, the symbol becomes null instead of
626 being an undefined reference.</dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000627
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000628 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
629 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000630 <dd>Some languages allow differing globals to be merged, such as two functions
631 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling5e721d72010-07-01 21:55:59 +0000632 that only equivalent globals are ever merged (the "one definition rule"
633 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000634 and <tt>weak_odr</tt> linkage types to indicate that the global will only
635 be merged with equivalent globals. These linkage types are otherwise the
636 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands667d4b82009-03-07 15:45:40 +0000637
Chris Lattnerfa730212004-12-09 16:11:40 +0000638 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000639 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000640 visible, meaning that it participates in linkage and can be used to
641 resolve external symbol references.</dd>
Reid Spencerc8910842007-04-11 23:49:50 +0000642</dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000643
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000644<p>The next two types of linkage are targeted for Microsoft Windows platform
645 only. They are designed to support importing (exporting) symbols from (to)
646 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000647
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000648<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000649 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000650 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000651 or variable via a global pointer to a pointer that is set up by the DLL
652 exporting the symbol. On Microsoft Windows targets, the pointer name is
653 formed by combining <code>__imp_</code> and the function or variable
654 name.</dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000655
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000656 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000657 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000658 pointer to a pointer in a DLL, so that it can be referenced with the
659 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
660 name is formed by combining <code>__imp_</code> and the function or
661 variable name.</dd>
Chris Lattnerfa730212004-12-09 16:11:40 +0000662</dl>
663
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000664<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
665 another module defined a "<tt>.LC0</tt>" variable and was linked with this
666 one, one of the two would be renamed, preventing a collision. Since
667 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
668 declarations), they are accessible outside of the current module.</p>
669
670<p>It is illegal for a function <i>declaration</i> to have any linkage type
671 other than "externally visible", <tt>dllimport</tt>
672 or <tt>extern_weak</tt>.</p>
673
Duncan Sands667d4b82009-03-07 15:45:40 +0000674<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000675 or <tt>weak_odr</tt> linkages.</p>
676
Chris Lattnerfa730212004-12-09 16:11:40 +0000677</div>
678
679<!-- ======================================================================= -->
680<div class="doc_subsection">
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000681 <a name="callingconv">Calling Conventions</a>
682</div>
683
684<div class="doc_text">
685
686<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000687 and <a href="#i_invoke">invokes</a> can all have an optional calling
688 convention specified for the call. The calling convention of any pair of
689 dynamic caller/callee must match, or the behavior of the program is
690 undefined. The following calling conventions are supported by LLVM, and more
691 may be added in the future:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000692
693<dl>
694 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000695 <dd>This calling convention (the default if no other calling convention is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000696 specified) matches the target C calling conventions. This calling
697 convention supports varargs function calls and tolerates some mismatch in
698 the declared prototype and implemented declaration of the function (as
699 does normal C).</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000700
701 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000702 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000703 (e.g. by passing things in registers). This calling convention allows the
704 target to use whatever tricks it wants to produce fast code for the
705 target, without having to conform to an externally specified ABI
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000706 (Application Binary Interface).
707 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattner29689432010-03-11 00:22:57 +0000708 when this or the GHC convention is used.</a> This calling convention
709 does not support varargs and requires the prototype of all callees to
710 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000711
712 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000713 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000714 as possible under the assumption that the call is not commonly executed.
715 As such, these calls often preserve all registers so that the call does
716 not break any live ranges in the caller side. This calling convention
717 does not support varargs and requires the prototype of all callees to
718 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000719
Chris Lattner29689432010-03-11 00:22:57 +0000720 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
721 <dd>This calling convention has been implemented specifically for use by the
722 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
723 It passes everything in registers, going to extremes to achieve this by
724 disabling callee save registers. This calling convention should not be
725 used lightly but only for specific situations such as an alternative to
726 the <em>register pinning</em> performance technique often used when
727 implementing functional programming languages.At the moment only X86
728 supports this convention and it has the following limitations:
729 <ul>
730 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
731 floating point types are supported.</li>
732 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
733 6 floating point parameters.</li>
734 </ul>
735 This calling convention supports
736 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
737 requires both the caller and callee are using it.
738 </dd>
739
Chris Lattnercfe6b372005-05-07 01:46:40 +0000740 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000741 <dd>Any calling convention may be specified by number, allowing
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000742 target-specific calling conventions to be used. Target specific calling
743 conventions start at 64.</dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000744</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000745
746<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000747 support Pascal conventions or any other well-known target-independent
748 convention.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000749
750</div>
751
752<!-- ======================================================================= -->
753<div class="doc_subsection">
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000754 <a name="visibility">Visibility Styles</a>
755</div>
756
757<div class="doc_text">
758
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000759<p>All Global Variables and Functions have one of the following visibility
760 styles:</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000761
762<dl>
763 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000764 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000765 that the declaration is visible to other modules and, in shared libraries,
766 means that the declared entity may be overridden. On Darwin, default
767 visibility means that the declaration is visible to other modules. Default
768 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000769
770 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000771 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000772 object if they are in the same shared object. Usually, hidden visibility
773 indicates that the symbol will not be placed into the dynamic symbol
774 table, so no other module (executable or shared library) can reference it
775 directly.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000776
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000777 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000778 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000779 the dynamic symbol table, but that references within the defining module
780 will bind to the local symbol. That is, the symbol cannot be overridden by
781 another module.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000782</dl>
783
784</div>
785
786<!-- ======================================================================= -->
787<div class="doc_subsection">
Chris Lattnere7886e42009-01-11 20:53:49 +0000788 <a name="namedtypes">Named Types</a>
789</div>
790
791<div class="doc_text">
792
793<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000794 it easier to read the IR and make the IR more condensed (particularly when
795 recursive types are involved). An example of a name specification is:</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000796
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000797<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +0000798%mytype = type { %mytype*, i32 }
799</pre>
Chris Lattnere7886e42009-01-11 20:53:49 +0000800
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000801<p>You may give a name to any <a href="#typesystem">type</a> except
Chris Lattnerdc65f222010-08-17 23:26:04 +0000802 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000803 is expected with the syntax "%mytype".</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000804
805<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000806 and that you can therefore specify multiple names for the same type. This
807 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
808 uses structural typing, the name is not part of the type. When printing out
809 LLVM IR, the printer will pick <em>one name</em> to render all types of a
810 particular shape. This means that if you have code where two different
811 source types end up having the same LLVM type, that the dumper will sometimes
812 print the "wrong" or unexpected type. This is an important design point and
813 isn't going to change.</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000814
815</div>
816
Chris Lattnere7886e42009-01-11 20:53:49 +0000817<!-- ======================================================================= -->
818<div class="doc_subsection">
Chris Lattnerfa730212004-12-09 16:11:40 +0000819 <a name="globalvars">Global Variables</a>
820</div>
821
822<div class="doc_text">
823
Chris Lattner3689a342005-02-12 19:30:21 +0000824<p>Global variables define regions of memory allocated at compilation time
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000825 instead of run-time. Global variables may optionally be initialized, may
826 have an explicit section to be placed in, and may have an optional explicit
827 alignment specified. A variable may be defined as "thread_local", which
828 means that it will not be shared by threads (each thread will have a
829 separated copy of the variable). A variable may be defined as a global
830 "constant," which indicates that the contents of the variable
831 will <b>never</b> be modified (enabling better optimization, allowing the
832 global data to be placed in the read-only section of an executable, etc).
833 Note that variables that need runtime initialization cannot be marked
834 "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000835
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000836<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
837 constant, even if the final definition of the global is not. This capability
838 can be used to enable slightly better optimization of the program, but
839 requires the language definition to guarantee that optimizations based on the
840 'constantness' are valid for the translation units that do not include the
841 definition.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000842
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000843<p>As SSA values, global variables define pointer values that are in scope
844 (i.e. they dominate) all basic blocks in the program. Global variables
845 always define a pointer to their "content" type because they describe a
846 region of memory, and all memory objects in LLVM are accessed through
847 pointers.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000848
Rafael Espindolabea46262011-01-08 16:42:36 +0000849<p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
850 that the address is not significant, only the content. Constants marked
Rafael Espindolaa5eaa862011-01-15 08:20:57 +0000851 like this can be merged with other constants if they have the same
852 initializer. Note that a constant with significant address <em>can</em>
853 be merged with a <tt>unnamed_addr</tt> constant, the result being a
854 constant whose address is significant.</p>
Rafael Espindolabea46262011-01-08 16:42:36 +0000855
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000856<p>A global variable may be declared to reside in a target-specific numbered
857 address space. For targets that support them, address spaces may affect how
858 optimizations are performed and/or what target instructions are used to
859 access the variable. The default address space is zero. The address space
860 qualifier must precede any other attributes.</p>
Christopher Lamb284d9922007-12-11 09:31:00 +0000861
Chris Lattner88f6c462005-11-12 00:45:07 +0000862<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000863 supports it, it will emit globals to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000864
Chris Lattnerce99fa92010-04-28 00:13:42 +0000865<p>An explicit alignment may be specified for a global, which must be a power
866 of 2. If not present, or if the alignment is set to zero, the alignment of
867 the global is set by the target to whatever it feels convenient. If an
868 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner2d4b8ee2010-04-28 00:31:12 +0000869 alignment. Targets and optimizers are not allowed to over-align the global
870 if the global has an assigned section. In this case, the extra alignment
871 could be observable: for example, code could assume that the globals are
872 densely packed in their section and try to iterate over them as an array,
873 alignment padding would break this iteration.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000874
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000875<p>For example, the following defines a global in a numbered address space with
876 an initializer, section, and alignment:</p>
Chris Lattner68027ea2007-01-14 00:27:09 +0000877
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000878<pre class="doc_code">
Dan Gohman398873c2009-01-11 00:40:00 +0000879@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner68027ea2007-01-14 00:27:09 +0000880</pre>
881
Chris Lattnerfa730212004-12-09 16:11:40 +0000882</div>
883
884
885<!-- ======================================================================= -->
886<div class="doc_subsection">
887 <a name="functionstructure">Functions</a>
888</div>
889
890<div class="doc_text">
891
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000892<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000893 optional <a href="#linkage">linkage type</a>, an optional
894 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000895 <a href="#callingconv">calling convention</a>,
896 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000897 <a href="#paramattrs">parameter attribute</a> for the return type, a function
898 name, a (possibly empty) argument list (each with optional
899 <a href="#paramattrs">parameter attributes</a>), optional
900 <a href="#fnattrs">function attributes</a>, an optional section, an optional
901 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
902 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000903
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000904<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
905 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000906 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000907 <a href="#callingconv">calling convention</a>,
908 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000909 <a href="#paramattrs">parameter attribute</a> for the return type, a function
910 name, a possibly empty list of arguments, an optional alignment, and an
911 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000912
Chris Lattnerd3eda892008-08-05 18:29:16 +0000913<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000914 (Control Flow Graph) for the function. Each basic block may optionally start
915 with a label (giving the basic block a symbol table entry), contains a list
916 of instructions, and ends with a <a href="#terminators">terminator</a>
917 instruction (such as a branch or function return).</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000918
Chris Lattner4a3c9012007-06-08 16:52:14 +0000919<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000920 executed on entrance to the function, and it is not allowed to have
921 predecessor basic blocks (i.e. there can not be any branches to the entry
922 block of a function). Because the block can have no predecessors, it also
923 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000924
Chris Lattner88f6c462005-11-12 00:45:07 +0000925<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000926 supports it, it will emit functions to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000927
Chris Lattner2cbdc452005-11-06 08:02:57 +0000928<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000929 the alignment is set to zero, the alignment of the function is set by the
930 target to whatever it feels convenient. If an explicit alignment is
931 specified, the function is forced to have at least that much alignment. All
932 alignments must be a power of 2.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000933
Rafael Espindolabea46262011-01-08 16:42:36 +0000934<p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
935 be significant and two identical functions can be merged</p>.
936
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000937<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000938<pre class="doc_code">
Chris Lattner50ad45c2008-10-13 16:55:18 +0000939define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000940 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
941 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
942 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
943 [<a href="#gc">gc</a>] { ... }
944</pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000945
Chris Lattnerfa730212004-12-09 16:11:40 +0000946</div>
947
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000948<!-- ======================================================================= -->
949<div class="doc_subsection">
950 <a name="aliasstructure">Aliases</a>
951</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000952
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000953<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000954
955<p>Aliases act as "second name" for the aliasee value (which can be either
956 function, global variable, another alias or bitcast of global value). Aliases
957 may have an optional <a href="#linkage">linkage type</a>, and an
958 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000959
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000960<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000961<pre class="doc_code">
Duncan Sands0b23ac12008-09-12 20:48:21 +0000962@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendlingaac388b2007-05-29 09:42:13 +0000963</pre>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000964
965</div>
966
Chris Lattner4e9aba72006-01-23 23:23:47 +0000967<!-- ======================================================================= -->
Devang Patelcd1fd252010-01-11 19:35:55 +0000968<div class="doc_subsection">
969 <a name="namedmetadatastructure">Named Metadata</a>
970</div>
971
972<div class="doc_text">
973
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000974<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
Dan Gohman872814a2010-07-21 18:54:18 +0000975 nodes</a> (but not metadata strings) are the only valid operands for
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000976 a named metadata.</p>
Devang Patelcd1fd252010-01-11 19:35:55 +0000977
978<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000979<pre class="doc_code">
Dan Gohman872814a2010-07-21 18:54:18 +0000980; Some unnamed metadata nodes, which are referenced by the named metadata.
981!0 = metadata !{metadata !"zero"}
Devang Patelcd1fd252010-01-11 19:35:55 +0000982!1 = metadata !{metadata !"one"}
Dan Gohman872814a2010-07-21 18:54:18 +0000983!2 = metadata !{metadata !"two"}
Dan Gohman1005bc52010-07-13 19:48:13 +0000984; A named metadata.
Dan Gohman872814a2010-07-21 18:54:18 +0000985!name = !{!0, !1, !2}
Devang Patelcd1fd252010-01-11 19:35:55 +0000986</pre>
Devang Patelcd1fd252010-01-11 19:35:55 +0000987
988</div>
989
990<!-- ======================================================================= -->
Reid Spencerca86e162006-12-31 07:07:53 +0000991<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Reid Spencerca86e162006-12-31 07:07:53 +0000992
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000993<div class="doc_text">
994
995<p>The return type and each parameter of a function type may have a set of
996 <i>parameter attributes</i> associated with them. Parameter attributes are
997 used to communicate additional information about the result or parameters of
998 a function. Parameter attributes are considered to be part of the function,
999 not of the function type, so functions with different parameter attributes
1000 can have the same function type.</p>
1001
1002<p>Parameter attributes are simple keywords that follow the type specified. If
1003 multiple parameter attributes are needed, they are space separated. For
1004 example:</p>
Bill Wendling2f7a8b02007-05-29 09:04:49 +00001005
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001006<pre class="doc_code">
Nick Lewyckyb6a7d252009-02-15 23:06:14 +00001007declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattner66d922c2008-10-04 18:33:34 +00001008declare i32 @atoi(i8 zeroext)
1009declare signext i8 @returns_signed_char()
Bill Wendling2f7a8b02007-05-29 09:04:49 +00001010</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +00001011
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001012<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1013 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerca86e162006-12-31 07:07:53 +00001014
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001015<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner47507de2008-01-11 06:20:47 +00001016
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001017<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001018 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001019 <dd>This indicates to the code generator that the parameter or return value
1020 should be zero-extended to a 32-bit value by the caller (for a parameter)
1021 or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +00001022
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001023 <dt><tt><b>signext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001024 <dd>This indicates to the code generator that the parameter or return value
1025 should be sign-extended to a 32-bit value by the caller (for a parameter)
1026 or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +00001027
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001028 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001029 <dd>This indicates that this parameter or return value should be treated in a
1030 special target-dependent fashion during while emitting code for a function
1031 call or return (usually, by putting it in a register as opposed to memory,
1032 though some targets use it to distinguish between two different kinds of
1033 registers). Use of this attribute is target-specific.</dd>
Chris Lattner47507de2008-01-11 06:20:47 +00001034
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001035 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Chris Lattnera6fd81d2010-11-20 23:49:06 +00001036 <dd><p>This indicates that the pointer parameter should really be passed by
1037 value to the function. The attribute implies that a hidden copy of the
1038 pointee
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001039 is made between the caller and the callee, so the callee is unable to
1040 modify the value in the callee. This attribute is only valid on LLVM
1041 pointer arguments. It is generally used to pass structs and arrays by
1042 value, but is also valid on pointers to scalars. The copy is considered
1043 to belong to the caller not the callee (for example,
1044 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1045 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnera6fd81d2010-11-20 23:49:06 +00001046 values.</p>
1047
1048 <p>The byval attribute also supports specifying an alignment with
1049 the align attribute. It indicates the alignment of the stack slot to
1050 form and the known alignment of the pointer specified to the call site. If
1051 the alignment is not specified, then the code generator makes a
1052 target-specific assumption.</p></dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001053
Dan Gohmanff235352010-07-02 23:18:08 +00001054 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001055 <dd>This indicates that the pointer parameter specifies the address of a
1056 structure that is the return value of the function in the source program.
1057 This pointer must be guaranteed by the caller to be valid: loads and
1058 stores to the structure may be assumed by the callee to not to trap. This
1059 may only be applied to the first parameter. This is not a valid attribute
1060 for return values. </dd>
1061
Dan Gohmanff235352010-07-02 23:18:08 +00001062 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohman1e109622010-07-02 18:41:32 +00001063 <dd>This indicates that pointer values
1064 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmanefca7f92010-07-02 23:46:54 +00001065 value do not alias pointer values which are not <i>based</i> on it,
1066 ignoring certain "irrelevant" dependencies.
1067 For a call to the parent function, dependencies between memory
1068 references from before or after the call and from those during the call
1069 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1070 return value used in that call.
Dan Gohman1e109622010-07-02 18:41:32 +00001071 The caller shares the responsibility with the callee for ensuring that
1072 these requirements are met.
1073 For further details, please see the discussion of the NoAlias response in
Dan Gohmanff70fe42010-07-06 15:26:33 +00001074 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1075<br>
John McCall191d4ee2010-07-06 21:07:14 +00001076 Note that this definition of <tt>noalias</tt> is intentionally
1077 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner211244a2010-07-06 20:51:35 +00001078 arguments, though it is slightly weaker.
Dan Gohmanff70fe42010-07-06 15:26:33 +00001079<br>
1080 For function return values, C99's <tt>restrict</tt> is not meaningful,
1081 while LLVM's <tt>noalias</tt> is.
1082 </dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001083
Dan Gohmanff235352010-07-02 23:18:08 +00001084 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001085 <dd>This indicates that the callee does not make any copies of the pointer
1086 that outlive the callee itself. This is not a valid attribute for return
1087 values.</dd>
1088
Dan Gohmanff235352010-07-02 23:18:08 +00001089 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001090 <dd>This indicates that the pointer parameter can be excised using the
1091 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1092 attribute for return values.</dd>
1093</dl>
Reid Spencerca86e162006-12-31 07:07:53 +00001094
Reid Spencerca86e162006-12-31 07:07:53 +00001095</div>
1096
1097<!-- ======================================================================= -->
Chris Lattner4e9aba72006-01-23 23:23:47 +00001098<div class="doc_subsection">
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001099 <a name="gc">Garbage Collector Names</a>
1100</div>
1101
1102<div class="doc_text">
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001103
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001104<p>Each function may specify a garbage collector name, which is simply a
1105 string:</p>
1106
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001107<pre class="doc_code">
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001108define void @f() gc "name" { ... }
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001109</pre>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001110
1111<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001112 collector which will cause the compiler to alter its output in order to
1113 support the named garbage collection algorithm.</p>
1114
Gordon Henriksen80a75bf2007-12-10 03:18:06 +00001115</div>
1116
1117<!-- ======================================================================= -->
1118<div class="doc_subsection">
Devang Patel2c9c3e72008-09-26 23:51:19 +00001119 <a name="fnattrs">Function Attributes</a>
Devang Patelf8b94812008-09-04 23:05:13 +00001120</div>
1121
1122<div class="doc_text">
Devang Patel2c9c3e72008-09-26 23:51:19 +00001123
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001124<p>Function attributes are set to communicate additional information about a
1125 function. Function attributes are considered to be part of the function, not
1126 of the function type, so functions with different parameter attributes can
1127 have the same function type.</p>
Devang Patel2c9c3e72008-09-26 23:51:19 +00001128
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001129<p>Function attributes are simple keywords that follow the type specified. If
1130 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelf8b94812008-09-04 23:05:13 +00001131
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001132<pre class="doc_code">
Devang Patel2c9c3e72008-09-26 23:51:19 +00001133define void @f() noinline { ... }
1134define void @f() alwaysinline { ... }
1135define void @f() alwaysinline optsize { ... }
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001136define void @f() optsize { ... }
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001137</pre>
Devang Patelf8b94812008-09-04 23:05:13 +00001138
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001139<dl>
Charles Davis1e063d12010-02-12 00:31:15 +00001140 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1141 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1142 the backend should forcibly align the stack pointer. Specify the
1143 desired alignment, which must be a power of two, in parentheses.
1144
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001145 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001146 <dd>This attribute indicates that the inliner should attempt to inline this
1147 function into callers whenever possible, ignoring any active inlining size
1148 threshold for this caller.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001149
Charles Davis970bfcc2010-10-25 15:37:09 +00001150 <dt><tt><b>hotpatch</b></tt></dt>
Charles Davis6f12e292010-10-25 16:29:03 +00001151 <dd>This attribute indicates that the function should be 'hotpatchable',
Charles Davis0076d202010-10-25 19:07:39 +00001152 meaning the function can be patched and/or hooked even while it is
1153 loaded into memory. On x86, the function prologue will be preceded
1154 by six bytes of padding and will begin with a two-byte instruction.
1155 Most of the functions in the Windows system DLLs in Windows XP SP2 or
1156 higher were compiled in this fashion.</dd>
Charles Davis970bfcc2010-10-25 15:37:09 +00001157
Jakob Stoklund Olesen570a4a52010-02-06 01:16:28 +00001158 <dt><tt><b>inlinehint</b></tt></dt>
1159 <dd>This attribute indicates that the source code contained a hint that inlining
1160 this function is desirable (such as the "inline" keyword in C/C++). It
1161 is just a hint; it imposes no requirements on the inliner.</dd>
1162
Nick Lewycky76ec37a2010-07-06 18:24:09 +00001163 <dt><tt><b>naked</b></tt></dt>
1164 <dd>This attribute disables prologue / epilogue emission for the function.
1165 This can have very system-specific consequences.</dd>
1166
1167 <dt><tt><b>noimplicitfloat</b></tt></dt>
1168 <dd>This attributes disables implicit floating point instructions.</dd>
1169
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001170 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001171 <dd>This attribute indicates that the inliner should never inline this
1172 function in any situation. This attribute may not be used together with
1173 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +00001174
Nick Lewycky76ec37a2010-07-06 18:24:09 +00001175 <dt><tt><b>noredzone</b></tt></dt>
1176 <dd>This attribute indicates that the code generator should not use a red
1177 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +00001178
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001179 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001180 <dd>This function attribute indicates that the function never returns
1181 normally. This produces undefined behavior at runtime if the function
1182 ever does dynamically return.</dd>
Bill Wendling31359ba2008-11-13 01:02:51 +00001183
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001184 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001185 <dd>This function attribute indicates that the function never returns with an
1186 unwind or exceptional control flow. If the function does unwind, its
1187 runtime behavior is undefined.</dd>
Bill Wendlingfbaa7ed2008-11-26 19:07:40 +00001188
Nick Lewycky76ec37a2010-07-06 18:24:09 +00001189 <dt><tt><b>optsize</b></tt></dt>
1190 <dd>This attribute suggests that optimization passes and code generator passes
1191 make choices that keep the code size of this function low, and otherwise
1192 do optimizations specifically to reduce code size.</dd>
1193
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001194 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001195 <dd>This attribute indicates that the function computes its result (or decides
1196 to unwind an exception) based strictly on its arguments, without
1197 dereferencing any pointer arguments or otherwise accessing any mutable
1198 state (e.g. memory, control registers, etc) visible to caller functions.
1199 It does not write through any pointer arguments
1200 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1201 changes any state visible to callers. This means that it cannot unwind
1202 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1203 could use the <tt>unwind</tt> instruction.</dd>
Devang Patel5d96fda2009-06-12 19:45:19 +00001204
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001205 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001206 <dd>This attribute indicates that the function does not write through any
1207 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1208 arguments) or otherwise modify any state (e.g. memory, control registers,
1209 etc) visible to caller functions. It may dereference pointer arguments
1210 and read state that may be set in the caller. A readonly function always
1211 returns the same value (or unwinds an exception identically) when called
1212 with the same set of arguments and global state. It cannot unwind an
1213 exception by calling the <tt>C++</tt> exception throwing methods, but may
1214 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovc5ec8a72009-07-17 18:07:26 +00001215
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001216 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001217 <dd>This attribute indicates that the function should emit a stack smashing
1218 protector. It is in the form of a "canary"&mdash;a random value placed on
1219 the stack before the local variables that's checked upon return from the
1220 function to see if it has been overwritten. A heuristic is used to
1221 determine if a function needs stack protectors or not.<br>
1222<br>
1223 If a function that has an <tt>ssp</tt> attribute is inlined into a
1224 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1225 function will have an <tt>ssp</tt> attribute.</dd>
1226
Bill Wendlingf82d40a2009-11-02 00:24:16 +00001227 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001228 <dd>This attribute indicates that the function should <em>always</em> emit a
1229 stack smashing protector. This overrides
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001230 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1231<br>
1232 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1233 function that doesn't have an <tt>sspreq</tt> attribute or which has
1234 an <tt>ssp</tt> attribute, then the resulting function will have
1235 an <tt>sspreq</tt> attribute.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001236</dl>
1237
Devang Patelf8b94812008-09-04 23:05:13 +00001238</div>
1239
1240<!-- ======================================================================= -->
1241<div class="doc_subsection">
Chris Lattner1eeeb0c2006-04-08 04:40:53 +00001242 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001243</div>
1244
1245<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001246
1247<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1248 the GCC "file scope inline asm" blocks. These blocks are internally
1249 concatenated by LLVM and treated as a single unit, but may be separated in
1250 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001251
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001252<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00001253module asm "inline asm code goes here"
1254module asm "more can go here"
1255</pre>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001256
1257<p>The strings can contain any character by escaping non-printable characters.
1258 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001259 for the number.</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +00001260
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001261<p>The inline asm code is simply printed to the machine code .s file when
1262 assembly code is generated.</p>
1263
Chris Lattner4e9aba72006-01-23 23:23:47 +00001264</div>
Chris Lattnerfa730212004-12-09 16:11:40 +00001265
Reid Spencerde151942007-02-19 23:54:10 +00001266<!-- ======================================================================= -->
1267<div class="doc_subsection">
1268 <a name="datalayout">Data Layout</a>
1269</div>
1270
1271<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001272
Reid Spencerde151942007-02-19 23:54:10 +00001273<p>A module may specify a target specific data layout string that specifies how
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001274 data is to be laid out in memory. The syntax for the data layout is
1275 simply:</p>
1276
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00001277<pre class="doc_code">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001278target datalayout = "<i>layout specification</i>"
1279</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001280
1281<p>The <i>layout specification</i> consists of a list of specifications
1282 separated by the minus sign character ('-'). Each specification starts with
1283 a letter and may include other information after the letter to define some
1284 aspect of the data layout. The specifications accepted are as follows:</p>
1285
Reid Spencerde151942007-02-19 23:54:10 +00001286<dl>
1287 <dt><tt>E</tt></dt>
1288 <dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001289 bits with the most significance have the lowest address location.</dd>
1290
Reid Spencerde151942007-02-19 23:54:10 +00001291 <dt><tt>e</tt></dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +00001292 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001293 the bits with the least significance have the lowest address
1294 location.</dd>
1295
Reid Spencerde151942007-02-19 23:54:10 +00001296 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001297 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001298 <i>preferred</i> alignments. All sizes are in bits. Specifying
1299 the <i>pref</i> alignment is optional. If omitted, the
1300 preceding <tt>:</tt> should be omitted too.</dd>
1301
Reid Spencerde151942007-02-19 23:54:10 +00001302 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1303 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001304 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1305
Reid Spencerde151942007-02-19 23:54:10 +00001306 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001307 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001308 <i>size</i>.</dd>
1309
Reid Spencerde151942007-02-19 23:54:10 +00001310 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001311 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesen9d8d2212010-05-28 18:54:47 +00001312 <i>size</i>. Only values of <i>size</i> that are supported by the target
1313 will work. 32 (float) and 64 (double) are supported on all targets;
1314 80 or 128 (different flavors of long double) are also supported on some
1315 targets.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001316
Reid Spencerde151942007-02-19 23:54:10 +00001317 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1318 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001319 <i>size</i>.</dd>
1320
Daniel Dunbar87bde0b2009-06-08 22:17:53 +00001321 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1322 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001323 <i>size</i>.</dd>
Chris Lattnere82bdc42009-11-07 09:35:34 +00001324
1325 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1326 <dd>This specifies a set of native integer widths for the target CPU
1327 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1328 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001329 this set are considered to support most general arithmetic
Chris Lattnere82bdc42009-11-07 09:35:34 +00001330 operations efficiently.</dd>
Reid Spencerde151942007-02-19 23:54:10 +00001331</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001332
Reid Spencerde151942007-02-19 23:54:10 +00001333<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman1c70c002010-04-28 00:36:01 +00001334 default set of specifications which are then (possibly) overridden by the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001335 specifications in the <tt>datalayout</tt> keyword. The default specifications
1336 are given in this list:</p>
1337
Reid Spencerde151942007-02-19 23:54:10 +00001338<ul>
1339 <li><tt>E</tt> - big endian</li>
Dan Gohmanfdf2e8c2010-02-23 02:44:03 +00001340 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
Reid Spencerde151942007-02-19 23:54:10 +00001341 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1342 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1343 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1344 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattnerd3eda892008-08-05 18:29:16 +00001345 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Reid Spencerde151942007-02-19 23:54:10 +00001346 alignment of 64-bits</li>
1347 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1348 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1349 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1350 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1351 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar87bde0b2009-06-08 22:17:53 +00001352 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencerde151942007-02-19 23:54:10 +00001353</ul>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001354
1355<p>When LLVM is determining the alignment for a given type, it uses the
1356 following rules:</p>
1357
Reid Spencerde151942007-02-19 23:54:10 +00001358<ol>
1359 <li>If the type sought is an exact match for one of the specifications, that
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001360 specification is used.</li>
1361
Reid Spencerde151942007-02-19 23:54:10 +00001362 <li>If no match is found, and the type sought is an integer type, then the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001363 smallest integer type that is larger than the bitwidth of the sought type
1364 is used. If none of the specifications are larger than the bitwidth then
1365 the the largest integer type is used. For example, given the default
1366 specifications above, the i7 type will use the alignment of i8 (next
1367 largest) while both i65 and i256 will use the alignment of i64 (largest
1368 specified).</li>
1369
Reid Spencerde151942007-02-19 23:54:10 +00001370 <li>If no match is found, and the type sought is a vector type, then the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001371 largest vector type that is smaller than the sought vector type will be
1372 used as a fall back. This happens because &lt;128 x double&gt; can be
1373 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencerde151942007-02-19 23:54:10 +00001374</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001375
Reid Spencerde151942007-02-19 23:54:10 +00001376</div>
Chris Lattnerfa730212004-12-09 16:11:40 +00001377
Dan Gohman556ca272009-07-27 18:07:55 +00001378<!-- ======================================================================= -->
1379<div class="doc_subsection">
1380 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1381</div>
1382
1383<div class="doc_text">
1384
Andreas Bolka55e459a2009-07-29 00:02:05 +00001385<p>Any memory access must be done through a pointer value associated
Andreas Bolka99a82052009-07-27 20:37:10 +00001386with an address range of the memory access, otherwise the behavior
Dan Gohman556ca272009-07-27 18:07:55 +00001387is undefined. Pointer values are associated with address ranges
1388according to the following rules:</p>
1389
1390<ul>
Dan Gohman1e109622010-07-02 18:41:32 +00001391 <li>A pointer value is associated with the addresses associated with
1392 any value it is <i>based</i> on.
Andreas Bolka55e459a2009-07-29 00:02:05 +00001393 <li>An address of a global variable is associated with the address
Dan Gohman556ca272009-07-27 18:07:55 +00001394 range of the variable's storage.</li>
1395 <li>The result value of an allocation instruction is associated with
1396 the address range of the allocated storage.</li>
1397 <li>A null pointer in the default address-space is associated with
Andreas Bolka55e459a2009-07-29 00:02:05 +00001398 no address.</li>
Dan Gohman556ca272009-07-27 18:07:55 +00001399 <li>An integer constant other than zero or a pointer value returned
1400 from a function not defined within LLVM may be associated with address
1401 ranges allocated through mechanisms other than those provided by
Andreas Bolka55e459a2009-07-29 00:02:05 +00001402 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman556ca272009-07-27 18:07:55 +00001403 allocated by mechanisms provided by LLVM.</li>
Dan Gohman1e109622010-07-02 18:41:32 +00001404</ul>
1405
1406<p>A pointer value is <i>based</i> on another pointer value according
1407 to the following rules:</p>
1408
1409<ul>
1410 <li>A pointer value formed from a
1411 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1412 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1413 <li>The result value of a
1414 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1415 of the <tt>bitcast</tt>.</li>
1416 <li>A pointer value formed by an
1417 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1418 pointer values that contribute (directly or indirectly) to the
1419 computation of the pointer's value.</li>
1420 <li>The "<i>based</i> on" relationship is transitive.</li>
1421</ul>
1422
1423<p>Note that this definition of <i>"based"</i> is intentionally
1424 similar to the definition of <i>"based"</i> in C99, though it is
1425 slightly weaker.</p>
Dan Gohman556ca272009-07-27 18:07:55 +00001426
1427<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +00001428<tt><a href="#i_load">load</a></tt> merely indicates the size and
1429alignment of the memory from which to load, as well as the
Dan Gohmanc22c0f32010-06-17 19:23:50 +00001430interpretation of the value. The first operand type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +00001431<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1432and alignment of the store.</p>
Dan Gohman556ca272009-07-27 18:07:55 +00001433
1434<p>Consequently, type-based alias analysis, aka TBAA, aka
1435<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1436LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1437additional information which specialized optimization passes may use
1438to implement type-based alias analysis.</p>
1439
1440</div>
1441
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00001442<!-- ======================================================================= -->
1443<div class="doc_subsection">
1444 <a name="volatile">Volatile Memory Accesses</a>
1445</div>
1446
1447<div class="doc_text">
1448
1449<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1450href="#i_store"><tt>store</tt></a>s, and <a
1451href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1452The optimizers must not change the number of volatile operations or change their
1453order of execution relative to other volatile operations. The optimizers
1454<i>may</i> change the order of volatile operations relative to non-volatile
1455operations. This is not Java's "volatile" and has no cross-thread
1456synchronization behavior.</p>
1457
1458</div>
1459
Chris Lattner00950542001-06-06 20:29:01 +00001460<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00001461<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1462<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +00001463
Misha Brukman9d0919f2003-11-08 01:05:38 +00001464<div class="doc_text">
Chris Lattnerfa730212004-12-09 16:11:40 +00001465
Misha Brukman9d0919f2003-11-08 01:05:38 +00001466<p>The LLVM type system is one of the most important features of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001467 intermediate representation. Being typed enables a number of optimizations
1468 to be performed on the intermediate representation directly, without having
1469 to do extra analyses on the side before the transformation. A strong type
1470 system makes it easier to read the generated code and enables novel analyses
1471 and transformations that are not feasible to perform on normal three address
1472 code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +00001473
1474</div>
1475
Chris Lattner00950542001-06-06 20:29:01 +00001476<!-- ======================================================================= -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001477<div class="doc_subsection"> <a name="t_classifications">Type
Chris Lattner261efe92003-11-25 01:02:51 +00001478Classifications</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001479
Misha Brukman9d0919f2003-11-08 01:05:38 +00001480<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001481
1482<p>The types fall into a few useful classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001483
1484<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00001485 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001486 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +00001487 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +00001488 <td><a href="#t_integer">integer</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +00001489 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +00001490 </tr>
1491 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +00001492 <td><a href="#t_floating">floating point</a></td>
1493 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +00001494 </tr>
1495 <tr>
1496 <td><a name="t_firstclass">first class</a></td>
Chris Lattner4f69f462008-01-04 04:32:38 +00001497 <td><a href="#t_integer">integer</a>,
1498 <a href="#t_floating">floating point</a>,
1499 <a href="#t_pointer">pointer</a>,
Dan Gohman0066db62008-06-18 18:42:13 +00001500 <a href="#t_vector">vector</a>,
Dan Gohmana334d5f2008-05-12 23:51:09 +00001501 <a href="#t_struct">structure</a>,
1502 <a href="#t_array">array</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001503 <a href="#t_label">label</a>,
1504 <a href="#t_metadata">metadata</a>.
Reid Spencerca86e162006-12-31 07:07:53 +00001505 </td>
Chris Lattner261efe92003-11-25 01:02:51 +00001506 </tr>
Chris Lattner4f69f462008-01-04 04:32:38 +00001507 <tr>
1508 <td><a href="#t_primitive">primitive</a></td>
1509 <td><a href="#t_label">label</a>,
1510 <a href="#t_void">void</a>,
Tobias Grosser05387292010-12-28 20:29:31 +00001511 <a href="#t_integer">integer</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001512 <a href="#t_floating">floating point</a>,
Dale Johannesen21fe99b2010-10-01 00:48:59 +00001513 <a href="#t_x86mmx">x86mmx</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001514 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner4f69f462008-01-04 04:32:38 +00001515 </tr>
1516 <tr>
1517 <td><a href="#t_derived">derived</a></td>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001518 <td><a href="#t_array">array</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +00001519 <a href="#t_function">function</a>,
1520 <a href="#t_pointer">pointer</a>,
1521 <a href="#t_struct">structure</a>,
1522 <a href="#t_pstruct">packed structure</a>,
1523 <a href="#t_vector">vector</a>,
1524 <a href="#t_opaque">opaque</a>.
Dan Gohman01ac1012008-10-14 16:32:04 +00001525 </td>
Chris Lattner4f69f462008-01-04 04:32:38 +00001526 </tr>
Chris Lattner261efe92003-11-25 01:02:51 +00001527 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001528</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001529
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001530<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1531 important. Values of these types are the only ones which can be produced by
Nick Lewyckyec38da42009-09-27 00:45:11 +00001532 instructions.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001533
Misha Brukman9d0919f2003-11-08 01:05:38 +00001534</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001535
Chris Lattner00950542001-06-06 20:29:01 +00001536<!-- ======================================================================= -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001537<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner8f8c7b72008-01-04 04:34:14 +00001538
Chris Lattner4f69f462008-01-04 04:32:38 +00001539<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001540
Chris Lattner4f69f462008-01-04 04:32:38 +00001541<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001542 system.</p>
Chris Lattner4f69f462008-01-04 04:32:38 +00001543
Chris Lattner8f8c7b72008-01-04 04:34:14 +00001544</div>
1545
Chris Lattner4f69f462008-01-04 04:32:38 +00001546<!-- _______________________________________________________________________ -->
Nick Lewyckyec38da42009-09-27 00:45:11 +00001547<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1548
1549<div class="doc_text">
1550
1551<h5>Overview:</h5>
1552<p>The integer type is a very simple type that simply specifies an arbitrary
1553 bit width for the integer type desired. Any bit width from 1 bit to
1554 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1555
1556<h5>Syntax:</h5>
1557<pre>
1558 iN
1559</pre>
1560
1561<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1562 value.</p>
1563
1564<h5>Examples:</h5>
1565<table class="layout">
1566 <tr class="layout">
1567 <td class="left"><tt>i1</tt></td>
1568 <td class="left">a single-bit integer.</td>
1569 </tr>
1570 <tr class="layout">
1571 <td class="left"><tt>i32</tt></td>
1572 <td class="left">a 32-bit integer.</td>
1573 </tr>
1574 <tr class="layout">
1575 <td class="left"><tt>i1942652</tt></td>
1576 <td class="left">a really big integer of over 1 million bits.</td>
1577 </tr>
1578</table>
1579
Nick Lewyckyec38da42009-09-27 00:45:11 +00001580</div>
1581
1582<!-- _______________________________________________________________________ -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001583<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1584
1585<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001586
1587<table>
1588 <tbody>
1589 <tr><th>Type</th><th>Description</th></tr>
1590 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1591 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1592 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1593 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1594 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1595 </tbody>
1596</table>
1597
Chris Lattner4f69f462008-01-04 04:32:38 +00001598</div>
1599
1600<!-- _______________________________________________________________________ -->
Dale Johannesen21fe99b2010-10-01 00:48:59 +00001601<div class="doc_subsubsection"> <a name="t_x86mmx">X86mmx Type</a> </div>
1602
1603<div class="doc_text">
1604
1605<h5>Overview:</h5>
1606<p>The x86mmx type represents a value held in an MMX register on an x86 machine. The operations allowed on it are quite limited: parameters and return values, load and store, and bitcast. User-specified MMX instructions are represented as intrinsic or asm calls with arguments and/or results of this type. There are no arrays, vectors or constants of this type.</p>
1607
1608<h5>Syntax:</h5>
1609<pre>
Dale Johannesen473a8c82010-10-01 01:07:02 +00001610 x86mmx
Dale Johannesen21fe99b2010-10-01 00:48:59 +00001611</pre>
1612
1613</div>
1614
1615<!-- _______________________________________________________________________ -->
Chris Lattner4f69f462008-01-04 04:32:38 +00001616<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1617
1618<div class="doc_text">
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001619
Chris Lattner4f69f462008-01-04 04:32:38 +00001620<h5>Overview:</h5>
1621<p>The void type does not represent any value and has no size.</p>
1622
1623<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +00001624<pre>
1625 void
1626</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001627
Chris Lattner4f69f462008-01-04 04:32:38 +00001628</div>
1629
1630<!-- _______________________________________________________________________ -->
1631<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1632
1633<div class="doc_text">
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001634
Chris Lattner4f69f462008-01-04 04:32:38 +00001635<h5>Overview:</h5>
1636<p>The label type represents code labels.</p>
1637
1638<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +00001639<pre>
1640 label
1641</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001642
Chris Lattner4f69f462008-01-04 04:32:38 +00001643</div>
1644
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001645<!-- _______________________________________________________________________ -->
1646<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1647
1648<div class="doc_text">
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001649
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001650<h5>Overview:</h5>
Nick Lewyckyc261df92009-09-27 23:27:42 +00001651<p>The metadata type represents embedded metadata. No derived types may be
1652 created from metadata except for <a href="#t_function">function</a>
1653 arguments.
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001654
1655<h5>Syntax:</h5>
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001656<pre>
1657 metadata
1658</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001659
Nick Lewycky7a0370f2009-05-30 05:06:04 +00001660</div>
1661
Chris Lattner4f69f462008-01-04 04:32:38 +00001662
1663<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00001664<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001665
Misha Brukman9d0919f2003-11-08 01:05:38 +00001666<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00001667
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001668<p>The real power in LLVM comes from the derived types in the system. This is
1669 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewyckyec38da42009-09-27 00:45:11 +00001670 useful types. Each of these types contain one or more element types which
1671 may be a primitive type, or another derived type. For example, it is
1672 possible to have a two dimensional array, using an array as the element type
1673 of another array.</p>
Dan Gohmand8791e52009-01-24 15:58:40 +00001674
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001675
1676</div>
1677
1678<!-- _______________________________________________________________________ -->
1679<div class="doc_subsubsection"> <a name="t_aggregate">Aggregate Types</a> </div>
1680
1681<div class="doc_text">
1682
1683<p>Aggregate Types are a subset of derived types that can contain multiple
1684 member types. <a href="#t_array">Arrays</a>,
Chris Lattner61c70e92010-08-28 04:09:24 +00001685 <a href="#t_struct">structs</a>, and <a href="#t_vector">vectors</a> are
1686 aggregate types.</p>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00001687
1688</div>
1689
Reid Spencer2b916312007-05-16 18:44:01 +00001690<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001691<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001692
Misha Brukman9d0919f2003-11-08 01:05:38 +00001693<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00001694
Chris Lattner00950542001-06-06 20:29:01 +00001695<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001696<p>The array type is a very simple derived type that arranges elements
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001697 sequentially in memory. The array type requires a size (number of elements)
1698 and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001699
Chris Lattner7faa8832002-04-14 06:13:44 +00001700<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001701<pre>
1702 [&lt;# elements&gt; x &lt;elementtype&gt;]
1703</pre>
1704
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001705<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1706 be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00001707
Chris Lattner7faa8832002-04-14 06:13:44 +00001708<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001709<table class="layout">
1710 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001711 <td class="left"><tt>[40 x i32]</tt></td>
1712 <td class="left">Array of 40 32-bit integer values.</td>
1713 </tr>
1714 <tr class="layout">
1715 <td class="left"><tt>[41 x i32]</tt></td>
1716 <td class="left">Array of 41 32-bit integer values.</td>
1717 </tr>
1718 <tr class="layout">
1719 <td class="left"><tt>[4 x i8]</tt></td>
1720 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001721 </tr>
Chris Lattner00950542001-06-06 20:29:01 +00001722</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001723<p>Here are some examples of multidimensional arrays:</p>
1724<table class="layout">
1725 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001726 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1727 <td class="left">3x4 array of 32-bit integer values.</td>
1728 </tr>
1729 <tr class="layout">
1730 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1731 <td class="left">12x10 array of single precision floating point values.</td>
1732 </tr>
1733 <tr class="layout">
1734 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1735 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001736 </tr>
1737</table>
Chris Lattnere67a9512005-06-24 17:22:57 +00001738
Dan Gohman7657f6b2009-11-09 19:01:53 +00001739<p>There is no restriction on indexing beyond the end of the array implied by
1740 a static type (though there are restrictions on indexing beyond the bounds
1741 of an allocated object in some cases). This means that single-dimension
1742 'variable sized array' addressing can be implemented in LLVM with a zero
1743 length array type. An implementation of 'pascal style arrays' in LLVM could
1744 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +00001745
Misha Brukman9d0919f2003-11-08 01:05:38 +00001746</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001747
Chris Lattner00950542001-06-06 20:29:01 +00001748<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001749<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001750
Misha Brukman9d0919f2003-11-08 01:05:38 +00001751<div class="doc_text">
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001752
Chris Lattner00950542001-06-06 20:29:01 +00001753<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001754<p>The function type can be thought of as a function signature. It consists of
1755 a return type and a list of formal parameter types. The return type of a
Chris Lattner61c70e92010-08-28 04:09:24 +00001756 function type is a first class type or a void type.</p>
Devang Patelc3fc6df2008-03-10 20:49:15 +00001757
Chris Lattner00950542001-06-06 20:29:01 +00001758<h5>Syntax:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001759<pre>
Nick Lewycky51386942009-09-27 07:55:32 +00001760 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00001761</pre>
1762
John Criswell0ec250c2005-10-24 16:17:18 +00001763<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001764 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1765 which indicates that the function takes a variable number of arguments.
1766 Variable argument functions can access their arguments with
1767 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner0724fbd2010-03-02 06:36:51 +00001768 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewyckyc261df92009-09-27 23:27:42 +00001769 <a href="#t_label">label</a>.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +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">
Reid Spencer92f82302006-12-31 07:18:34 +00001774 <td class="left"><tt>i32 (i32)</tt></td>
1775 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001776 </td>
Reid Spencer92f82302006-12-31 07:18:34 +00001777 </tr><tr class="layout">
Chris Lattner0724fbd2010-03-02 06:36:51 +00001778 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencerf17a0b72006-12-31 07:20:23 +00001779 </tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001780 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner0724fbd2010-03-02 06:36:51 +00001781 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1782 returning <tt>float</tt>.
Reid Spencer92f82302006-12-31 07:18:34 +00001783 </td>
1784 </tr><tr class="layout">
1785 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00001786 <td class="left">A vararg function that takes at least one
1787 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1788 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer92f82302006-12-31 07:18:34 +00001789 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +00001790 </td>
Devang Patela582f402008-03-24 05:35:41 +00001791 </tr><tr class="layout">
1792 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky51386942009-09-27 07:55:32 +00001793 <td class="left">A function taking an <tt>i32</tt>, returning a
1794 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patela582f402008-03-24 05:35:41 +00001795 </td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001796 </tr>
Chris Lattner00950542001-06-06 20:29:01 +00001797</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00001798
Misha Brukman9d0919f2003-11-08 01:05:38 +00001799</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001800
Chris Lattner00950542001-06-06 20:29:01 +00001801<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001802<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001803
Misha Brukman9d0919f2003-11-08 01:05:38 +00001804<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001805
Chris Lattner00950542001-06-06 20:29:01 +00001806<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001807<p>The structure type is used to represent a collection of data members together
1808 in memory. The packing of the field types is defined to match the ABI of the
1809 underlying processor. The elements of a structure may be any type that has a
1810 size.</p>
1811
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +00001812<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1813 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1814 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1815 Structures in registers are accessed using the
1816 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1817 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattner00950542001-06-06 20:29:01 +00001818<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001819<pre>
1820 { &lt;type list&gt; }
1821</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001822
Chris Lattner00950542001-06-06 20:29:01 +00001823<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001824<table class="layout">
1825 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +00001826 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1827 <td class="left">A triple of three <tt>i32</tt> values</td>
1828 </tr><tr class="layout">
1829 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1830 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1831 second element is a <a href="#t_pointer">pointer</a> to a
1832 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1833 an <tt>i32</tt>.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001834 </tr>
Chris Lattner00950542001-06-06 20:29:01 +00001835</table>
Dan Gohmand8791e52009-01-24 15:58:40 +00001836
Misha Brukman9d0919f2003-11-08 01:05:38 +00001837</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001838
Chris Lattner00950542001-06-06 20:29:01 +00001839<!-- _______________________________________________________________________ -->
Andrew Lenharth75e10682006-12-08 17:13:00 +00001840<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1841</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001842
Andrew Lenharth75e10682006-12-08 17:13:00 +00001843<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001844
Andrew Lenharth75e10682006-12-08 17:13:00 +00001845<h5>Overview:</h5>
1846<p>The packed structure type is used to represent a collection of data members
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001847 together in memory. There is no padding between fields. Further, the
1848 alignment of a packed structure is 1 byte. The elements of a packed
1849 structure may be any type that has a size.</p>
1850
1851<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1852 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1853 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1854
Andrew Lenharth75e10682006-12-08 17:13:00 +00001855<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001856<pre>
1857 &lt; { &lt;type list&gt; } &gt;
1858</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001859
Andrew Lenharth75e10682006-12-08 17:13:00 +00001860<h5>Examples:</h5>
1861<table class="layout">
1862 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +00001863 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1864 <td class="left">A triple of three <tt>i32</tt> values</td>
1865 </tr><tr class="layout">
Bill Wendlinge36dccc2008-09-07 10:26:33 +00001866 <td class="left">
1867<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00001868 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1869 second element is a <a href="#t_pointer">pointer</a> to a
1870 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1871 an <tt>i32</tt>.</td>
Andrew Lenharth75e10682006-12-08 17:13:00 +00001872 </tr>
1873</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001874
Andrew Lenharth75e10682006-12-08 17:13:00 +00001875</div>
1876
1877<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00001878<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Chris Lattner0fd4a272009-02-08 19:53:29 +00001879
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001880<div class="doc_text">
1881
1882<h5>Overview:</h5>
Dan Gohmanff3ef322010-02-25 16:50:07 +00001883<p>The pointer type is used to specify memory locations.
1884 Pointers are commonly used to reference objects in memory.</p>
1885
1886<p>Pointer types may have an optional address space attribute defining the
1887 numbered address space where the pointed-to object resides. The default
1888 address space is number zero. The semantics of non-zero address
1889 spaces are target-specific.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001890
1891<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1892 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner0fd4a272009-02-08 19:53:29 +00001893
Chris Lattner7faa8832002-04-14 06:13:44 +00001894<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00001895<pre>
1896 &lt;type&gt; *
1897</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001898
Chris Lattner7faa8832002-04-14 06:13:44 +00001899<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001900<table class="layout">
1901 <tr class="layout">
Dan Gohman2a08c532009-01-04 23:44:43 +00001902 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +00001903 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1904 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1905 </tr>
1906 <tr class="layout">
Dan Gohmanfe47aae2010-05-28 17:13:49 +00001907 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +00001908 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +00001909 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner23ff1f92007-12-19 05:04:11 +00001910 <tt>i32</tt>.</td>
1911 </tr>
1912 <tr class="layout">
1913 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1914 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1915 that resides in address space #5.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001916 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +00001917</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001918
Misha Brukman9d0919f2003-11-08 01:05:38 +00001919</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001920
Chris Lattnera58561b2004-08-12 19:12:28 +00001921<!-- _______________________________________________________________________ -->
Reid Spencer485bad12007-02-15 03:07:05 +00001922<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001923
Misha Brukman9d0919f2003-11-08 01:05:38 +00001924<div class="doc_text">
Chris Lattner69c11bb2005-04-25 17:34:15 +00001925
Chris Lattnera58561b2004-08-12 19:12:28 +00001926<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001927<p>A vector type is a simple derived type that represents a vector of elements.
1928 Vector types are used when multiple primitive data are operated in parallel
1929 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sandsd40d14e2009-11-27 13:38:03 +00001930 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001931 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001932
Chris Lattnera58561b2004-08-12 19:12:28 +00001933<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001934<pre>
1935 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1936</pre>
1937
Chris Lattner7d2e7be2010-10-10 18:20:35 +00001938<p>The number of elements is a constant integer value larger than 0; elementtype
1939 may be any integer or floating point type. Vectors of size zero are not
1940 allowed, and pointers are not allowed as the element type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001941
Chris Lattnera58561b2004-08-12 19:12:28 +00001942<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001943<table class="layout">
1944 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001945 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1946 <td class="left">Vector of 4 32-bit integer values.</td>
1947 </tr>
1948 <tr class="layout">
1949 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1950 <td class="left">Vector of 8 32-bit floating-point values.</td>
1951 </tr>
1952 <tr class="layout">
1953 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1954 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +00001955 </tr>
1956</table>
Dan Gohmand8791e52009-01-24 15:58:40 +00001957
Misha Brukman9d0919f2003-11-08 01:05:38 +00001958</div>
1959
Chris Lattner69c11bb2005-04-25 17:34:15 +00001960<!-- _______________________________________________________________________ -->
1961<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1962<div class="doc_text">
1963
1964<h5>Overview:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001965<p>Opaque types are used to represent unknown types in the system. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001966 corresponds (for example) to the C notion of a forward declared structure
1967 type. In LLVM, opaque types can eventually be resolved to any type (not just
1968 a structure type).</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001969
1970<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001971<pre>
1972 opaque
1973</pre>
1974
1975<h5>Examples:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001976<table class="layout">
1977 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +00001978 <td class="left"><tt>opaque</tt></td>
1979 <td class="left">An opaque type.</td>
Chris Lattner69c11bb2005-04-25 17:34:15 +00001980 </tr>
1981</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001982
Chris Lattner69c11bb2005-04-25 17:34:15 +00001983</div>
1984
Chris Lattner242d61d2009-02-02 07:32:36 +00001985<!-- ======================================================================= -->
1986<div class="doc_subsection">
1987 <a name="t_uprefs">Type Up-references</a>
1988</div>
1989
1990<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001991
Chris Lattner242d61d2009-02-02 07:32:36 +00001992<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00001993<p>An "up reference" allows you to refer to a lexically enclosing type without
1994 requiring it to have a name. For instance, a structure declaration may
1995 contain a pointer to any of the types it is lexically a member of. Example
1996 of up references (with their equivalent as named type declarations)
1997 include:</p>
Chris Lattner242d61d2009-02-02 07:32:36 +00001998
1999<pre>
Chris Lattner3060f5b2009-02-09 10:00:56 +00002000 { \2 * } %x = type { %x* }
Chris Lattner242d61d2009-02-02 07:32:36 +00002001 { \2 }* %y = type { %y }*
2002 \1* %z = type %z*
2003</pre>
2004
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002005<p>An up reference is needed by the asmprinter for printing out cyclic types
2006 when there is no declared name for a type in the cycle. Because the
2007 asmprinter does not want to print out an infinite type string, it needs a
2008 syntax to handle recursive types that have no names (all names are optional
2009 in llvm IR).</p>
Chris Lattner242d61d2009-02-02 07:32:36 +00002010
2011<h5>Syntax:</h5>
2012<pre>
2013 \&lt;level&gt;
2014</pre>
2015
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002016<p>The level is the count of the lexical type that is being referred to.</p>
Chris Lattner242d61d2009-02-02 07:32:36 +00002017
2018<h5>Examples:</h5>
Chris Lattner242d61d2009-02-02 07:32:36 +00002019<table class="layout">
2020 <tr class="layout">
2021 <td class="left"><tt>\1*</tt></td>
2022 <td class="left">Self-referential pointer.</td>
2023 </tr>
2024 <tr class="layout">
2025 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
2026 <td class="left">Recursive structure where the upref refers to the out-most
2027 structure.</td>
2028 </tr>
2029</table>
Chris Lattner242d61d2009-02-02 07:32:36 +00002030
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002031</div>
Chris Lattner69c11bb2005-04-25 17:34:15 +00002032
Chris Lattnerc3f59762004-12-09 17:30:23 +00002033<!-- *********************************************************************** -->
2034<div class="doc_section"> <a name="constants">Constants</a> </div>
2035<!-- *********************************************************************** -->
2036
2037<div class="doc_text">
2038
2039<p>LLVM has several different basic types of constants. This section describes
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002040 them all and their syntax.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002041
2042</div>
2043
2044<!-- ======================================================================= -->
Reid Spencercc16dc32004-12-09 18:02:53 +00002045<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002046
2047<div class="doc_text">
2048
2049<dl>
2050 <dt><b>Boolean constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002051 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewyckyec38da42009-09-27 00:45:11 +00002052 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002053
2054 <dt><b>Integer constants</b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002055 <dd>Standard integers (such as '4') are constants of
2056 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2057 with integer types.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002058
2059 <dt><b>Floating point constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002060 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002061 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2062 notation (see below). The assembler requires the exact decimal value of a
2063 floating-point constant. For example, the assembler accepts 1.25 but
2064 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2065 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002066
2067 <dt><b>Null pointer constants</b></dt>
John Criswell9e2485c2004-12-10 15:51:16 +00002068 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002069 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002070</dl>
2071
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002072<p>The one non-intuitive notation for constants is the hexadecimal form of
2073 floating point constants. For example, the form '<tt>double
2074 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2075 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2076 constants are required (and the only time that they are generated by the
2077 disassembler) is when a floating point constant must be emitted but it cannot
2078 be represented as a decimal floating point number in a reasonable number of
2079 digits. For example, NaN's, infinities, and other special values are
2080 represented in their IEEE hexadecimal format so that assembly and disassembly
2081 do not cause any bits to change in the constants.</p>
2082
Dale Johannesenbd5e5a82009-02-11 22:14:51 +00002083<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002084 represented using the 16-digit form shown above (which matches the IEEE754
2085 representation for double); float values must, however, be exactly
2086 representable as IEE754 single precision. Hexadecimal format is always used
2087 for long double, and there are three forms of long double. The 80-bit format
2088 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2089 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2090 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2091 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2092 currently supported target uses this format. Long doubles will only work if
2093 they match the long double format on your target. All hexadecimal formats
2094 are big-endian (sign bit at the left).</p>
2095
Dale Johannesen21fe99b2010-10-01 00:48:59 +00002096<p>There are no constants of type x86mmx.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002097</div>
2098
2099<!-- ======================================================================= -->
Chris Lattner70882792009-02-28 18:32:25 +00002100<div class="doc_subsection">
Bill Wendlingd9fe2982009-07-20 02:32:41 +00002101<a name="aggregateconstants"></a> <!-- old anchor -->
2102<a name="complexconstants">Complex Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002103</div>
2104
2105<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002106
Chris Lattner70882792009-02-28 18:32:25 +00002107<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002108 constants and smaller complex constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002109
2110<dl>
2111 <dt><b>Structure constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002112 <dd>Structure constants are represented with notation similar to structure
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002113 type definitions (a comma separated list of elements, surrounded by braces
2114 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2115 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2116 Structure constants must have <a href="#t_struct">structure type</a>, and
2117 the number and types of elements must match those specified by the
2118 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002119
2120 <dt><b>Array constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002121 <dd>Array constants are represented with notation similar to array type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002122 definitions (a comma separated list of elements, surrounded by square
2123 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2124 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2125 the number and types of elements must match those specified by the
2126 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002127
Reid Spencer485bad12007-02-15 03:07:05 +00002128 <dt><b>Vector constants</b></dt>
Reid Spencer485bad12007-02-15 03:07:05 +00002129 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002130 definitions (a comma separated list of elements, surrounded by
2131 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2132 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2133 have <a href="#t_vector">vector type</a>, and the number and types of
2134 elements must match those specified by the type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002135
2136 <dt><b>Zero initialization</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002137 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattnerfdfeb692010-02-12 20:49:41 +00002138 value to zero of <em>any</em> type, including scalar and
2139 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002140 This is often used to avoid having to print large zero initializers
2141 (e.g. for large arrays) and is always exactly equivalent to using explicit
2142 zero initializers.</dd>
Nick Lewycky21cc4462009-04-04 07:22:01 +00002143
2144 <dt><b>Metadata node</b></dt>
Nick Lewycky1e8c7a62009-05-30 16:08:30 +00002145 <dd>A metadata node is a structure-like constant with
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002146 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2147 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2148 be interpreted as part of the instruction stream, metadata is a place to
2149 attach additional information such as debug info.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002150</dl>
2151
2152</div>
2153
2154<!-- ======================================================================= -->
2155<div class="doc_subsection">
2156 <a name="globalconstants">Global Variable and Function Addresses</a>
2157</div>
2158
2159<div class="doc_text">
2160
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002161<p>The addresses of <a href="#globalvars">global variables</a>
2162 and <a href="#functionstructure">functions</a> are always implicitly valid
2163 (link-time) constants. These constants are explicitly referenced when
2164 the <a href="#identifiers">identifier for the global</a> is used and always
2165 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2166 legal LLVM file:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002167
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002168<pre class="doc_code">
Chris Lattnera18a4242007-06-06 18:28:13 +00002169@X = global i32 17
2170@Y = global i32 42
2171@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +00002172</pre>
2173
2174</div>
2175
2176<!-- ======================================================================= -->
Reid Spencer2dc45b82004-12-09 18:13:12 +00002177<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002178<div class="doc_text">
2179
Chris Lattner48a109c2009-09-07 22:52:39 +00002180<p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
Benjamin Kramer8040cd32009-10-12 14:46:08 +00002181 indicates that the user of the value may receive an unspecified bit-pattern.
Bill Wendling1b383ba2010-10-27 01:07:41 +00002182 Undefined values may be of any type (other than '<tt>label</tt>'
2183 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002184
Chris Lattnerc608cb12009-09-11 01:49:31 +00002185<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner48a109c2009-09-07 22:52:39 +00002186 program is well defined no matter what value is used. This gives the
2187 compiler more freedom to optimize. Here are some examples of (potentially
2188 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002189
Chris Lattner48a109c2009-09-07 22:52:39 +00002190
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002191<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002192 %A = add %X, undef
2193 %B = sub %X, undef
2194 %C = xor %X, undef
2195Safe:
2196 %A = undef
2197 %B = undef
2198 %C = undef
2199</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002200
2201<p>This is safe because all of the output bits are affected by the undef bits.
Bill Wendling1b383ba2010-10-27 01:07:41 +00002202 Any output bit can have a zero or one depending on the input bits.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +00002203
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002204<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002205 %A = or %X, undef
2206 %B = and %X, undef
2207Safe:
2208 %A = -1
2209 %B = 0
2210Unsafe:
2211 %A = undef
2212 %B = undef
2213</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002214
2215<p>These logical operations have bits that are not always affected by the input.
Bill Wendling1b383ba2010-10-27 01:07:41 +00002216 For example, if <tt>%X</tt> has a zero bit, then the output of the
2217 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2218 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2219 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2220 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2221 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2222 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2223 set, allowing the '<tt>or</tt>' to be folded to -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
Bill Wendling1b383ba2010-10-27 01:07:41 +00002239<p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2240 branch) conditions can go <em>either way</em>, but they have to come from one
2241 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2242 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2243 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2244 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2245 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2246 eliminated.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +00002247
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002248<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +00002249 %A = xor undef, undef
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002250
Chris Lattner48a109c2009-09-07 22:52:39 +00002251 %B = undef
2252 %C = xor %B, %B
2253
2254 %D = undef
2255 %E = icmp lt %D, 4
2256 %F = icmp gte %D, 4
2257
2258Safe:
2259 %A = undef
2260 %B = undef
2261 %C = undef
2262 %D = undef
2263 %E = undef
2264 %F = undef
2265</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +00002266
Bill Wendling1b383ba2010-10-27 01:07:41 +00002267<p>This example points out that two '<tt>undef</tt>' operands are not
2268 necessarily the same. This can be surprising to people (and also matches C
2269 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2270 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2271 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2272 its value over its "live range". This is true because the variable doesn't
2273 actually <em>have a live range</em>. Instead, the value is logically read
2274 from arbitrary registers that happen to be around when needed, so the value
2275 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2276 need to have the same semantics or the core LLVM "replace all uses with"
2277 concept would not hold.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002278
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002279<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +00002280 %A = fdiv undef, %X
2281 %B = fdiv %X, undef
2282Safe:
2283 %A = undef
2284b: unreachable
2285</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002286
2287<p>These examples show the crucial difference between an <em>undefined
Bill Wendling1b383ba2010-10-27 01:07:41 +00002288 value</em> and <em>undefined behavior</em>. An undefined value (like
2289 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2290 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2291 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2292 defined on SNaN's. However, in the second example, we can make a more
2293 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2294 arbitrary value, we are allowed to assume that it could be zero. Since a
2295 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2296 the operation does not execute at all. This allows us to delete the divide and
2297 all code after it. Because the undefined operation "can't happen", the
2298 optimizer can assume that it occurs in dead code.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002299
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002300<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +00002301a: store undef -> %X
2302b: store %X -> undef
2303Safe:
2304a: &lt;deleted&gt;
2305b: unreachable
2306</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002307
Bill Wendling1b383ba2010-10-27 01:07:41 +00002308<p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2309 undefined value can be assumed to not have any effect; we can assume that the
2310 value is overwritten with bits that happen to match what was already there.
2311 However, a store <em>to</em> an undefined location could clobber arbitrary
2312 memory, therefore, it has undefined behavior.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +00002313
Chris Lattnerc3f59762004-12-09 17:30:23 +00002314</div>
2315
2316<!-- ======================================================================= -->
Dan Gohmanfff6c532010-04-22 23:14:21 +00002317<div class="doc_subsection"><a name="trapvalues">Trap Values</a></div>
2318<div class="doc_text">
2319
Dan Gohmanc68ce062010-04-26 20:21:21 +00002320<p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmanfff6c532010-04-22 23:14:21 +00002321 instead of representing an unspecified bit pattern, they represent the
2322 fact that an instruction or constant expression which cannot evoke side
2323 effects has nevertheless detected a condition which results in undefined
Dan Gohmanc68ce062010-04-26 20:21:21 +00002324 behavior.</p>
Dan Gohmanfff6c532010-04-22 23:14:21 +00002325
Dan Gohman34b3d992010-04-28 00:49:41 +00002326<p>There is currently no way of representing a trap value in the IR; they
Dan Gohman855abed2010-05-03 14:51:43 +00002327 only exist when produced by operations such as
Dan Gohman34b3d992010-04-28 00:49:41 +00002328 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002329
Dan Gohman34b3d992010-04-28 00:49:41 +00002330<p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002331
Dan Gohman34b3d992010-04-28 00:49:41 +00002332<ul>
2333<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2334 their operands.</li>
2335
2336<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2337 to their dynamic predecessor basic block.</li>
2338
2339<li>Function arguments depend on the corresponding actual argument values in
2340 the dynamic callers of their functions.</li>
2341
2342<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2343 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2344 control back to them.</li>
2345
Dan Gohmanb5328162010-05-03 14:55:22 +00002346<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2347 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2348 or exception-throwing call instructions that dynamically transfer control
2349 back to them.</li>
2350
Dan Gohman34b3d992010-04-28 00:49:41 +00002351<li>Non-volatile loads and stores depend on the most recent stores to all of the
2352 referenced memory addresses, following the order in the IR
2353 (including loads and stores implied by intrinsics such as
2354 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2355
Dan Gohman7c24ff12010-05-03 14:59:34 +00002356<!-- TODO: In the case of multiple threads, this only applies if the store
2357 "happens-before" the load or store. -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002358
Dan Gohman34b3d992010-04-28 00:49:41 +00002359<!-- TODO: floating-point exception state -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002360
Dan Gohman34b3d992010-04-28 00:49:41 +00002361<li>An instruction with externally visible side effects depends on the most
2362 recent preceding instruction with externally visible side effects, following
Dan Gohmanff70fe42010-07-06 15:26:33 +00002363 the order in the IR. (This includes
2364 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanfff6c532010-04-22 23:14:21 +00002365
Dan Gohmanb5328162010-05-03 14:55:22 +00002366<li>An instruction <i>control-depends</i> on a
2367 <a href="#terminators">terminator instruction</a>
2368 if the terminator instruction has multiple successors and the instruction
2369 is always executed when control transfers to one of the successors, and
2370 may not be executed when control is transfered to another.</li>
Dan Gohman34b3d992010-04-28 00:49:41 +00002371
2372<li>Dependence is transitive.</li>
2373
2374</ul>
Dan Gohman34b3d992010-04-28 00:49:41 +00002375
2376<p>Whenever a trap value is generated, all values which depend on it evaluate
2377 to trap. If they have side effects, the evoke their side effects as if each
2378 operand with a trap value were undef. If they have externally-visible side
2379 effects, the behavior is undefined.</p>
2380
2381<p>Here are some examples:</p>
Dan Gohmanc30f6e12010-04-26 20:54:53 +00002382
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002383<pre class="doc_code">
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002384entry:
2385 %trap = sub nuw i32 0, 1 ; Results in a trap value.
Dan Gohman34b3d992010-04-28 00:49:41 +00002386 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2387 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2388 store i32 0, i32* %trap_yet_again ; undefined behavior
2389
2390 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2391 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2392
2393 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2394
2395 %narrowaddr = bitcast i32* @g to i16*
2396 %wideaddr = bitcast i32* @g to i64*
2397 %trap3 = load 16* %narrowaddr ; Returns a trap value.
2398 %trap4 = load i64* %widaddr ; Returns a trap value.
2399
2400 %cmp = icmp i32 slt %trap, 0 ; Returns a trap value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002401 %br i1 %cmp, %true, %end ; Branch to either destination.
2402
2403true:
Dan Gohman34b3d992010-04-28 00:49:41 +00002404 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2405 ; it has undefined behavior.
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002406 br label %end
2407
2408end:
2409 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2410 ; Both edges into this PHI are
2411 ; control-dependent on %cmp, so this
Dan Gohman34b3d992010-04-28 00:49:41 +00002412 ; always results in a trap value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002413
2414 volatile store i32 0, i32* @g ; %end is control-equivalent to %entry
2415 ; so this is defined (ignoring earlier
2416 ; undefined behavior in this example).
Dan Gohmanae11c3f2010-04-26 23:36:52 +00002417</pre>
Dan Gohmanfff6c532010-04-22 23:14:21 +00002418
Dan Gohmanfff6c532010-04-22 23:14:21 +00002419</div>
2420
2421<!-- ======================================================================= -->
Chris Lattnerf9d078e2009-10-27 21:19:13 +00002422<div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2423 Blocks</a></div>
Chris Lattnerc6f44362009-10-27 21:01:34 +00002424<div class="doc_text">
2425
Chris Lattnercdfc9402009-11-01 01:27:45 +00002426<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerc6f44362009-10-27 21:01:34 +00002427
2428<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner2dfdf2a2009-10-27 21:49:40 +00002429 basic block in the specified function, and always has an i8* type. Taking
Chris Lattnercdfc9402009-11-01 01:27:45 +00002430 the address of the entry block is illegal.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002431
Chris Lattnerc6f44362009-10-27 21:01:34 +00002432<p>This value only has defined behavior when used as an operand to the
Bill Wendling1b383ba2010-10-27 01:07:41 +00002433 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2434 comparisons against null. Pointer equality tests between labels addresses
2435 results in undefined behavior &mdash; though, again, comparison against null
2436 is ok, and no label is equal to the null pointer. This may be passed around
2437 as an opaque pointer sized value as long as the bits are not inspected. This
2438 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2439 long as the original value is reconstituted before the <tt>indirectbr</tt>
2440 instruction.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00002441
Bill Wendling1b383ba2010-10-27 01:07:41 +00002442<p>Finally, some targets may provide defined semantics when using the value as
2443 the operand to an inline assembly, but that is target specific.</p>
Chris Lattnerc6f44362009-10-27 21:01:34 +00002444
2445</div>
2446
2447
2448<!-- ======================================================================= -->
Chris Lattnerc3f59762004-12-09 17:30:23 +00002449<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2450</div>
2451
2452<div class="doc_text">
2453
2454<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002455 to be used as constants. Constant expressions may be of
2456 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2457 operation that does not have side effects (e.g. load and call are not
Bill Wendling1b383ba2010-10-27 01:07:41 +00002458 supported). The following is the syntax for constant expressions:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002459
2460<dl>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002461 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002462 <dd>Truncate a constant to another type. The bit size of CST must be larger
2463 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002464
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002465 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002466 <dd>Zero extend a constant to another type. The bit size of CST must be
Duncan Sands28afd432010-07-13 12:06:14 +00002467 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002468
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002469 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002470 <dd>Sign extend a constant to another type. The bit size of CST must be
Duncan Sands28afd432010-07-13 12:06:14 +00002471 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002472
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002473 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002474 <dd>Truncate a floating point constant to another floating point type. The
2475 size of CST must be larger than the size of TYPE. Both types must be
2476 floating point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002477
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002478 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002479 <dd>Floating point extend a constant to another type. The size of CST must be
2480 smaller or equal to the size of TYPE. Both types must be floating
2481 point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002482
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002483 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002484 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002485 constant. TYPE must be a scalar or vector integer type. CST must be of
2486 scalar or vector floating point type. Both CST and TYPE must be scalars,
2487 or vectors of the same number of elements. If the value won't fit in the
2488 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002489
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002490 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002491 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002492 constant. TYPE must be a scalar or vector integer type. CST must be of
2493 scalar or vector floating point type. Both CST and TYPE must be scalars,
2494 or vectors of the same number of elements. If the value won't fit in the
2495 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002496
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002497 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002498 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002499 constant. TYPE must be a scalar or vector floating point type. CST must be
2500 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2501 vectors of the same number of elements. If the value won't fit in the
2502 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002503
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002504 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002505 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002506 constant. TYPE must be a scalar or vector floating point type. CST must be
2507 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2508 vectors of the same number of elements. If the value won't fit in the
2509 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00002510
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002511 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5c0ef472006-11-11 23:08:07 +00002512 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002513 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2514 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2515 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +00002516
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002517 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002518 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2519 type. CST must be of integer type. The CST value is zero extended,
2520 truncated, or unchanged to make it fit in a pointer size. This one is
2521 <i>really</i> dangerous!</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +00002522
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002523 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner03bbad62009-02-28 18:27:03 +00002524 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2525 are the same as those for the <a href="#i_bitcast">bitcast
2526 instruction</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002527
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002528 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2529 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002530 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002531 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2532 instruction, the index list may have zero or more indexes, which are
2533 required to make sense for the type of "CSTPTR".</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002534
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002535 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002536 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer01c42592006-12-04 19:23:19 +00002537
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002538 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +00002539 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2540
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002541 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +00002542 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +00002543
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002544 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002545 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2546 constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +00002547
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002548 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002549 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2550 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +00002551
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002552 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002553 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2554 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +00002555
Nick Lewycky9e130ce2010-05-29 06:44:15 +00002556 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2557 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2558 constants. The index list is interpreted in a similar manner as indices in
2559 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2560 index value must be specified.</dd>
2561
2562 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2563 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2564 constants. The index list is interpreted in a similar manner as indices in
2565 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2566 index value must be specified.</dd>
2567
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00002568 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002569 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2570 be any of the <a href="#binaryops">binary</a>
2571 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2572 on operands are the same as those for the corresponding instruction
2573 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002574</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002575
Chris Lattnerc3f59762004-12-09 17:30:23 +00002576</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +00002577
Chris Lattner00950542001-06-06 20:29:01 +00002578<!-- *********************************************************************** -->
Chris Lattnere87d6532006-01-25 23:47:57 +00002579<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2580<!-- *********************************************************************** -->
2581
2582<!-- ======================================================================= -->
2583<div class="doc_subsection">
2584<a name="inlineasm">Inline Assembler Expressions</a>
2585</div>
2586
2587<div class="doc_text">
2588
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002589<p>LLVM supports inline assembler expressions (as opposed
2590 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2591 a special value. This value represents the inline assembler as a string
2592 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen09fed252009-10-13 21:56:55 +00002593 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +00002594 expression has side effects, and a flag indicating whether the function
2595 containing the asm needs to align its stack conservatively. An example
2596 inline assembler expression is:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002597
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002598<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00002599i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +00002600</pre>
2601
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002602<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2603 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2604 have:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002605
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002606<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00002607%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattnere87d6532006-01-25 23:47:57 +00002608</pre>
2609
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002610<p>Inline asms with side effects not visible in the constraint list must be
2611 marked as having side effects. This is done through the use of the
2612 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002613
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002614<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00002615call void asm sideeffect "eieio", ""()
Chris Lattnere87d6532006-01-25 23:47:57 +00002616</pre>
2617
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +00002618<p>In some cases inline asms will contain code that will not work unless the
2619 stack is aligned in some way, such as calls or SSE instructions on x86,
2620 yet will not contain code that does that alignment within the asm.
2621 The compiler should make conservative assumptions about what the asm might
2622 contain and should generate its usual stack alignment code in the prologue
2623 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen09fed252009-10-13 21:56:55 +00002624
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002625<pre class="doc_code">
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +00002626call void asm alignstack "eieio", ""()
Dale Johannesen09fed252009-10-13 21:56:55 +00002627</pre>
Dale Johannesen09fed252009-10-13 21:56:55 +00002628
2629<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2630 first.</p>
2631
Chris Lattnere87d6532006-01-25 23:47:57 +00002632<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002633 documented here. Constraints on what can be done (e.g. duplication, moving,
2634 etc need to be documented). This is probably best done by reference to
2635 another document that covers inline asm from a holistic perspective.</p>
Chris Lattnercf9a4152010-04-07 05:38:05 +00002636</div>
2637
2638<div class="doc_subsubsection">
2639<a name="inlineasm_md">Inline Asm Metadata</a>
2640</div>
2641
2642<div class="doc_text">
2643
2644<p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
Chris Lattnerce1b9ad2010-11-17 08:20:42 +00002645 attached to it that contains a list of constant integers. If present, the
2646 code generator will use the integer as the location cookie value when report
Chris Lattnercf9a4152010-04-07 05:38:05 +00002647 errors through the LLVMContext error reporting mechanisms. This allows a
Dan Gohman1c70c002010-04-28 00:36:01 +00002648 front-end to correlate backend errors that occur with inline asm back to the
Chris Lattnercf9a4152010-04-07 05:38:05 +00002649 source code that produced it. For example:</p>
2650
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002651<pre class="doc_code">
Chris Lattnercf9a4152010-04-07 05:38:05 +00002652call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2653...
2654!42 = !{ i32 1234567 }
2655</pre>
Chris Lattnercf9a4152010-04-07 05:38:05 +00002656
2657<p>It is up to the front-end to make sense of the magic numbers it places in the
Chris Lattnerce1b9ad2010-11-17 08:20:42 +00002658 IR. If the MDNode contains multiple constants, the code generator will use
2659 the one that corresponds to the line of the asm that the error occurs on.</p>
Chris Lattnere87d6532006-01-25 23:47:57 +00002660
2661</div>
2662
Chris Lattnere6a5ddd2010-01-15 21:50:19 +00002663<!-- ======================================================================= -->
2664<div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2665 Strings</a>
2666</div>
2667
2668<div class="doc_text">
2669
2670<p>LLVM IR allows metadata to be attached to instructions in the program that
2671 can convey extra information about the code to the optimizers and code
2672 generator. One example application of metadata is source-level debug
2673 information. There are two metadata primitives: strings and nodes. All
2674 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2675 preceding exclamation point ('<tt>!</tt>').</p>
2676
2677<p>A metadata string is a string surrounded by double quotes. It can contain
2678 any character by escaping non-printable characters with "\xx" where "xx" is
2679 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2680
2681<p>Metadata nodes are represented with notation similar to structure constants
2682 (a comma separated list of elements, surrounded by braces and preceded by an
2683 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2684 10}</tt>". Metadata nodes can have any values as their operand.</p>
2685
2686<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2687 metadata nodes, which can be looked up in the module symbol table. For
2688 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2689
Devang Patele1d50cd2010-03-04 23:44:48 +00002690<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002691 function is using two metadata arguments.</p>
Devang Patele1d50cd2010-03-04 23:44:48 +00002692
Bill Wendling9ff5de92011-03-02 02:17:11 +00002693<div class="doc_code">
2694<pre>
2695call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2696</pre>
2697</div>
Devang Patele1d50cd2010-03-04 23:44:48 +00002698
2699<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00002700 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.</p>
Devang Patele1d50cd2010-03-04 23:44:48 +00002701
Bill Wendling9ff5de92011-03-02 02:17:11 +00002702<div class="doc_code">
2703<pre>
2704%indvar.next = add i64 %indvar, 1, !dbg !21
2705</pre>
2706</div>
2707
Chris Lattnere6a5ddd2010-01-15 21:50:19 +00002708</div>
2709
Chris Lattner857755c2009-07-20 05:55:19 +00002710
2711<!-- *********************************************************************** -->
2712<div class="doc_section">
2713 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2714</div>
2715<!-- *********************************************************************** -->
2716
2717<p>LLVM has a number of "magic" global variables that contain data that affect
2718code generation or other IR semantics. These are documented here. All globals
Chris Lattner401e10c2009-07-20 06:14:25 +00002719of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2720section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2721by LLVM.</p>
Chris Lattner857755c2009-07-20 05:55:19 +00002722
2723<!-- ======================================================================= -->
2724<div class="doc_subsection">
2725<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2726</div>
2727
2728<div class="doc_text">
2729
2730<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2731href="#linkage_appending">appending linkage</a>. This array contains a list of
2732pointers to global variables and functions which may optionally have a pointer
2733cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2734
2735<pre>
2736 @X = global i8 4
2737 @Y = global i32 123
2738
2739 @llvm.used = appending global [2 x i8*] [
2740 i8* @X,
2741 i8* bitcast (i32* @Y to i8*)
2742 ], section "llvm.metadata"
2743</pre>
2744
2745<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2746compiler, assembler, and linker are required to treat the symbol as if there is
2747a reference to the global that it cannot see. For example, if a variable has
2748internal linkage and no references other than that from the <tt>@llvm.used</tt>
2749list, it cannot be deleted. This is commonly used to represent references from
2750inline asms and other things the compiler cannot "see", and corresponds to
2751"attribute((used))" in GNU C.</p>
2752
2753<p>On some targets, the code generator must emit a directive to the assembler or
2754object file to prevent the assembler and linker from molesting the symbol.</p>
2755
2756</div>
2757
2758<!-- ======================================================================= -->
2759<div class="doc_subsection">
Chris Lattner401e10c2009-07-20 06:14:25 +00002760<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2761</div>
2762
2763<div class="doc_text">
2764
2765<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2766<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2767touching the symbol. On targets that support it, this allows an intelligent
2768linker to optimize references to the symbol without being impeded as it would be
2769by <tt>@llvm.used</tt>.</p>
2770
2771<p>This is a rare construct that should only be used in rare circumstances, and
2772should not be exposed to source languages.</p>
2773
2774</div>
2775
2776<!-- ======================================================================= -->
2777<div class="doc_subsection">
Chris Lattner857755c2009-07-20 05:55:19 +00002778<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2779</div>
2780
2781<div class="doc_text">
David Chisnalle31e9962010-04-30 19:23:49 +00002782<pre>
2783%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +00002784@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalle31e9962010-04-30 19:23:49 +00002785</pre>
2786<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.
2787</p>
Chris Lattner857755c2009-07-20 05:55:19 +00002788
2789</div>
2790
2791<!-- ======================================================================= -->
2792<div class="doc_subsection">
2793<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2794</div>
2795
2796<div class="doc_text">
David Chisnalle31e9962010-04-30 19:23:49 +00002797<pre>
2798%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +00002799@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalle31e9962010-04-30 19:23:49 +00002800</pre>
Chris Lattner857755c2009-07-20 05:55:19 +00002801
David Chisnalle31e9962010-04-30 19:23:49 +00002802<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.
2803</p>
Chris Lattner857755c2009-07-20 05:55:19 +00002804
2805</div>
2806
2807
Chris Lattnere87d6532006-01-25 23:47:57 +00002808<!-- *********************************************************************** -->
Chris Lattner261efe92003-11-25 01:02:51 +00002809<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2810<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +00002811
Misha Brukman9d0919f2003-11-08 01:05:38 +00002812<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00002813
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002814<p>The LLVM instruction set consists of several different classifications of
2815 instructions: <a href="#terminators">terminator
2816 instructions</a>, <a href="#binaryops">binary instructions</a>,
2817 <a href="#bitwiseops">bitwise binary instructions</a>,
2818 <a href="#memoryops">memory instructions</a>, and
2819 <a href="#otherops">other instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002820
Misha Brukman9d0919f2003-11-08 01:05:38 +00002821</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002822
Chris Lattner00950542001-06-06 20:29:01 +00002823<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00002824<div class="doc_subsection"> <a name="terminators">Terminator
2825Instructions</a> </div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002826
Misha Brukman9d0919f2003-11-08 01:05:38 +00002827<div class="doc_text">
Chris Lattnerc3f59762004-12-09 17:30:23 +00002828
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002829<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2830 in a program ends with a "Terminator" instruction, which indicates which
2831 block should be executed after the current block is finished. These
2832 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2833 control flow, not values (the one exception being the
2834 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2835
Duncan Sands83821c82010-04-15 20:35:54 +00002836<p>There are seven different terminator instructions: the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002837 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2838 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2839 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
Bill Wendling21c346e2009-11-02 00:25:26 +00002840 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002841 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2842 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2843 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002844
Misha Brukman9d0919f2003-11-08 01:05:38 +00002845</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +00002846
Chris Lattner00950542001-06-06 20:29:01 +00002847<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00002848<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2849Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002850
Misha Brukman9d0919f2003-11-08 01:05:38 +00002851<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002852
Chris Lattner00950542001-06-06 20:29:01 +00002853<h5>Syntax:</h5>
Dan Gohmanb1e6b962008-10-04 19:00:07 +00002854<pre>
2855 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00002856 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +00002857</pre>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002858
Chris Lattner00950542001-06-06 20:29:01 +00002859<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002860<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2861 a value) from a function back to the caller.</p>
2862
2863<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2864 value and then causes control flow, and one that just causes control flow to
2865 occur.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002866
Chris Lattner00950542001-06-06 20:29:01 +00002867<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002868<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2869 return value. The type of the return value must be a
2870 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmanb1e6b962008-10-04 19:00:07 +00002871
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002872<p>A function is not <a href="#wellformed">well formed</a> if it it has a
2873 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2874 value or a return value with a type that does not match its type, or if it
2875 has a void return type and contains a '<tt>ret</tt>' instruction with a
2876 return value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002877
Chris Lattner00950542001-06-06 20:29:01 +00002878<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002879<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2880 the calling function's context. If the caller is a
2881 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2882 instruction after the call. If the caller was an
2883 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2884 the beginning of the "normal" destination block. If the instruction returns
2885 a value, that value shall set the call or invoke instruction's return
2886 value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002887
Chris Lattner00950542001-06-06 20:29:01 +00002888<h5>Example:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +00002889<pre>
2890 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +00002891 ret void <i>; Return from a void function</i>
Bill Wendling0a4bbbf2009-02-28 22:12:54 +00002892 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Chris Lattner00950542001-06-06 20:29:01 +00002893</pre>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +00002894
Misha Brukman9d0919f2003-11-08 01:05:38 +00002895</div>
Chris Lattner00950542001-06-06 20:29:01 +00002896<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00002897<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002898
Misha Brukman9d0919f2003-11-08 01:05:38 +00002899<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002900
Chris Lattner00950542001-06-06 20:29:01 +00002901<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002902<pre>
2903 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 +00002904</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002905
Chris Lattner00950542001-06-06 20:29:01 +00002906<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002907<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2908 different basic block in the current function. There are two forms of this
2909 instruction, corresponding to a conditional branch and an unconditional
2910 branch.</p>
2911
Chris Lattner00950542001-06-06 20:29:01 +00002912<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002913<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2914 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2915 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2916 target.</p>
2917
Chris Lattner00950542001-06-06 20:29:01 +00002918<h5>Semantics:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +00002919<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002920 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2921 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2922 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2923
Chris Lattner00950542001-06-06 20:29:01 +00002924<h5>Example:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +00002925<pre>
2926Test:
2927 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2928 br i1 %cond, label %IfEqual, label %IfUnequal
2929IfEqual:
2930 <a href="#i_ret">ret</a> i32 1
2931IfUnequal:
2932 <a href="#i_ret">ret</a> i32 0
2933</pre>
2934
Misha Brukman9d0919f2003-11-08 01:05:38 +00002935</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002936
Chris Lattner00950542001-06-06 20:29:01 +00002937<!-- _______________________________________________________________________ -->
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002938<div class="doc_subsubsection">
2939 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2940</div>
2941
Misha Brukman9d0919f2003-11-08 01:05:38 +00002942<div class="doc_text">
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002943
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002944<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002945<pre>
2946 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2947</pre>
2948
Chris Lattner00950542001-06-06 20:29:01 +00002949<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002950<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002951 several different places. It is a generalization of the '<tt>br</tt>'
2952 instruction, allowing a branch to occur to one of many possible
2953 destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002954
Chris Lattner00950542001-06-06 20:29:01 +00002955<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002956<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002957 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2958 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2959 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002960
Chris Lattner00950542001-06-06 20:29:01 +00002961<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +00002962<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002963 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2964 is searched for the given value. If the value is found, control flow is
Benjamin Kramer8040cd32009-10-12 14:46:08 +00002965 transferred to the corresponding destination; otherwise, control flow is
2966 transferred to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +00002967
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002968<h5>Implementation:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002969<p>Depending on properties of the target machine and the particular
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002970 <tt>switch</tt> instruction, this instruction may be code generated in
2971 different ways. For example, it could be generated as a series of chained
2972 conditional branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002973
2974<h5>Example:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002975<pre>
2976 <i>; Emulate a conditional br instruction</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00002977 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman2a08c532009-01-04 23:44:43 +00002978 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002979
2980 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +00002981 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +00002982
2983 <i>; Implement a jump table:</i>
Dan Gohman2a08c532009-01-04 23:44:43 +00002984 switch i32 %val, label %otherwise [ i32 0, label %onzero
2985 i32 1, label %onone
2986 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +00002987</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00002988
Misha Brukman9d0919f2003-11-08 01:05:38 +00002989</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00002990
Chris Lattnerf9be95f2009-10-27 19:13:16 +00002991
2992<!-- _______________________________________________________________________ -->
2993<div class="doc_subsubsection">
Chris Lattnerab21db72009-10-28 00:19:10 +00002994 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
Chris Lattnerf9be95f2009-10-27 19:13:16 +00002995</div>
2996
2997<div class="doc_text">
2998
2999<h5>Syntax:</h5>
3000<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +00003001 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003002</pre>
3003
3004<h5>Overview:</h5>
3005
Chris Lattnerab21db72009-10-28 00:19:10 +00003006<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003007 within the current function, whose address is specified by
Chris Lattnerc6f44362009-10-27 21:01:34 +00003008 "<tt>address</tt>". Address must be derived from a <a
3009 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003010
3011<h5>Arguments:</h5>
3012
3013<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3014 rest of the arguments indicate the full set of possible destinations that the
3015 address may point to. Blocks are allowed to occur multiple times in the
3016 destination list, though this isn't particularly useful.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003017
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003018<p>This destination list is required so that dataflow analysis has an accurate
3019 understanding of the CFG.</p>
3020
3021<h5>Semantics:</h5>
3022
3023<p>Control transfers to the block specified in the address argument. All
3024 possible destination blocks must be listed in the label list, otherwise this
3025 instruction has undefined behavior. This implies that jumps to labels
3026 defined in other functions have undefined behavior as well.</p>
3027
3028<h5>Implementation:</h5>
3029
3030<p>This is typically implemented with a jump through a register.</p>
3031
3032<h5>Example:</h5>
3033<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +00003034 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +00003035</pre>
3036
3037</div>
3038
3039
Chris Lattner00950542001-06-06 20:29:01 +00003040<!-- _______________________________________________________________________ -->
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003041<div class="doc_subsubsection">
3042 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3043</div>
3044
Misha Brukman9d0919f2003-11-08 01:05:38 +00003045<div class="doc_text">
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003046
Chris Lattner00950542001-06-06 20:29:01 +00003047<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003048<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +00003049 &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 +00003050 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003051</pre>
3052
Chris Lattner6536cfe2002-05-06 22:08:29 +00003053<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003054<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003055 function, with the possibility of control flow transfer to either the
3056 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3057 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3058 control flow will return to the "normal" label. If the callee (or any
3059 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3060 instruction, control is interrupted and continued at the dynamically nearest
3061 "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003062
Chris Lattner00950542001-06-06 20:29:01 +00003063<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003064<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003065
Chris Lattner00950542001-06-06 20:29:01 +00003066<ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003067 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3068 convention</a> the call should use. If none is specified, the call
3069 defaults to using C calling conventions.</li>
Devang Patelf642f472008-10-06 18:50:38 +00003070
3071 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003072 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3073 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelf642f472008-10-06 18:50:38 +00003074
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003075 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003076 function value being invoked. In most cases, this is a direct function
3077 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3078 off an arbitrary pointer to function value.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003079
3080 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003081 function to be invoked. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003082
3083 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +00003084 signature argument types and parameter attributes. All arguments must be
3085 of <a href="#t_firstclass">first class</a> type. If the function
3086 signature indicates the function accepts a variable number of arguments,
3087 the extra arguments can be specified.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003088
3089 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003090 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003091
3092 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003093 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003094
Devang Patel307e8ab2008-10-07 17:48:33 +00003095 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003096 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3097 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner00950542001-06-06 20:29:01 +00003098</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003099
Chris Lattner00950542001-06-06 20:29:01 +00003100<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003101<p>This instruction is designed to operate as a standard
3102 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3103 primary difference is that it establishes an association with a label, which
3104 is used by the runtime library to unwind the stack.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003105
3106<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003107 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3108 exception. Additionally, this is important for implementation of
3109 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003110
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003111<p>For the purposes of the SSA form, the definition of the value returned by the
3112 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3113 block to the "normal" label. If the callee unwinds then no return value is
3114 available.</p>
Dan Gohmanf96a4992009-05-22 21:47:08 +00003115
Chris Lattnerdf7a6802010-01-15 18:08:37 +00003116<p>Note that the code generator does not yet completely support unwind, and
3117that the invoke/unwind semantics are likely to change in future versions.</p>
3118
Chris Lattner00950542001-06-06 20:29:01 +00003119<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +00003120<pre>
Nick Lewyckyd703f652008-03-16 07:18:12 +00003121 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +00003122 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckyd703f652008-03-16 07:18:12 +00003123 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +00003124 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +00003125</pre>
Chris Lattner35eca582004-10-16 18:04:13 +00003126
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003127</div>
Chris Lattner35eca582004-10-16 18:04:13 +00003128
Chris Lattner27f71f22003-09-03 00:41:47 +00003129<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +00003130
Chris Lattner261efe92003-11-25 01:02:51 +00003131<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
3132Instruction</a> </div>
Chris Lattner35eca582004-10-16 18:04:13 +00003133
Misha Brukman9d0919f2003-11-08 01:05:38 +00003134<div class="doc_text">
Chris Lattner35eca582004-10-16 18:04:13 +00003135
Chris Lattner27f71f22003-09-03 00:41:47 +00003136<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003137<pre>
3138 unwind
3139</pre>
3140
Chris Lattner27f71f22003-09-03 00:41:47 +00003141<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003142<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003143 at the first callee in the dynamic call stack which used
3144 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3145 This is primarily used to implement exception handling.</p>
Chris Lattner35eca582004-10-16 18:04:13 +00003146
Chris Lattner27f71f22003-09-03 00:41:47 +00003147<h5>Semantics:</h5>
Chris Lattner72ed2002008-04-19 21:01:16 +00003148<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003149 immediately halt. The dynamic call stack is then searched for the
3150 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3151 Once found, execution continues at the "exceptional" destination block
3152 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3153 instruction in the dynamic call chain, undefined behavior results.</p>
3154
Chris Lattnerdf7a6802010-01-15 18:08:37 +00003155<p>Note that the code generator does not yet completely support unwind, and
3156that the invoke/unwind semantics are likely to change in future versions.</p>
3157
Misha Brukman9d0919f2003-11-08 01:05:38 +00003158</div>
Chris Lattner35eca582004-10-16 18:04:13 +00003159
3160<!-- _______________________________________________________________________ -->
3161
3162<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
3163Instruction</a> </div>
3164
3165<div class="doc_text">
3166
3167<h5>Syntax:</h5>
3168<pre>
3169 unreachable
3170</pre>
3171
3172<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003173<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003174 instruction is used to inform the optimizer that a particular portion of the
3175 code is not reachable. This can be used to indicate that the code after a
3176 no-return function cannot be reached, and other facts.</p>
Chris Lattner35eca582004-10-16 18:04:13 +00003177
3178<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +00003179<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003180
Chris Lattner35eca582004-10-16 18:04:13 +00003181</div>
3182
Chris Lattner00950542001-06-06 20:29:01 +00003183<!-- ======================================================================= -->
Chris Lattner261efe92003-11-25 01:02:51 +00003184<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003185
Misha Brukman9d0919f2003-11-08 01:05:38 +00003186<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003187
3188<p>Binary operators are used to do most of the computation in a program. They
3189 require two operands of the same type, execute an operation on them, and
3190 produce a single value. The operands might represent multiple data, as is
3191 the case with the <a href="#t_vector">vector</a> data type. The result value
3192 has the same type as its operands.</p>
3193
Misha Brukman9d0919f2003-11-08 01:05:38 +00003194<p>There are several different binary operators:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003195
Misha Brukman9d0919f2003-11-08 01:05:38 +00003196</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003197
Chris Lattner00950542001-06-06 20:29:01 +00003198<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003199<div class="doc_subsubsection">
3200 <a name="i_add">'<tt>add</tt>' Instruction</a>
3201</div>
3202
Misha Brukman9d0919f2003-11-08 01:05:38 +00003203<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003204
Chris Lattner00950542001-06-06 20:29:01 +00003205<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003206<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003207 &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 +00003208 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3209 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3210 &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 +00003211</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003212
Chris Lattner00950542001-06-06 20:29:01 +00003213<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003214<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003215
Chris Lattner00950542001-06-06 20:29:01 +00003216<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003217<p>The two arguments to the '<tt>add</tt>' instruction must
3218 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3219 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003220
Chris Lattner00950542001-06-06 20:29:01 +00003221<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003222<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003223
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003224<p>If the sum has unsigned overflow, the result returned is the mathematical
3225 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003226
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003227<p>Because LLVM integers use a two's complement representation, this instruction
3228 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003229
Dan Gohman08d012e2009-07-22 22:44:56 +00003230<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3231 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3232 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +00003233 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3234 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003235
Chris Lattner00950542001-06-06 20:29:01 +00003236<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003237<pre>
3238 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00003239</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003240
Misha Brukman9d0919f2003-11-08 01:05:38 +00003241</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003242
Chris Lattner00950542001-06-06 20:29:01 +00003243<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003244<div class="doc_subsubsection">
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003245 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3246</div>
3247
3248<div class="doc_text">
3249
3250<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003251<pre>
3252 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3253</pre>
3254
3255<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003256<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3257
3258<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003259<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003260 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3261 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003262
3263<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003264<p>The value produced is the floating point sum of the two operands.</p>
3265
3266<h5>Example:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003267<pre>
3268 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3269</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003270
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003271</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003272
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003273<!-- _______________________________________________________________________ -->
3274<div class="doc_subsubsection">
Chris Lattner5568e942008-05-20 20:48:21 +00003275 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3276</div>
3277
Misha Brukman9d0919f2003-11-08 01:05:38 +00003278<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003279
Chris Lattner00950542001-06-06 20:29:01 +00003280<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003281<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +00003282 &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 +00003283 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3284 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3285 &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 +00003286</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003287
Chris Lattner00950542001-06-06 20:29:01 +00003288<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003289<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003290 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003291
3292<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003293 '<tt>neg</tt>' instruction present in most other intermediate
3294 representations.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003295
Chris Lattner00950542001-06-06 20:29:01 +00003296<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003297<p>The two arguments to the '<tt>sub</tt>' instruction must
3298 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3299 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003300
Chris Lattner00950542001-06-06 20:29:01 +00003301<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003302<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003303
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003304<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003305 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3306 result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003307
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003308<p>Because LLVM integers use a two's complement representation, this instruction
3309 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003310
Dan Gohman08d012e2009-07-22 22:44:56 +00003311<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3312 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3313 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +00003314 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3315 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003316
Chris Lattner00950542001-06-06 20:29:01 +00003317<h5>Example:</h5>
Bill Wendlingaac388b2007-05-29 09:42:13 +00003318<pre>
3319 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003320 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +00003321</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003322
Misha Brukman9d0919f2003-11-08 01:05:38 +00003323</div>
Chris Lattner5568e942008-05-20 20:48:21 +00003324
Chris Lattner00950542001-06-06 20:29:01 +00003325<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003326<div class="doc_subsubsection">
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003327 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3328</div>
3329
3330<div class="doc_text">
3331
3332<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003333<pre>
3334 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3335</pre>
3336
3337<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003338<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003339 operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003340
3341<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003342 '<tt>fneg</tt>' instruction present in most other intermediate
3343 representations.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003344
3345<h5>Arguments:</h5>
Bill Wendlingd9fe2982009-07-20 02:32:41 +00003346<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003347 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3348 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003349
3350<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003351<p>The value produced is the floating point difference of the two operands.</p>
3352
3353<h5>Example:</h5>
3354<pre>
3355 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3356 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3357</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003358
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003359</div>
3360
3361<!-- _______________________________________________________________________ -->
3362<div class="doc_subsubsection">
Chris Lattner5568e942008-05-20 20:48:21 +00003363 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3364</div>
3365
Misha Brukman9d0919f2003-11-08 01:05:38 +00003366<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003367
Chris Lattner00950542001-06-06 20:29:01 +00003368<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003369<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +00003370 &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 +00003371 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3372 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3373 &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 +00003374</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003375
Chris Lattner00950542001-06-06 20:29:01 +00003376<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003377<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003378
Chris Lattner00950542001-06-06 20:29:01 +00003379<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003380<p>The two arguments to the '<tt>mul</tt>' instruction must
3381 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3382 integer values. Both arguments must have identical types.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003383
Chris Lattner00950542001-06-06 20:29:01 +00003384<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003385<p>The value produced is the integer product of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003386
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003387<p>If the result of the multiplication has unsigned overflow, the result
3388 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3389 width of the result.</p>
3390
3391<p>Because LLVM integers use a two's complement representation, and the result
3392 is the same width as the operands, this instruction returns the correct
3393 result for both signed and unsigned integers. If a full product
3394 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3395 be sign-extended or zero-extended as appropriate to the width of the full
3396 product.</p>
3397
Dan Gohman08d012e2009-07-22 22:44:56 +00003398<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3399 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3400 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +00003401 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3402 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003403
Chris Lattner00950542001-06-06 20:29:01 +00003404<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003405<pre>
3406 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +00003407</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003408
Misha Brukman9d0919f2003-11-08 01:05:38 +00003409</div>
Chris Lattner5568e942008-05-20 20:48:21 +00003410
Chris Lattner00950542001-06-06 20:29:01 +00003411<!-- _______________________________________________________________________ -->
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003412<div class="doc_subsubsection">
3413 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3414</div>
3415
3416<div class="doc_text">
3417
3418<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003419<pre>
3420 &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 +00003421</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003422
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003423<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003424<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003425
3426<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003427<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003428 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3429 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003430
3431<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003432<p>The value produced is the floating point product of the two operands.</p>
3433
3434<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003435<pre>
3436 &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003437</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003438
Dan Gohmanae3a0be2009-06-04 22:49:04 +00003439</div>
3440
3441<!-- _______________________________________________________________________ -->
Reid Spencer1628cec2006-10-26 06:15:43 +00003442<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3443</a></div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003444
Reid Spencer1628cec2006-10-26 06:15:43 +00003445<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003446
Reid Spencer1628cec2006-10-26 06:15:43 +00003447<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003448<pre>
Chris Lattner35bda892011-02-06 21:44:57 +00003449 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3450 &lt;result&gt; = udiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003451</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003452
Reid Spencer1628cec2006-10-26 06:15:43 +00003453<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003454<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003455
Reid Spencer1628cec2006-10-26 06:15:43 +00003456<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003457<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003458 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3459 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003460
Reid Spencer1628cec2006-10-26 06:15:43 +00003461<h5>Semantics:</h5>
Chris Lattner5ec89832008-01-28 00:36:27 +00003462<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003463
Chris Lattner5ec89832008-01-28 00:36:27 +00003464<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003465 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3466
Chris Lattner5ec89832008-01-28 00:36:27 +00003467<p>Division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003468
Chris Lattner35bda892011-02-06 21:44:57 +00003469<p>If the <tt>exact</tt> keyword is present, the result value of the
3470 <tt>udiv</tt> is a <a href="#trapvalues">trap value</a> if %op1 is not a
3471 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
3472
3473
Reid Spencer1628cec2006-10-26 06:15:43 +00003474<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003475<pre>
3476 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003477</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003478
Reid Spencer1628cec2006-10-26 06:15:43 +00003479</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003480
Reid Spencer1628cec2006-10-26 06:15:43 +00003481<!-- _______________________________________________________________________ -->
3482<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3483</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003484
Reid Spencer1628cec2006-10-26 06:15:43 +00003485<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003486
Reid Spencer1628cec2006-10-26 06:15:43 +00003487<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003488<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003489 &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 +00003490 &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 +00003491</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003492
Reid Spencer1628cec2006-10-26 06:15:43 +00003493<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003494<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003495
Reid Spencer1628cec2006-10-26 06:15:43 +00003496<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003497<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003498 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3499 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003500
Reid Spencer1628cec2006-10-26 06:15:43 +00003501<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003502<p>The value produced is the signed integer quotient of the two operands rounded
3503 towards zero.</p>
3504
Chris Lattner5ec89832008-01-28 00:36:27 +00003505<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003506 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3507
Chris Lattner5ec89832008-01-28 00:36:27 +00003508<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003509 undefined behavior; this is a rare case, but can occur, for example, by doing
3510 a 32-bit division of -2147483648 by -1.</p>
3511
Dan Gohman9c5beed2009-07-22 00:04:19 +00003512<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman27ef9972010-04-23 15:23:32 +00003513 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
Dan Gohman38da9272010-07-11 00:08:34 +00003514 be rounded.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +00003515
Reid Spencer1628cec2006-10-26 06:15:43 +00003516<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003517<pre>
3518 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +00003519</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003520
Reid Spencer1628cec2006-10-26 06:15:43 +00003521</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003522
Reid Spencer1628cec2006-10-26 06:15:43 +00003523<!-- _______________________________________________________________________ -->
3524<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner261efe92003-11-25 01:02:51 +00003525Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003526
Misha Brukman9d0919f2003-11-08 01:05:38 +00003527<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003528
Chris Lattner00950542001-06-06 20:29:01 +00003529<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003530<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +00003531 &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 +00003532</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003533
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003534<h5>Overview:</h5>
3535<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003536
Chris Lattner261efe92003-11-25 01:02:51 +00003537<h5>Arguments:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00003538<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003539 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3540 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003541
Chris Lattner261efe92003-11-25 01:02:51 +00003542<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +00003543<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003544
Chris Lattner261efe92003-11-25 01:02:51 +00003545<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003546<pre>
3547 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +00003548</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003549
Chris Lattner261efe92003-11-25 01:02:51 +00003550</div>
Chris Lattner5568e942008-05-20 20:48:21 +00003551
Chris Lattner261efe92003-11-25 01:02:51 +00003552<!-- _______________________________________________________________________ -->
Reid Spencer0a783f72006-11-02 01:53:59 +00003553<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3554</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003555
Reid Spencer0a783f72006-11-02 01:53:59 +00003556<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003557
Reid Spencer0a783f72006-11-02 01:53:59 +00003558<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003559<pre>
3560 &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 +00003561</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003562
Reid Spencer0a783f72006-11-02 01:53:59 +00003563<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003564<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3565 division of its two arguments.</p>
3566
Reid Spencer0a783f72006-11-02 01:53:59 +00003567<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003568<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003569 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3570 values. Both arguments must have identical types.</p>
3571
Reid Spencer0a783f72006-11-02 01:53:59 +00003572<h5>Semantics:</h5>
3573<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003574 This instruction always performs an unsigned division to get the
3575 remainder.</p>
3576
Chris Lattner5ec89832008-01-28 00:36:27 +00003577<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003578 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3579
Chris Lattner5ec89832008-01-28 00:36:27 +00003580<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003581
Reid Spencer0a783f72006-11-02 01:53:59 +00003582<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003583<pre>
3584 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +00003585</pre>
3586
3587</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003588
Reid Spencer0a783f72006-11-02 01:53:59 +00003589<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003590<div class="doc_subsubsection">
3591 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3592</div>
3593
Chris Lattner261efe92003-11-25 01:02:51 +00003594<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003595
Chris Lattner261efe92003-11-25 01:02:51 +00003596<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003597<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +00003598 &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 +00003599</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003600
Chris Lattner261efe92003-11-25 01:02:51 +00003601<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003602<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3603 division of its two operands. This instruction can also take
3604 <a href="#t_vector">vector</a> versions of the values in which case the
3605 elements must be integers.</p>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +00003606
Chris Lattner261efe92003-11-25 01:02:51 +00003607<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003608<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003609 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3610 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003611
Chris Lattner261efe92003-11-25 01:02:51 +00003612<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +00003613<p>This instruction returns the <i>remainder</i> of a division (where the result
Duncan Sandsdea3a5e2011-03-07 09:12:24 +00003614 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
3615 <i>modulo</i> operator (where the result is either zero or has the same sign
3616 as the divisor, <tt>op2</tt>) of a value.
3617 For more information about the difference,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003618 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3619 Math Forum</a>. For a table of how this is implemented in various languages,
3620 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3621 Wikipedia: modulo operation</a>.</p>
3622
Chris Lattner5ec89832008-01-28 00:36:27 +00003623<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003624 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3625
Chris Lattner5ec89832008-01-28 00:36:27 +00003626<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003627 Overflow also leads to undefined behavior; this is a rare case, but can
3628 occur, for example, by taking the remainder of a 32-bit division of
3629 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3630 lets srem be implemented using instructions that return both the result of
3631 the division and the remainder.)</p>
3632
Chris Lattner261efe92003-11-25 01:02:51 +00003633<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003634<pre>
3635 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +00003636</pre>
3637
3638</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003639
Reid Spencer0a783f72006-11-02 01:53:59 +00003640<!-- _______________________________________________________________________ -->
Chris Lattner5568e942008-05-20 20:48:21 +00003641<div class="doc_subsubsection">
3642 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3643
Reid Spencer0a783f72006-11-02 01:53:59 +00003644<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003645
Reid Spencer0a783f72006-11-02 01:53:59 +00003646<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003647<pre>
3648 &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 +00003649</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003650
Reid Spencer0a783f72006-11-02 01:53:59 +00003651<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003652<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3653 its two operands.</p>
3654
Reid Spencer0a783f72006-11-02 01:53:59 +00003655<h5>Arguments:</h5>
3656<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003657 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3658 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003659
Reid Spencer0a783f72006-11-02 01:53:59 +00003660<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003661<p>This instruction returns the <i>remainder</i> of a division. The remainder
3662 has the same sign as the dividend.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003663
Reid Spencer0a783f72006-11-02 01:53:59 +00003664<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003665<pre>
3666 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +00003667</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003668
Misha Brukman9d0919f2003-11-08 01:05:38 +00003669</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +00003670
Reid Spencer8e11bf82007-02-02 13:57:07 +00003671<!-- ======================================================================= -->
3672<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3673Operations</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003674
Reid Spencer8e11bf82007-02-02 13:57:07 +00003675<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003676
3677<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3678 program. They are generally very efficient instructions and can commonly be
3679 strength reduced from other instructions. They require two operands of the
3680 same type, execute an operation on them, and produce a single value. The
3681 resulting value is the same type as its operands.</p>
3682
Reid Spencer8e11bf82007-02-02 13:57:07 +00003683</div>
3684
Reid Spencer569f2fa2007-01-31 21:39:12 +00003685<!-- _______________________________________________________________________ -->
3686<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3687Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003688
Reid Spencer569f2fa2007-01-31 21:39:12 +00003689<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003690
Reid Spencer569f2fa2007-01-31 21:39:12 +00003691<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003692<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +00003693 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3694 &lt;result&gt; = shl nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3695 &lt;result&gt; = shl nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3696 &lt;result&gt; = shl nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003697</pre>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003698
Reid Spencer569f2fa2007-01-31 21:39:12 +00003699<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003700<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3701 a specified number of bits.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003702
Reid Spencer569f2fa2007-01-31 21:39:12 +00003703<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003704<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3705 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3706 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003707
Reid Spencer569f2fa2007-01-31 21:39:12 +00003708<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003709<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3710 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3711 is (statically or dynamically) negative or equal to or larger than the number
3712 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3713 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3714 shift amount in <tt>op2</tt>.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003715
Chris Lattnerf067d582011-02-07 16:40:21 +00003716<p>If the <tt>nuw</tt> keyword is present, then the shift produces a
3717 <a href="#trapvalues">trap value</a> if it shifts out any non-zero bits. If
Chris Lattner66298c12011-02-09 16:44:44 +00003718 the <tt>nsw</tt> keyword is present, then the shift produces a
Chris Lattnerf067d582011-02-07 16:40:21 +00003719 <a href="#trapvalues">trap value</a> if it shifts out any bits that disagree
3720 with the resultant sign bit. As such, NUW/NSW have the same semantics as
3721 they would if the shift were expressed as a mul instruction with the same
3722 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
3723
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003724<h5>Example:</h5>
3725<pre>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003726 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3727 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3728 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003729 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +00003730 &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 +00003731</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003732
Reid Spencer569f2fa2007-01-31 21:39:12 +00003733</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003734
Reid Spencer569f2fa2007-01-31 21:39:12 +00003735<!-- _______________________________________________________________________ -->
3736<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3737Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003738
Reid Spencer569f2fa2007-01-31 21:39:12 +00003739<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003740
Reid Spencer569f2fa2007-01-31 21:39:12 +00003741<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003742<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +00003743 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3744 &lt;result&gt; = lshr exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003745</pre>
3746
3747<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003748<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3749 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003750
3751<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003752<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003753 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3754 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003755
3756<h5>Semantics:</h5>
3757<p>This instruction always performs a logical shift right operation. The most
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003758 significant bits of the result will be filled with zero bits after the shift.
3759 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3760 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3761 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3762 shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003763
Chris Lattnerf067d582011-02-07 16:40:21 +00003764<p>If the <tt>exact</tt> keyword is present, the result value of the
3765 <tt>lshr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
3766 shifted out are non-zero.</p>
3767
3768
Reid Spencer569f2fa2007-01-31 21:39:12 +00003769<h5>Example:</h5>
3770<pre>
3771 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3772 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3773 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3774 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003775 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +00003776 &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 +00003777</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003778
Reid Spencer569f2fa2007-01-31 21:39:12 +00003779</div>
3780
Reid Spencer8e11bf82007-02-02 13:57:07 +00003781<!-- _______________________________________________________________________ -->
Reid Spencer569f2fa2007-01-31 21:39:12 +00003782<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3783Instruction</a> </div>
3784<div class="doc_text">
3785
3786<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003787<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +00003788 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3789 &lt;result&gt; = ashr exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003790</pre>
3791
3792<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003793<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3794 operand shifted to the right a specified number of bits with sign
3795 extension.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003796
3797<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003798<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003799 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3800 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003801
3802<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003803<p>This instruction always performs an arithmetic shift right operation, The
3804 most significant bits of the result will be filled with the sign bit
3805 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3806 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3807 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3808 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +00003809
Chris Lattnerf067d582011-02-07 16:40:21 +00003810<p>If the <tt>exact</tt> keyword is present, the result value of the
3811 <tt>ashr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
3812 shifted out are non-zero.</p>
3813
Reid Spencer569f2fa2007-01-31 21:39:12 +00003814<h5>Example:</h5>
3815<pre>
3816 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3817 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3818 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3819 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +00003820 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +00003821 &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 +00003822</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003823
Reid Spencer569f2fa2007-01-31 21:39:12 +00003824</div>
3825
Chris Lattner00950542001-06-06 20:29:01 +00003826<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00003827<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3828Instruction</a> </div>
Chris Lattner5568e942008-05-20 20:48:21 +00003829
Misha Brukman9d0919f2003-11-08 01:05:38 +00003830<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00003831
Chris Lattner00950542001-06-06 20:29:01 +00003832<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003833<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +00003834 &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 +00003835</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00003836
Chris Lattner00950542001-06-06 20:29:01 +00003837<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003838<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3839 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003840
Chris Lattner00950542001-06-06 20:29:01 +00003841<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003842<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003843 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3844 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003845
Chris Lattner00950542001-06-06 20:29:01 +00003846<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003847<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003848
Misha Brukman9d0919f2003-11-08 01:05:38 +00003849<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +00003850 <tbody>
3851 <tr>
3852 <td>In0</td>
3853 <td>In1</td>
3854 <td>Out</td>
3855 </tr>
3856 <tr>
3857 <td>0</td>
3858 <td>0</td>
3859 <td>0</td>
3860 </tr>
3861 <tr>
3862 <td>0</td>
3863 <td>1</td>
3864 <td>0</td>
3865 </tr>
3866 <tr>
3867 <td>1</td>
3868 <td>0</td>
3869 <td>0</td>
3870 </tr>
3871 <tr>
3872 <td>1</td>
3873 <td>1</td>
3874 <td>1</td>
3875 </tr>
3876 </tbody>
3877</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003878
Chris Lattner00950542001-06-06 20:29:01 +00003879<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00003880<pre>
3881 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003882 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3883 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +00003884</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003885</div>
Chris Lattner00950542001-06-06 20:29:01 +00003886<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00003887<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattner5568e942008-05-20 20:48:21 +00003888
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003889<div class="doc_text">
3890
3891<h5>Syntax:</h5>
3892<pre>
3893 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3894</pre>
3895
3896<h5>Overview:</h5>
3897<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3898 two operands.</p>
3899
3900<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003901<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003902 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3903 values. Both arguments must have identical types.</p>
3904
Chris Lattner00950542001-06-06 20:29:01 +00003905<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003906<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003907
Chris Lattner261efe92003-11-25 01:02:51 +00003908<table border="1" cellspacing="0" cellpadding="4">
3909 <tbody>
3910 <tr>
3911 <td>In0</td>
3912 <td>In1</td>
3913 <td>Out</td>
3914 </tr>
3915 <tr>
3916 <td>0</td>
3917 <td>0</td>
3918 <td>0</td>
3919 </tr>
3920 <tr>
3921 <td>0</td>
3922 <td>1</td>
3923 <td>1</td>
3924 </tr>
3925 <tr>
3926 <td>1</td>
3927 <td>0</td>
3928 <td>1</td>
3929 </tr>
3930 <tr>
3931 <td>1</td>
3932 <td>1</td>
3933 <td>1</td>
3934 </tr>
3935 </tbody>
3936</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003937
Chris Lattner00950542001-06-06 20:29:01 +00003938<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003939<pre>
3940 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00003941 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3942 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +00003943</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003944
Misha Brukman9d0919f2003-11-08 01:05:38 +00003945</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003946
Chris Lattner00950542001-06-06 20:29:01 +00003947<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00003948<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3949Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003950
Misha Brukman9d0919f2003-11-08 01:05:38 +00003951<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003952
Chris Lattner00950542001-06-06 20:29:01 +00003953<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003954<pre>
3955 &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 +00003956</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003957
Chris Lattner00950542001-06-06 20:29:01 +00003958<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003959<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3960 its two operands. The <tt>xor</tt> is used to implement the "one's
3961 complement" operation, which is the "~" operator in C.</p>
3962
Chris Lattner00950542001-06-06 20:29:01 +00003963<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00003964<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003965 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3966 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00003967
Chris Lattner00950542001-06-06 20:29:01 +00003968<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00003969<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00003970
Chris Lattner261efe92003-11-25 01:02:51 +00003971<table border="1" cellspacing="0" cellpadding="4">
3972 <tbody>
3973 <tr>
3974 <td>In0</td>
3975 <td>In1</td>
3976 <td>Out</td>
3977 </tr>
3978 <tr>
3979 <td>0</td>
3980 <td>0</td>
3981 <td>0</td>
3982 </tr>
3983 <tr>
3984 <td>0</td>
3985 <td>1</td>
3986 <td>1</td>
3987 </tr>
3988 <tr>
3989 <td>1</td>
3990 <td>0</td>
3991 <td>1</td>
3992 </tr>
3993 <tr>
3994 <td>1</td>
3995 <td>1</td>
3996 <td>0</td>
3997 </tr>
3998 </tbody>
3999</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004000
Chris Lattner00950542001-06-06 20:29:01 +00004001<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004002<pre>
4003 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +00004004 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4005 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4006 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +00004007</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004008
Misha Brukman9d0919f2003-11-08 01:05:38 +00004009</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004010
Chris Lattner00950542001-06-06 20:29:01 +00004011<!-- ======================================================================= -->
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004012<div class="doc_subsection">
Chris Lattner3df241e2006-04-08 23:07:04 +00004013 <a name="vectorops">Vector Operations</a>
4014</div>
4015
4016<div class="doc_text">
4017
4018<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004019 target-independent manner. These instructions cover the element-access and
4020 vector-specific operations needed to process vectors effectively. While LLVM
4021 does directly support these vector operations, many sophisticated algorithms
4022 will want to use target-specific intrinsics to take full advantage of a
4023 specific target.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004024
4025</div>
4026
4027<!-- _______________________________________________________________________ -->
4028<div class="doc_subsubsection">
4029 <a name="i_extractelement">'<tt>extractelement</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>
Reid Spencerca86e162006-12-31 07:07:53 +00004036 &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 +00004037</pre>
4038
4039<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004040<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4041 from a vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004042
4043
4044<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004045<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4046 of <a href="#t_vector">vector</a> type. The second operand is an index
4047 indicating the position from which to extract the element. The index may be
4048 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 scalar of the same type as the element type of
4052 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4053 <tt>val</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; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattner3df241e2006-04-08 23:07:04 +00004059</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +00004060
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004061</div>
Chris Lattner3df241e2006-04-08 23:07:04 +00004062
4063<!-- _______________________________________________________________________ -->
4064<div class="doc_subsubsection">
4065 <a name="i_insertelement">'<tt>insertelement</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>
Dan Gohmanf3480b92008-05-12 23:38:42 +00004072 &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 +00004073</pre>
4074
4075<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004076<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4077 vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004078
4079<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004080<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4081 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4082 whose type must equal the element type of the first operand. The third
4083 operand is an index indicating the position at which to insert the value.
4084 The index may be a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004085
4086<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004087<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4088 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4089 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4090 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004091
4092<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004093<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +00004094 &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 +00004095</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004096
Chris Lattner3df241e2006-04-08 23:07:04 +00004097</div>
4098
4099<!-- _______________________________________________________________________ -->
4100<div class="doc_subsubsection">
4101 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4102</div>
4103
4104<div class="doc_text">
4105
4106<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004107<pre>
Mon P Wangaeb06d22008-11-10 04:46:22 +00004108 &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 +00004109</pre>
4110
4111<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004112<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4113 from two input vectors, returning a vector with the same element type as the
4114 input and length that is the same as the shuffle mask.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004115
4116<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004117<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4118 with types that match each other. The third argument is a shuffle mask whose
4119 element type is always 'i32'. The result of the instruction is a vector
4120 whose length is the same as the shuffle mask and whose element type is the
4121 same as the element type of the first two operands.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004122
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004123<p>The shuffle mask operand is required to be a constant vector with either
4124 constant integer or undef values.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004125
4126<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004127<p>The elements of the two input vectors are numbered from left to right across
4128 both of the vectors. The shuffle mask operand specifies, for each element of
4129 the result vector, which element of the two input vectors the result element
4130 gets. The element selector may be undef (meaning "don't care") and the
4131 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +00004132
4133<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +00004134<pre>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004135 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen6f1cc772007-04-22 01:17:39 +00004136 &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 +00004137 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Reid Spencerca86e162006-12-31 07:07:53 +00004138 &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 +00004139 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wangaeb06d22008-11-10 04:46:22 +00004140 &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 +00004141 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wangaeb06d22008-11-10 04:46:22 +00004142 &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 +00004143</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +00004144
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004145</div>
Tanya Lattner09474292006-04-14 19:24:33 +00004146
Chris Lattner3df241e2006-04-08 23:07:04 +00004147<!-- ======================================================================= -->
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004148<div class="doc_subsection">
Dan Gohmana334d5f2008-05-12 23:51:09 +00004149 <a name="aggregateops">Aggregate Operations</a>
4150</div>
4151
4152<div class="doc_text">
4153
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004154<p>LLVM supports several instructions for working with
4155 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004156
4157</div>
4158
4159<!-- _______________________________________________________________________ -->
4160<div class="doc_subsubsection">
4161 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4162</div>
4163
4164<div class="doc_text">
4165
4166<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004167<pre>
4168 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4169</pre>
4170
4171<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004172<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4173 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004174
4175<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004176<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +00004177 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004178 <a href="#t_array">array</a> type. The operands are constant indices to
4179 specify which value to extract in a similar manner as indices in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004180 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Frits van Bommel13242892010-12-05 20:54:38 +00004181 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4182 <ul>
4183 <li>Since the value being indexed is not a pointer, the first index is
4184 omitted and assumed to be zero.</li>
4185 <li>At least one index must be specified.</li>
4186 <li>Not only struct indices but also array indices must be in
4187 bounds.</li>
4188 </ul>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004189
4190<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004191<p>The result is the value at the position in the aggregate specified by the
4192 index operands.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004193
4194<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004195<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +00004196 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004197</pre>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004198
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004199</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004200
4201<!-- _______________________________________________________________________ -->
4202<div class="doc_subsubsection">
4203 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4204</div>
4205
4206<div class="doc_text">
4207
4208<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004209<pre>
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +00004210 &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 +00004211</pre>
4212
4213<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004214<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4215 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004216
4217<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004218<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +00004219 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004220 <a href="#t_array">array</a> type. The second operand is a first-class
4221 value to insert. The following operands are constant indices indicating
4222 the position at which to insert the value in a similar manner as indices in a
Frits van Bommel13242892010-12-05 20:54:38 +00004223 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004224 value to insert must have the same type as the value identified by the
4225 indices.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004226
4227<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004228<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4229 that of <tt>val</tt> except that the value at the position specified by the
4230 indices is that of <tt>elt</tt>.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004231
4232<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004233<pre>
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +00004234 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4235 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +00004236</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004237
Dan Gohmana334d5f2008-05-12 23:51:09 +00004238</div>
4239
4240
4241<!-- ======================================================================= -->
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004242<div class="doc_subsection">
Chris Lattner884a9702006-08-15 00:45:58 +00004243 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004244</div>
4245
Misha Brukman9d0919f2003-11-08 01:05:38 +00004246<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00004247
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004248<p>A key design point of an SSA-based representation is how it represents
4249 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez2fee2942009-10-26 23:44:29 +00004250 very simple. This section describes how to read, write, and allocate
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004251 memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004252
Misha Brukman9d0919f2003-11-08 01:05:38 +00004253</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004254
Chris Lattner00950542001-06-06 20:29:01 +00004255<!-- _______________________________________________________________________ -->
Chris Lattner2cbdc452005-11-06 08:02:57 +00004256<div class="doc_subsubsection">
Chris Lattner2cbdc452005-11-06 08:02:57 +00004257 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4258</div>
4259
Misha Brukman9d0919f2003-11-08 01:05:38 +00004260<div class="doc_text">
Chris Lattner2cbdc452005-11-06 08:02:57 +00004261
Chris Lattner00950542001-06-06 20:29:01 +00004262<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004263<pre>
Dan Gohmanf75a7d32010-05-28 01:14:11 +00004264 &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 +00004265</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004266
Chris Lattner00950542001-06-06 20:29:01 +00004267<h5>Overview:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00004268<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004269 currently executing function, to be automatically released when this function
4270 returns to its caller. The object is always allocated in the generic address
4271 space (address space zero).</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004272
Chris Lattner00950542001-06-06 20:29:01 +00004273<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004274<p>The '<tt>alloca</tt>' instruction
4275 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4276 runtime stack, returning a pointer of the appropriate type to the program.
4277 If "NumElements" is specified, it is the number of elements allocated,
4278 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4279 specified, the value result of the allocation is guaranteed to be aligned to
4280 at least that boundary. If not specified, or if zero, the target can choose
4281 to align the allocation on any convenient boundary compatible with the
4282 type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004283
Misha Brukman9d0919f2003-11-08 01:05:38 +00004284<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004285
Chris Lattner00950542001-06-06 20:29:01 +00004286<h5>Semantics:</h5>
Bill Wendling871eb0a2009-05-08 20:49:29 +00004287<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004288 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4289 memory is automatically released when the function returns. The
4290 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4291 variables that must have an address available. When the function returns
4292 (either with the <tt><a href="#i_ret">ret</a></tt>
4293 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4294 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004295
Chris Lattner00950542001-06-06 20:29:01 +00004296<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004297<pre>
Dan Gohman81e21672009-01-04 23:49:44 +00004298 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4299 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4300 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4301 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +00004302</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004303
Misha Brukman9d0919f2003-11-08 01:05:38 +00004304</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +00004305
Chris Lattner00950542001-06-06 20:29:01 +00004306<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00004307<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
4308Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004309
Misha Brukman9d0919f2003-11-08 01:05:38 +00004310<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004311
Chris Lattner2b7d3202002-05-06 03:03:22 +00004312<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004313<pre>
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004314 &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4315 &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4316 !&lt;index&gt; = !{ i32 1 }
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004317</pre>
4318
Chris Lattner2b7d3202002-05-06 03:03:22 +00004319<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004320<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004321
Chris Lattner2b7d3202002-05-06 03:03:22 +00004322<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004323<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4324 from which to load. The pointer must point to
4325 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4326 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00004327 number or order of execution of this <tt>load</tt> with other <a
4328 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004329
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004330<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004331 operation (that is, the alignment of the memory address). A value of 0 or an
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004332 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004333 alignment for the target. It is the responsibility of the code emitter to
4334 ensure that the alignment information is correct. Overestimating the
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004335 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004336 produce less efficient code. An alignment of 1 is always safe.</p>
4337
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004338<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4339 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004340 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendling7c78dbb2010-02-25 21:23:24 +00004341 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4342 and code generator that this load is not expected to be reused in the cache.
4343 The code generator may select special instructions to save cache bandwidth,
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004344 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +00004345
Chris Lattner2b7d3202002-05-06 03:03:22 +00004346<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004347<p>The location of memory pointed to is loaded. If the value being loaded is of
4348 scalar type then the number of bytes read does not exceed the minimum number
4349 of bytes needed to hold all bits of the type. For example, loading an
4350 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4351 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4352 is undefined if the value was not originally written using a store of the
4353 same type.</p>
4354
Chris Lattner2b7d3202002-05-06 03:03:22 +00004355<h5>Examples:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004356<pre>
4357 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4358 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerca86e162006-12-31 07:07:53 +00004359 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00004360</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004361
Misha Brukman9d0919f2003-11-08 01:05:38 +00004362</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004363
Chris Lattner2b7d3202002-05-06 03:03:22 +00004364<!-- _______________________________________________________________________ -->
Chris Lattner261efe92003-11-25 01:02:51 +00004365<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4366Instruction</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004367
Reid Spencer035ab572006-11-09 21:18:01 +00004368<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004369
Chris Lattner2b7d3202002-05-06 03:03:22 +00004370<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004371<pre>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004372 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>
4373 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 +00004374</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004375
Chris Lattner2b7d3202002-05-06 03:03:22 +00004376<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004377<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004378
Chris Lattner2b7d3202002-05-06 03:03:22 +00004379<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004380<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4381 and an address at which to store it. The type of the
4382 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4383 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00004384 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
4385 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4386 order of execution of this <tt>store</tt> with other <a
4387 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004388
4389<p>The optional constant "align" argument specifies the alignment of the
4390 operation (that is, the alignment of the memory address). A value of 0 or an
4391 omitted "align" argument means that the operation has the preferential
4392 alignment for the target. It is the responsibility of the code emitter to
4393 ensure that the alignment information is correct. Overestimating the
4394 alignment results in an undefined behavior. Underestimating the alignment may
4395 produce less efficient code. An alignment of 1 is always safe.</p>
4396
David Greene8939b0d2010-02-16 20:50:18 +00004397<p>The optional !nontemporal metadata must reference a single metatadata
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004398 name &lt;index&gt; corresponding to a metadata node with one i32 entry of
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004399 value 1. The existence of the !nontemporal metatadata on the
David Greene8939b0d2010-02-16 20:50:18 +00004400 instruction tells the optimizer and code generator that this load is
4401 not expected to be reused in the cache. The code generator may
4402 select special instructions to save cache bandwidth, such as the
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00004403 MOVNT instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +00004404
4405
Chris Lattner261efe92003-11-25 01:02:51 +00004406<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004407<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4408 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4409 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4410 does not exceed the minimum number of bytes needed to hold all bits of the
4411 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4412 writing a value of a type like <tt>i20</tt> with a size that is not an
4413 integral number of bytes, it is unspecified what happens to the extra bits
4414 that do not belong to the type, but they will typically be overwritten.</p>
4415
Chris Lattner2b7d3202002-05-06 03:03:22 +00004416<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004417<pre>
4418 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8c6c72d2007-10-22 05:10:05 +00004419 store i32 3, i32* %ptr <i>; yields {void}</i>
4420 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +00004421</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004422
Reid Spencer47ce1792006-11-09 21:15:49 +00004423</div>
4424
Chris Lattner2b7d3202002-05-06 03:03:22 +00004425<!-- _______________________________________________________________________ -->
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004426<div class="doc_subsubsection">
4427 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4428</div>
4429
Misha Brukman9d0919f2003-11-08 01:05:38 +00004430<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004431
Chris Lattner7faa8832002-04-14 06:13:44 +00004432<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004433<pre>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004434 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmandd8004d2009-07-27 21:53:46 +00004435 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004436</pre>
4437
Chris Lattner7faa8832002-04-14 06:13:44 +00004438<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004439<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004440 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4441 It performs address calculation only and does not access memory.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004442
Chris Lattner7faa8832002-04-14 06:13:44 +00004443<h5>Arguments:</h5>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004444<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnerc8eef442009-07-29 06:44:13 +00004445 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004446 elements of the aggregate object are indexed. The interpretation of each
4447 index is dependent on the type being indexed into. The first index always
4448 indexes the pointer value given as the first argument, the second index
4449 indexes a value of the type pointed to (not necessarily the value directly
4450 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004451 indexed into must be a pointer value, subsequent types can be arrays,
Chris Lattner61c70e92010-08-28 04:09:24 +00004452 vectors, and structs. Note that subsequent types being indexed into
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004453 can never be pointers, since that would require loading the pointer before
4454 continuing calculation.</p>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004455
4456<p>The type of each index argument depends on the type it is indexing into.
Chris Lattner61c70e92010-08-28 04:09:24 +00004457 When indexing into a (optionally packed) structure, only <tt>i32</tt>
Chris Lattnerfdfeb692010-02-12 20:49:41 +00004458 integer <b>constants</b> are allowed. When indexing into an array, pointer
4459 or vector, integers of any width are allowed, and they are not required to be
Chris Lattnerc8eef442009-07-29 06:44:13 +00004460 constant.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004461
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004462<p>For example, let's consider a C code fragment and how it gets compiled to
4463 LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004464
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004465<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004466struct RT {
4467 char A;
Chris Lattnercabc8462007-05-29 15:43:56 +00004468 int B[10][20];
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004469 char C;
4470};
4471struct ST {
Chris Lattnercabc8462007-05-29 15:43:56 +00004472 int X;
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004473 double Y;
4474 struct RT Z;
4475};
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004476
Chris Lattnercabc8462007-05-29 15:43:56 +00004477int *foo(struct ST *s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004478 return &amp;s[1].Z.B[5][13];
4479}
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004480</pre>
4481
Misha Brukman9d0919f2003-11-08 01:05:38 +00004482<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004483
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00004484<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +00004485%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4486%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004487
Dan Gohman4df605b2009-07-25 02:23:48 +00004488define i32* @foo(%ST* %s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +00004489entry:
4490 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4491 ret i32* %reg
4492}
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004493</pre>
4494
Chris Lattner7faa8832002-04-14 06:13:44 +00004495<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004496<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004497 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4498 }</tt>' type, a structure. The second index indexes into the third element
4499 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4500 i8 }</tt>' type, another structure. The third index indexes into the second
4501 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4502 array. The two dimensions of the array are subscripted into, yielding an
4503 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4504 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004505
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004506<p>Note that it is perfectly legal to index partially through a structure,
4507 returning a pointer to an inner element. Because of this, the LLVM code for
4508 the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004509
4510<pre>
Dan Gohman4df605b2009-07-25 02:23:48 +00004511 define i32* @foo(%ST* %s) {
Reid Spencerca86e162006-12-31 07:07:53 +00004512 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
Jeff Cohen6f1cc772007-04-22 01:17:39 +00004513 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4514 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
Reid Spencerca86e162006-12-31 07:07:53 +00004515 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4516 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4517 ret i32* %t5
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004518 }
Chris Lattner6536cfe2002-05-06 22:08:29 +00004519</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +00004520
Dan Gohmandd8004d2009-07-27 21:53:46 +00004521<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman27ef9972010-04-23 15:23:32 +00004522 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
4523 base pointer is not an <i>in bounds</i> address of an allocated object,
4524 or if any of the addresses that would be formed by successive addition of
4525 the offsets implied by the indices to the base address with infinitely
4526 precise arithmetic are not an <i>in bounds</i> address of that allocated
4527 object. The <i>in bounds</i> addresses for an allocated object are all
4528 the addresses that point into the object, plus the address one byte past
4529 the end.</p>
Dan Gohmandd8004d2009-07-27 21:53:46 +00004530
4531<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4532 the base address with silently-wrapping two's complement arithmetic, and
4533 the result value of the <tt>getelementptr</tt> may be outside the object
4534 pointed to by the base pointer. The result value may not necessarily be
4535 used to access memory though, even if it happens to point into allocated
4536 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4537 section for more information.</p>
4538
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004539<p>The getelementptr instruction is often confusing. For some more insight into
4540 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner884a9702006-08-15 00:45:58 +00004541
Chris Lattner7faa8832002-04-14 06:13:44 +00004542<h5>Example:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004543<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004544 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004545 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4546 <i>; yields i8*:vptr</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +00004547 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +00004548 <i>; yields i8*:eptr</i>
4549 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta9f805c22009-04-25 07:27:44 +00004550 <i>; yields i32*:iptr</i>
Sanjiv Gupta16ffa802009-04-24 16:38:13 +00004551 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004552</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004553
Chris Lattnerf74d5c72004-04-05 01:30:49 +00004554</div>
Reid Spencer47ce1792006-11-09 21:15:49 +00004555
Chris Lattner00950542001-06-06 20:29:01 +00004556<!-- ======================================================================= -->
Reid Spencer2fd21e62006-11-08 01:18:52 +00004557<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman9d0919f2003-11-08 01:05:38 +00004558</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004559
Misha Brukman9d0919f2003-11-08 01:05:38 +00004560<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004561
Reid Spencer2fd21e62006-11-08 01:18:52 +00004562<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004563 which all take a single operand and a type. They perform various bit
4564 conversions on the operand.</p>
4565
Misha Brukman9d0919f2003-11-08 01:05:38 +00004566</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004567
Chris Lattner6536cfe2002-05-06 22:08:29 +00004568<!-- _______________________________________________________________________ -->
Chris Lattnercc37aae2004-03-12 05:50:16 +00004569<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004570 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4571</div>
4572<div class="doc_text">
4573
4574<h5>Syntax:</h5>
4575<pre>
4576 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4577</pre>
4578
4579<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004580<p>The '<tt>trunc</tt>' instruction truncates its operand to the
4581 type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004582
4583<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +00004584<p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
4585 Both types must be of <a href="#t_integer">integer</a> types, or vectors
4586 of the same number of integers.
4587 The bit size of the <tt>value</tt> must be larger than
4588 the bit size of the destination type, <tt>ty2</tt>.
4589 Equal sized types are not allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004590
4591<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004592<p>The '<tt>trunc</tt>' instruction truncates the high order bits
4593 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4594 source size must be larger than the destination size, <tt>trunc</tt> cannot
4595 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004596
4597<h5>Example:</h5>
4598<pre>
Nadav Rotem8c20ec52011-02-24 21:01:34 +00004599 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4600 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4601 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
4602 %W = trunc &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i8&gt; <i>; yields &lt;i8 8, i8 7&gt;</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004603</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004604
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004605</div>
4606
4607<!-- _______________________________________________________________________ -->
4608<div class="doc_subsubsection">
4609 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4610</div>
4611<div class="doc_text">
4612
4613<h5>Syntax:</h5>
4614<pre>
4615 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4616</pre>
4617
4618<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004619<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004620 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004621
4622
4623<h5>Arguments:</h5>
Nadav Rotemed9b9342011-02-20 12:37:50 +00004624<p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
4625 Both types must be of <a href="#t_integer">integer</a> types, or vectors
4626 of the same number of integers.
4627 The bit size of the <tt>value</tt> must be smaller than
4628 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004629 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004630
4631<h5>Semantics:</h5>
4632<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004633 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004634
Reid Spencerb5929522007-01-12 15:46:11 +00004635<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004636
4637<h5>Example:</h5>
4638<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004639 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00004640 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Nadav Rotemed9b9342011-02-20 12:37:50 +00004641 %Z = zext &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i32&gt; <i>; yields &lt;i32 8, i32 7&gt;</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004642</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004643
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004644</div>
4645
4646<!-- _______________________________________________________________________ -->
4647<div class="doc_subsubsection">
4648 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4649</div>
4650<div class="doc_text">
4651
4652<h5>Syntax:</h5>
4653<pre>
4654 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4655</pre>
4656
4657<h5>Overview:</h5>
4658<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4659
4660<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +00004661<p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
4662 Both types must be of <a href="#t_integer">integer</a> types, or vectors
4663 of the same number of integers.
4664 The bit size of the <tt>value</tt> must be smaller than
4665 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004666 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004667
4668<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004669<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4670 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4671 of the type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004672
Reid Spencerc78f3372007-01-12 03:35:51 +00004673<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004674
4675<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004676<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004677 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencerc78f3372007-01-12 03:35:51 +00004678 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Nadav Rotem8c20ec52011-02-24 21:01:34 +00004679 %Z = sext &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i32&gt; <i>; yields &lt;i32 8, i32 7&gt;</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004680</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004681
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004682</div>
4683
4684<!-- _______________________________________________________________________ -->
4685<div class="doc_subsubsection">
Reid Spencer3fa91b02006-11-09 21:48:10 +00004686 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4687</div>
4688
4689<div class="doc_text">
4690
4691<h5>Syntax:</h5>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004692<pre>
4693 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4694</pre>
4695
4696<h5>Overview:</h5>
4697<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004698 <tt>ty2</tt>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004699
4700<h5>Arguments:</h5>
4701<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004702 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4703 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004704 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004705 <i>no-op cast</i>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004706
4707<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004708<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004709 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004710 <a href="#t_floating">floating point</a> type. If the value cannot fit
4711 within the destination type, <tt>ty2</tt>, then the results are
4712 undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +00004713
4714<h5>Example:</h5>
4715<pre>
4716 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4717 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4718</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004719
Reid Spencer3fa91b02006-11-09 21:48:10 +00004720</div>
4721
4722<!-- _______________________________________________________________________ -->
4723<div class="doc_subsubsection">
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004724 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4725</div>
4726<div class="doc_text">
4727
4728<h5>Syntax:</h5>
4729<pre>
4730 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4731</pre>
4732
4733<h5>Overview:</h5>
4734<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004735 floating point value.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004736
4737<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004738<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004739 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4740 a <a href="#t_floating">floating point</a> type to cast it to. The source
4741 type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004742
4743<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +00004744<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004745 <a href="#t_floating">floating point</a> type to a larger
4746 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4747 used to make a <i>no-op cast</i> because it always changes bits. Use
4748 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004749
4750<h5>Example:</h5>
4751<pre>
4752 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4753 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4754</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004755
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004756</div>
4757
4758<!-- _______________________________________________________________________ -->
4759<div class="doc_subsubsection">
Reid Spencer24d6da52007-01-21 00:29:26 +00004760 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004761</div>
4762<div class="doc_text">
4763
4764<h5>Syntax:</h5>
4765<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +00004766 &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 +00004767</pre>
4768
4769<h5>Overview:</h5>
Reid Spencer1539a1c2007-07-31 14:40:14 +00004770<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004771 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004772
4773<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004774<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4775 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4776 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4777 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4778 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004779
4780<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004781<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004782 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4783 towards zero) unsigned integer value. If the value cannot fit
4784 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004785
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004786<h5>Example:</h5>
4787<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +00004788 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner88519042007-09-22 03:17:52 +00004789 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004790 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004791</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004792
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004793</div>
4794
4795<!-- _______________________________________________________________________ -->
4796<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +00004797 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004798</div>
4799<div class="doc_text">
4800
4801<h5>Syntax:</h5>
4802<pre>
Reid Spencerd4448792006-11-09 23:03:26 +00004803 &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 +00004804</pre>
4805
4806<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004807<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004808 <a href="#t_floating">floating point</a> <tt>value</tt> to
4809 type <tt>ty2</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004810
Chris Lattner6536cfe2002-05-06 22:08:29 +00004811<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004812<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4813 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4814 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4815 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4816 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004817
Chris Lattner6536cfe2002-05-06 22:08:29 +00004818<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00004819<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004820 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4821 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4822 the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004823
Chris Lattner33ba0d92001-07-09 00:26:23 +00004824<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00004825<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +00004826 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner88519042007-09-22 03:17:52 +00004827 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004828 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004829</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004830
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004831</div>
4832
4833<!-- _______________________________________________________________________ -->
4834<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +00004835 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004836</div>
4837<div class="doc_text">
4838
4839<h5>Syntax:</h5>
4840<pre>
Reid Spencerd4448792006-11-09 23:03:26 +00004841 &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 +00004842</pre>
4843
4844<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +00004845<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004846 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004847
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004848<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +00004849<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004850 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4851 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4852 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4853 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004854
4855<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +00004856<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004857 integer quantity and converts it to the corresponding floating point
4858 value. If the value cannot fit in the floating point value, the results are
4859 undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004860
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004861<h5>Example:</h5>
4862<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004863 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +00004864 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004865</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004866
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004867</div>
4868
4869<!-- _______________________________________________________________________ -->
4870<div class="doc_subsubsection">
Reid Spencerd4448792006-11-09 23:03:26 +00004871 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004872</div>
4873<div class="doc_text">
4874
4875<h5>Syntax:</h5>
4876<pre>
Reid Spencerd4448792006-11-09 23:03:26 +00004877 &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 +00004878</pre>
4879
4880<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004881<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4882 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004883
4884<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +00004885<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004886 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4887 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4888 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4889 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004890
4891<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004892<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4893 quantity and converts it to the corresponding floating point value. If the
4894 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004895
4896<h5>Example:</h5>
4897<pre>
Reid Spencerca86e162006-12-31 07:07:53 +00004898 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +00004899 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004900</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004901
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004902</div>
4903
4904<!-- _______________________________________________________________________ -->
4905<div class="doc_subsubsection">
Reid Spencer72679252006-11-11 21:00:47 +00004906 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4907</div>
4908<div class="doc_text">
4909
4910<h5>Syntax:</h5>
4911<pre>
4912 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4913</pre>
4914
4915<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004916<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4917 the integer type <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004918
4919<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004920<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4921 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4922 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004923
4924<h5>Semantics:</h5>
4925<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004926 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4927 truncating or zero extending that value to the size of the integer type. If
4928 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4929 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4930 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4931 change.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004932
4933<h5>Example:</h5>
4934<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +00004935 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4936 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +00004937</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004938
Reid Spencer72679252006-11-11 21:00:47 +00004939</div>
4940
4941<!-- _______________________________________________________________________ -->
4942<div class="doc_subsubsection">
4943 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4944</div>
4945<div class="doc_text">
4946
4947<h5>Syntax:</h5>
4948<pre>
4949 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4950</pre>
4951
4952<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004953<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4954 pointer type, <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004955
4956<h5>Arguments:</h5>
Duncan Sands8036ca42007-03-30 12:22:09 +00004957<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004958 value to cast, and a type to cast it to, which must be a
4959 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +00004960
4961<h5>Semantics:</h5>
4962<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004963 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4964 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4965 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4966 than the size of a pointer then a zero extension is done. If they are the
4967 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencer72679252006-11-11 21:00:47 +00004968
4969<h5>Example:</h5>
4970<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +00004971 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greif6a292012009-10-28 09:21:30 +00004972 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4973 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +00004974</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004975
Reid Spencer72679252006-11-11 21:00:47 +00004976</div>
4977
4978<!-- _______________________________________________________________________ -->
4979<div class="doc_subsubsection">
Reid Spencer5c0ef472006-11-11 23:08:07 +00004980 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004981</div>
4982<div class="doc_text">
4983
4984<h5>Syntax:</h5>
4985<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +00004986 &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 +00004987</pre>
4988
4989<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +00004990<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004991 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00004992
4993<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00004994<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4995 non-aggregate first class value, and a type to cast it to, which must also be
4996 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4997 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4998 identical. If the source type is a pointer, the destination type must also be
4999 a pointer. This instruction supports bitwise conversion of vectors to
5000 integers and to vectors of other types (as long as they have the same
5001 size).</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00005002
5003<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +00005004<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005005 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5006 this conversion. The conversion is done as if the <tt>value</tt> had been
5007 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
5008 be converted to other pointer types with this instruction. To convert
5009 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5010 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +00005011
5012<h5>Example:</h5>
5013<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005014 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerca86e162006-12-31 07:07:53 +00005015 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005016 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +00005017</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005018
Misha Brukman9d0919f2003-11-08 01:05:38 +00005019</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005020
Reid Spencer2fd21e62006-11-08 01:18:52 +00005021<!-- ======================================================================= -->
5022<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005023
Reid Spencer2fd21e62006-11-08 01:18:52 +00005024<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005025
5026<p>The instructions in this category are the "miscellaneous" instructions, which
5027 defy better classification.</p>
5028
Reid Spencer2fd21e62006-11-08 01:18:52 +00005029</div>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005030
5031<!-- _______________________________________________________________________ -->
5032<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5033</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005034
Reid Spencerf3a70a62006-11-18 21:50:54 +00005035<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005036
Reid Spencerf3a70a62006-11-18 21:50:54 +00005037<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005038<pre>
5039 &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 +00005040</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005041
Reid Spencerf3a70a62006-11-18 21:50:54 +00005042<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005043<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5044 boolean values based on comparison of its two integer, integer vector, or
5045 pointer operands.</p>
5046
Reid Spencerf3a70a62006-11-18 21:50:54 +00005047<h5>Arguments:</h5>
5048<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005049 the condition code indicating the kind of comparison to perform. It is not a
5050 value, just a keyword. The possible condition code are:</p>
5051
Reid Spencerf3a70a62006-11-18 21:50:54 +00005052<ol>
5053 <li><tt>eq</tt>: equal</li>
5054 <li><tt>ne</tt>: not equal </li>
5055 <li><tt>ugt</tt>: unsigned greater than</li>
5056 <li><tt>uge</tt>: unsigned greater or equal</li>
5057 <li><tt>ult</tt>: unsigned less than</li>
5058 <li><tt>ule</tt>: unsigned less or equal</li>
5059 <li><tt>sgt</tt>: signed greater than</li>
5060 <li><tt>sge</tt>: signed greater or equal</li>
5061 <li><tt>slt</tt>: signed less than</li>
5062 <li><tt>sle</tt>: signed less or equal</li>
5063</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005064
Chris Lattner3b19d652007-01-15 01:54:13 +00005065<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005066 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5067 typed. They must also be identical types.</p>
5068
Reid Spencerf3a70a62006-11-18 21:50:54 +00005069<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005070<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5071 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewyckyec38da42009-09-27 00:45:11 +00005072 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005073 result, as follows:</p>
5074
Reid Spencerf3a70a62006-11-18 21:50:54 +00005075<ol>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005076 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005077 <tt>false</tt> otherwise. No sign interpretation is necessary or
5078 performed.</li>
5079
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005080 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005081 <tt>false</tt> otherwise. No sign interpretation is necessary or
5082 performed.</li>
5083
Reid Spencerf3a70a62006-11-18 21:50:54 +00005084 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005085 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5086
Reid Spencerf3a70a62006-11-18 21:50:54 +00005087 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005088 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5089 to <tt>op2</tt>.</li>
5090
Reid Spencerf3a70a62006-11-18 21:50:54 +00005091 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005092 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5093
Reid Spencerf3a70a62006-11-18 21:50:54 +00005094 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005095 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5096
Reid Spencerf3a70a62006-11-18 21:50:54 +00005097 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005098 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5099
Reid Spencerf3a70a62006-11-18 21:50:54 +00005100 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005101 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5102 to <tt>op2</tt>.</li>
5103
Reid Spencerf3a70a62006-11-18 21:50:54 +00005104 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005105 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5106
Reid Spencerf3a70a62006-11-18 21:50:54 +00005107 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005108 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005109</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005110
Reid Spencerf3a70a62006-11-18 21:50:54 +00005111<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005112 values are compared as if they were integers.</p>
5113
5114<p>If the operands are integer vectors, then they are compared element by
5115 element. The result is an <tt>i1</tt> vector with the same number of elements
5116 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005117
5118<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005119<pre>
5120 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerca86e162006-12-31 07:07:53 +00005121 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
5122 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
5123 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5124 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5125 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005126</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +00005127
5128<p>Note that the code generator does not yet support vector types with
5129 the <tt>icmp</tt> instruction.</p>
5130
Reid Spencerf3a70a62006-11-18 21:50:54 +00005131</div>
5132
5133<!-- _______________________________________________________________________ -->
5134<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5135</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005136
Reid Spencerf3a70a62006-11-18 21:50:54 +00005137<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005138
Reid Spencerf3a70a62006-11-18 21:50:54 +00005139<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005140<pre>
5141 &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 +00005142</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005143
Reid Spencerf3a70a62006-11-18 21:50:54 +00005144<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005145<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5146 values based on comparison of its operands.</p>
5147
5148<p>If the operands are floating point scalars, then the result type is a boolean
Nick Lewyckyec38da42009-09-27 00:45:11 +00005149(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005150
5151<p>If the operands are floating point vectors, then the result type is a vector
5152 of boolean with the same number of elements as the operands being
5153 compared.</p>
5154
Reid Spencerf3a70a62006-11-18 21:50:54 +00005155<h5>Arguments:</h5>
5156<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005157 the condition code indicating the kind of comparison to perform. It is not a
5158 value, just a keyword. The possible condition code are:</p>
5159
Reid Spencerf3a70a62006-11-18 21:50:54 +00005160<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +00005161 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005162 <li><tt>oeq</tt>: ordered and equal</li>
5163 <li><tt>ogt</tt>: ordered and greater than </li>
5164 <li><tt>oge</tt>: ordered and greater than or equal</li>
5165 <li><tt>olt</tt>: ordered and less than </li>
5166 <li><tt>ole</tt>: ordered and less than or equal</li>
5167 <li><tt>one</tt>: ordered and not equal</li>
5168 <li><tt>ord</tt>: ordered (no nans)</li>
5169 <li><tt>ueq</tt>: unordered or equal</li>
5170 <li><tt>ugt</tt>: unordered or greater than </li>
5171 <li><tt>uge</tt>: unordered or greater than or equal</li>
5172 <li><tt>ult</tt>: unordered or less than </li>
5173 <li><tt>ule</tt>: unordered or less than or equal</li>
5174 <li><tt>une</tt>: unordered or not equal</li>
5175 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +00005176 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005177</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005178
Jeff Cohenb627eab2007-04-29 01:07:00 +00005179<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005180 <i>unordered</i> means that either operand may be a QNAN.</p>
5181
5182<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5183 a <a href="#t_floating">floating point</a> type or
5184 a <a href="#t_vector">vector</a> of floating point type. They must have
5185 identical types.</p>
5186
Reid Spencerf3a70a62006-11-18 21:50:54 +00005187<h5>Semantics:</h5>
Gabor Greiffb224a22008-08-07 21:46:00 +00005188<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005189 according to the condition code given as <tt>cond</tt>. If the operands are
5190 vectors, then the vectors are compared element by element. Each comparison
Nick Lewyckyec38da42009-09-27 00:45:11 +00005191 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005192 follows:</p>
5193
Reid Spencerf3a70a62006-11-18 21:50:54 +00005194<ol>
5195 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005196
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005197 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005198 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5199
Reid Spencerb7f26282006-11-19 03:00:14 +00005200 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00005201 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005202
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005203 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005204 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5205
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005206 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005207 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5208
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005209 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005210 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5211
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005212 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005213 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5214
Reid Spencerb7f26282006-11-19 03:00:14 +00005215 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005216
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005217 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005218 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5219
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005220 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005221 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5222
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005223 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005224 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5225
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005226 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005227 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5228
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005229 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005230 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5231
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005232 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005233 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5234
Reid Spencerb7f26282006-11-19 03:00:14 +00005235 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005236
Reid Spencerf3a70a62006-11-18 21:50:54 +00005237 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5238</ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005239
5240<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005241<pre>
5242 &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanf72fb672008-09-09 01:02:47 +00005243 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5244 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5245 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +00005246</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +00005247
5248<p>Note that the code generator does not yet support vector types with
5249 the <tt>fcmp</tt> instruction.</p>
5250
Reid Spencerf3a70a62006-11-18 21:50:54 +00005251</div>
5252
Reid Spencer2fd21e62006-11-08 01:18:52 +00005253<!-- _______________________________________________________________________ -->
Nate Begemanac80ade2008-05-12 19:01:56 +00005254<div class="doc_subsubsection">
Chris Lattner5568e942008-05-20 20:48:21 +00005255 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5256</div>
5257
Reid Spencer2fd21e62006-11-08 01:18:52 +00005258<div class="doc_text">
Chris Lattner5568e942008-05-20 20:48:21 +00005259
Reid Spencer2fd21e62006-11-08 01:18:52 +00005260<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005261<pre>
5262 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
5263</pre>
Chris Lattner5568e942008-05-20 20:48:21 +00005264
Reid Spencer2fd21e62006-11-08 01:18:52 +00005265<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005266<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
5267 SSA graph representing the function.</p>
5268
Reid Spencer2fd21e62006-11-08 01:18:52 +00005269<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005270<p>The type of the incoming values is specified with the first type field. After
5271 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5272 one pair for each predecessor basic block of the current block. Only values
5273 of <a href="#t_firstclass">first class</a> type may be used as the value
5274 arguments to the PHI node. Only labels may be used as the label
5275 arguments.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00005276
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005277<p>There must be no non-phi instructions between the start of a basic block and
5278 the PHI instructions: i.e. PHI instructions must be first in a basic
5279 block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00005280
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005281<p>For the purposes of the SSA form, the use of each incoming value is deemed to
5282 occur on the edge from the corresponding predecessor block to the current
5283 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5284 value on the same edge).</p>
Jay Foadd2449092009-06-03 10:20:10 +00005285
Reid Spencer2fd21e62006-11-08 01:18:52 +00005286<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005287<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005288 specified by the pair corresponding to the predecessor basic block that
5289 executed just prior to the current block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +00005290
Reid Spencer2fd21e62006-11-08 01:18:52 +00005291<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +00005292<pre>
5293Loop: ; Infinite loop that counts from 0 on up...
5294 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5295 %nextindvar = add i32 %indvar, 1
5296 br label %Loop
5297</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005298
Reid Spencer2fd21e62006-11-08 01:18:52 +00005299</div>
5300
Chris Lattnercc37aae2004-03-12 05:50:16 +00005301<!-- _______________________________________________________________________ -->
5302<div class="doc_subsubsection">
5303 <a name="i_select">'<tt>select</tt>' Instruction</a>
5304</div>
5305
5306<div class="doc_text">
5307
5308<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005309<pre>
Dan Gohmanf72fb672008-09-09 01:02:47 +00005310 &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>
5311
Dan Gohman0e451ce2008-10-14 16:51:45 +00005312 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnercc37aae2004-03-12 05:50:16 +00005313</pre>
5314
5315<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005316<p>The '<tt>select</tt>' instruction is used to choose one value based on a
5317 condition, without branching.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005318
5319
5320<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005321<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5322 values indicating the condition, and two values of the
5323 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5324 vectors and the condition is a scalar, then entire vectors are selected, not
5325 individual elements.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005326
5327<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005328<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5329 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005330
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005331<p>If the condition is a vector of i1, then the value arguments must be vectors
5332 of the same size, and the selection is done element by element.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005333
5334<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005335<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +00005336 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +00005337</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +00005338
5339<p>Note that the code generator does not yet support conditions
5340 with vector type.</p>
5341
Chris Lattnercc37aae2004-03-12 05:50:16 +00005342</div>
5343
Robert Bocchino05ccd702006-01-15 20:48:27 +00005344<!-- _______________________________________________________________________ -->
5345<div class="doc_subsubsection">
Chris Lattner2bff5242005-05-06 05:47:36 +00005346 <a name="i_call">'<tt>call</tt>' Instruction</a>
5347</div>
5348
Misha Brukman9d0919f2003-11-08 01:05:38 +00005349<div class="doc_text">
Chris Lattner2bff5242005-05-06 05:47:36 +00005350
Chris Lattner00950542001-06-06 20:29:01 +00005351<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00005352<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +00005353 &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 +00005354</pre>
5355
Chris Lattner00950542001-06-06 20:29:01 +00005356<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00005357<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00005358
Chris Lattner00950542001-06-06 20:29:01 +00005359<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00005360<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00005361
Chris Lattner6536cfe2002-05-06 22:08:29 +00005362<ol>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005363 <li>The optional "tail" marker indicates that the callee function does not
5364 access any allocas or varargs in the caller. Note that calls may be
5365 marked "tail" even if they do not occur before
5366 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5367 present, the function call is eligible for tail call optimization,
5368 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Chengdc444e92010-03-08 21:05:02 +00005369 optimized into a jump</a>. The code generator may optimize calls marked
5370 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5371 sibling call optimization</a> when the caller and callee have
5372 matching signatures, or 2) forced tail call optimization when the
5373 following extra requirements are met:
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005374 <ul>
5375 <li>Caller and callee both have the calling
5376 convention <tt>fastcc</tt>.</li>
5377 <li>The call is in tail position (ret immediately follows call and ret
5378 uses value of call or is void).</li>
5379 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohmanfbbee8d2010-03-02 01:08:11 +00005380 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005381 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5382 constraints are met.</a></li>
5383 </ul>
5384 </li>
Devang Patelf642f472008-10-06 18:50:38 +00005385
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005386 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5387 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +00005388 defaults to using C calling conventions. The calling convention of the
5389 call must match the calling convention of the target function, or else the
5390 behavior is undefined.</li>
Devang Patelf642f472008-10-06 18:50:38 +00005391
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005392 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5393 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5394 '<tt>inreg</tt>' attributes are valid here.</li>
5395
5396 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5397 type of the return value. Functions that return no value are marked
5398 <tt><a href="#t_void">void</a></tt>.</li>
5399
5400 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5401 being invoked. The argument types must match the types implied by this
5402 signature. This type can be omitted if the function is not varargs and if
5403 the function type does not return a pointer to a function.</li>
5404
5405 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5406 be invoked. In most cases, this is a direct function invocation, but
5407 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5408 to function value.</li>
5409
5410 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +00005411 signature argument types and parameter attributes. All arguments must be
5412 of <a href="#t_firstclass">first class</a> type. If the function
5413 signature indicates the function accepts a variable number of arguments,
5414 the extra arguments can be specified.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005415
5416 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5417 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5418 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner6536cfe2002-05-06 22:08:29 +00005419</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +00005420
Chris Lattner00950542001-06-06 20:29:01 +00005421<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005422<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5423 a specified function, with its incoming arguments bound to the specified
5424 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5425 function, control flow continues with the instruction after the function
5426 call, and the return value of the function is bound to the result
5427 argument.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +00005428
Chris Lattner00950542001-06-06 20:29:01 +00005429<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +00005430<pre>
Nick Lewyckydb7e3c92007-09-08 13:57:50 +00005431 %retval = call i32 @test(i32 %argc)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005432 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattner772fccf2008-03-21 17:24:17 +00005433 %X = tail call i32 @foo() <i>; yields i32</i>
5434 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5435 call void %foo(i8 97 signext)
Devang Patelc3fc6df2008-03-10 20:49:15 +00005436
5437 %struct.A = type { i32, i8 }
Devang Patelf642f472008-10-06 18:50:38 +00005438 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmanb1e6b962008-10-04 19:00:07 +00005439 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5440 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner85a350f2008-10-08 06:26:11 +00005441 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmancb73d192008-10-07 10:03:45 +00005442 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattner2bff5242005-05-06 05:47:36 +00005443</pre>
5444
Dale Johannesen07de8d12009-09-24 18:38:21 +00005445<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen9f8380b2009-09-25 17:04:42 +00005446standard C99 library as being the C99 library functions, and may perform
5447optimizations or generate code for them under that assumption. This is
5448something we'd like to change in the future to provide better support for
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00005449freestanding environments and non-C-based languages.</p>
Dale Johannesen07de8d12009-09-24 18:38:21 +00005450
Misha Brukman9d0919f2003-11-08 01:05:38 +00005451</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005452
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005453<!-- _______________________________________________________________________ -->
Chris Lattnere19d7a72004-09-27 21:51:25 +00005454<div class="doc_subsubsection">
Chris Lattnerfb6977d2006-01-13 23:26:01 +00005455 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005456</div>
5457
Misha Brukman9d0919f2003-11-08 01:05:38 +00005458<div class="doc_text">
Chris Lattnere19d7a72004-09-27 21:51:25 +00005459
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005460<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005461<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005462 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattnere19d7a72004-09-27 21:51:25 +00005463</pre>
5464
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005465<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005466<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005467 the "variable argument" area of a function call. It is used to implement the
5468 <tt>va_arg</tt> macro in C.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005469
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005470<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005471<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5472 argument. It returns a value of the specified argument type and increments
5473 the <tt>va_list</tt> to point to the next argument. The actual type
5474 of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005475
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005476<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005477<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5478 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5479 to the next argument. For more information, see the variable argument
5480 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005481
5482<p>It is legal for this instruction to be called in a function which does not
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005483 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5484 function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005485
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005486<p><tt>va_arg</tt> is an LLVM instruction instead of
5487 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5488 argument.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005489
Chris Lattner8d1a81d2003-10-18 05:51:36 +00005490<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +00005491<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5492
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005493<p>Note that the code generator does not yet fully support va_arg on many
5494 targets. Also, it does not currently support va_arg with aggregate types on
5495 any target.</p>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +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 Lattner261efe92003-11-25 01:02:51 +00005500<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5501<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005502
Misha Brukman9d0919f2003-11-08 01:05:38 +00005503<div class="doc_text">
Chris Lattner33aec9e2004-02-12 17:01:32 +00005504
5505<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005506 well known names and semantics and are required to follow certain
5507 restrictions. Overall, these intrinsics represent an extension mechanism for
5508 the LLVM language that does not require changing all of the transformations
5509 in LLVM when adding to the language (or the bitcode reader/writer, the
5510 parser, etc...).</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005511
John Criswellfc6b8952005-05-16 16:17:45 +00005512<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005513 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5514 begin with this prefix. Intrinsic functions must always be external
5515 functions: you cannot define the body of intrinsic functions. Intrinsic
5516 functions may only be used in call or invoke instructions: it is illegal to
5517 take the address of an intrinsic function. Additionally, because intrinsic
5518 functions are part of the LLVM language, it is required if any are added that
5519 they be documented here.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005520
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005521<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5522 family of functions that perform the same operation but on different data
5523 types. Because LLVM can represent over 8 million different integer types,
5524 overloading is used commonly to allow an intrinsic function to operate on any
5525 integer type. One or more of the argument types or the result type can be
5526 overloaded to accept any integer type. Argument types may also be defined as
5527 exactly matching a previous argument's type or the result type. This allows
5528 an intrinsic function which accepts multiple arguments, but needs all of them
5529 to be of the same type, to only be overloaded with respect to a single
5530 argument or the result.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005531
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005532<p>Overloaded intrinsics will have the names of its overloaded argument types
5533 encoded into its function name, each preceded by a period. Only those types
5534 which are overloaded result in a name suffix. Arguments whose type is matched
5535 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5536 can take an integer of any width and returns an integer of exactly the same
5537 integer width. This leads to a family of functions such as
5538 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5539 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5540 suffix is required. Because the argument's type is matched against the return
5541 type, it does not require its own name suffix.</p>
Reid Spencer409e28f2007-04-01 08:04:23 +00005542
Eric Christopher6c7e8a02009-12-05 02:46:03 +00005543<p>To learn how to add an intrinsic function, please see the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005544 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00005545
Misha Brukman9d0919f2003-11-08 01:05:38 +00005546</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005547
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005548<!-- ======================================================================= -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005549<div class="doc_subsection">
5550 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5551</div>
5552
Misha Brukman9d0919f2003-11-08 01:05:38 +00005553<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00005554
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005555<p>Variable argument support is defined in LLVM with
5556 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5557 intrinsic functions. These functions are related to the similarly named
5558 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005559
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005560<p>All of these functions operate on arguments that use a target-specific value
5561 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5562 not define what this type is, so all transformations should be prepared to
5563 handle these functions regardless of the type used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005564
Chris Lattner374ab302006-05-15 17:26:46 +00005565<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005566 instruction and the variable argument handling intrinsic functions are
5567 used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005568
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00005569<pre class="doc_code">
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005570define i32 @test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +00005571 ; Initialize variable argument processing
Jeff Cohenb627eab2007-04-29 01:07:00 +00005572 %ap = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +00005573 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005574 call void @llvm.va_start(i8* %ap2)
Chris Lattner33aec9e2004-02-12 17:01:32 +00005575
5576 ; Read a single integer argument
Jeff Cohenb627eab2007-04-29 01:07:00 +00005577 %tmp = va_arg i8** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +00005578
5579 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohenb627eab2007-04-29 01:07:00 +00005580 %aq = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +00005581 %aq2 = bitcast i8** %aq to i8*
Jeff Cohenb627eab2007-04-29 01:07:00 +00005582 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005583 call void @llvm.va_end(i8* %aq2)
Chris Lattner33aec9e2004-02-12 17:01:32 +00005584
5585 ; Stop processing of arguments.
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005586 call void @llvm.va_end(i8* %ap2)
Reid Spencerca86e162006-12-31 07:07:53 +00005587 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +00005588}
Anton Korobeynikov5d522f32007-03-21 23:58:04 +00005589
5590declare void @llvm.va_start(i8*)
5591declare void @llvm.va_copy(i8*, i8*)
5592declare void @llvm.va_end(i8*)
Chris Lattner33aec9e2004-02-12 17:01:32 +00005593</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +00005594
Bill Wendling2f7a8b02007-05-29 09:04:49 +00005595</div>
5596
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005597<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005598<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005599 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00005600</div>
5601
5602
Misha Brukman9d0919f2003-11-08 01:05:38 +00005603<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005604
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005605<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005606<pre>
5607 declare void %llvm.va_start(i8* &lt;arglist&gt;)
5608</pre>
5609
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005610<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005611<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
5612 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005613
5614<h5>Arguments:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +00005615<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005616
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005617<h5>Semantics:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +00005618<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005619 macro available in C. In a target-dependent way, it initializes
5620 the <tt>va_list</tt> element to which the argument points, so that the next
5621 call to <tt>va_arg</tt> will produce the first variable argument passed to
5622 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5623 need to know the last argument of the function as the compiler can figure
5624 that out.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005625
Misha Brukman9d0919f2003-11-08 01:05:38 +00005626</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005627
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005628<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005629<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005630 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00005631</div>
5632
Misha Brukman9d0919f2003-11-08 01:05:38 +00005633<div class="doc_text">
Chris Lattnerb75137d2007-01-08 07:55:15 +00005634
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005635<h5>Syntax:</h5>
5636<pre>
5637 declare void @llvm.va_end(i8* &lt;arglist&gt;)
5638</pre>
5639
5640<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005641<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005642 which has been initialized previously
5643 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5644 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +00005645
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005646<h5>Arguments:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005647<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +00005648
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005649<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +00005650<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005651 macro available in C. In a target-dependent way, it destroys
5652 the <tt>va_list</tt> element to which the argument points. Calls
5653 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5654 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5655 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +00005656
Misha Brukman9d0919f2003-11-08 01:05:38 +00005657</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005658
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005659<!-- _______________________________________________________________________ -->
Chris Lattner8ff75902004-01-06 05:31:32 +00005660<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005661 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00005662</div>
5663
Misha Brukman9d0919f2003-11-08 01:05:38 +00005664<div class="doc_text">
Chris Lattnerd7923912004-05-23 21:06:01 +00005665
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005666<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005667<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +00005668 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +00005669</pre>
5670
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005671<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005672<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005673 from the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005674
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005675<h5>Arguments:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +00005676<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005677 The second argument is a pointer to a <tt>va_list</tt> element to copy
5678 from.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005679
Chris Lattnerd9ad5b32003-05-08 04:57:36 +00005680<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +00005681<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005682 macro available in C. In a target-dependent way, it copies the
5683 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5684 element. This intrinsic is necessary because
5685 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5686 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005687
Misha Brukman9d0919f2003-11-08 01:05:38 +00005688</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00005689
Chris Lattner33aec9e2004-02-12 17:01:32 +00005690<!-- ======================================================================= -->
5691<div class="doc_subsection">
Chris Lattnerd7923912004-05-23 21:06:01 +00005692 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5693</div>
5694
5695<div class="doc_text">
5696
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005697<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattnerd3eda892008-08-05 18:29:16 +00005698Collection</a> (GC) requires the implementation and generation of these
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005699intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5700roots on the stack</a>, as well as garbage collector implementations that
5701require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5702barriers. Front-ends for type-safe garbage collected languages should generate
5703these intrinsics to make use of the LLVM garbage collectors. For more details,
5704see <a href="GarbageCollection.html">Accurate Garbage Collection with
5705LLVM</a>.</p>
Christopher Lamb303dae92007-12-17 01:00:21 +00005706
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005707<p>The garbage collection intrinsics only operate on objects in the generic
5708 address space (address space zero).</p>
Christopher Lamb303dae92007-12-17 01:00:21 +00005709
Chris Lattnerd7923912004-05-23 21:06:01 +00005710</div>
5711
5712<!-- _______________________________________________________________________ -->
5713<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005714 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +00005715</div>
5716
5717<div class="doc_text">
5718
5719<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005720<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005721 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +00005722</pre>
5723
5724<h5>Overview:</h5>
John Criswell9e2485c2004-12-10 15:51:16 +00005725<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005726 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005727
5728<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005729<p>The first argument specifies the address of a stack object that contains the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005730 root pointer. The second pointer (which must be either a constant or a
5731 global value address) contains the meta-data to be associated with the
5732 root.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005733
5734<h5>Semantics:</h5>
Chris Lattner05d67092008-04-24 05:59:56 +00005735<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005736 location. At compile-time, the code generator generates information to allow
5737 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5738 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5739 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005740
5741</div>
5742
Chris Lattnerd7923912004-05-23 21:06:01 +00005743<!-- _______________________________________________________________________ -->
5744<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005745 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +00005746</div>
5747
5748<div class="doc_text">
5749
5750<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005751<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005752 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +00005753</pre>
5754
5755<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005756<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005757 locations, allowing garbage collector implementations that require read
5758 barriers.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005759
5760<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +00005761<p>The second argument is the address to read from, which should be an address
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005762 allocated from the garbage collector. The first object is a pointer to the
5763 start of the referenced object, if needed by the language runtime (otherwise
5764 null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005765
5766<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005767<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005768 instruction, but may be replaced with substantially more complex code by the
5769 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5770 may only be used in a function which <a href="#gc">specifies a GC
5771 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005772
5773</div>
5774
Chris Lattnerd7923912004-05-23 21:06:01 +00005775<!-- _______________________________________________________________________ -->
5776<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005777 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
Chris Lattnerd7923912004-05-23 21:06:01 +00005778</div>
5779
5780<div class="doc_text">
5781
5782<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005783<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005784 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +00005785</pre>
5786
5787<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005788<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005789 locations, allowing garbage collector implementations that require write
5790 barriers (such as generational or reference counting collectors).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005791
5792<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +00005793<p>The first argument is the reference to store, the second is the start of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005794 object to store it to, and the third is the address of the field of Obj to
5795 store to. If the runtime does not require a pointer to the object, Obj may
5796 be null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005797
5798<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +00005799<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005800 instruction, but may be replaced with substantially more complex code by the
5801 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5802 may only be used in a function which <a href="#gc">specifies a GC
5803 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +00005804
5805</div>
5806
Chris Lattnerd7923912004-05-23 21:06:01 +00005807<!-- ======================================================================= -->
5808<div class="doc_subsection">
Chris Lattner10610642004-02-14 04:08:35 +00005809 <a name="int_codegen">Code Generator Intrinsics</a>
5810</div>
5811
5812<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005813
5814<p>These intrinsics are provided by LLVM to expose special features that may
5815 only be implemented with code generator support.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005816
5817</div>
5818
5819<!-- _______________________________________________________________________ -->
5820<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005821 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
Chris Lattner10610642004-02-14 04:08:35 +00005822</div>
5823
5824<div class="doc_text">
5825
5826<h5>Syntax:</h5>
5827<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +00005828 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00005829</pre>
5830
5831<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005832<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5833 target-specific value indicating the return address of the current function
5834 or one of its callers.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005835
5836<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005837<p>The argument to this intrinsic indicates which function to return the address
5838 for. Zero indicates the calling function, one indicates its caller, etc.
5839 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005840
5841<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005842<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5843 indicating the return address of the specified call frame, or zero if it
5844 cannot be identified. The value returned by this intrinsic is likely to be
5845 incorrect or 0 for arguments other than zero, so it should only be used for
5846 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005847
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005848<p>Note that calling this intrinsic does not prevent function inlining or other
5849 aggressive transformations, so the value returned may not be that of the
5850 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005851
Chris Lattner10610642004-02-14 04:08:35 +00005852</div>
5853
Chris Lattner10610642004-02-14 04:08:35 +00005854<!-- _______________________________________________________________________ -->
5855<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005856 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
Chris Lattner10610642004-02-14 04:08:35 +00005857</div>
5858
5859<div class="doc_text">
5860
5861<h5>Syntax:</h5>
5862<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005863 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00005864</pre>
5865
5866<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005867<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5868 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005869
5870<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005871<p>The argument to this intrinsic indicates which function to return the frame
5872 pointer for. Zero indicates the calling function, one indicates its caller,
5873 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005874
5875<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005876<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5877 indicating the frame address of the specified call frame, or zero if it
5878 cannot be identified. The value returned by this intrinsic is likely to be
5879 incorrect or 0 for arguments other than zero, so it should only be used for
5880 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005881
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005882<p>Note that calling this intrinsic does not prevent function inlining or other
5883 aggressive transformations, so the value returned may not be that of the
5884 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +00005885
Chris Lattner10610642004-02-14 04:08:35 +00005886</div>
5887
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005888<!-- _______________________________________________________________________ -->
5889<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005890 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
Chris Lattner57e1f392006-01-13 02:03:13 +00005891</div>
5892
5893<div class="doc_text">
5894
5895<h5>Syntax:</h5>
5896<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005897 declare i8* @llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +00005898</pre>
5899
5900<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005901<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5902 of the function stack, for use
5903 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5904 useful for implementing language features like scoped automatic variable
5905 sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005906
5907<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005908<p>This intrinsic returns a opaque pointer value that can be passed
5909 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5910 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5911 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5912 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5913 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5914 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005915
5916</div>
5917
5918<!-- _______________________________________________________________________ -->
5919<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005920 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
Chris Lattner57e1f392006-01-13 02:03:13 +00005921</div>
5922
5923<div class="doc_text">
5924
5925<h5>Syntax:</h5>
5926<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00005927 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +00005928</pre>
5929
5930<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005931<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5932 the function stack to the state it was in when the
5933 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5934 executed. This is useful for implementing language features like scoped
5935 automatic variable sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005936
5937<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005938<p>See the description
5939 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +00005940
5941</div>
5942
Chris Lattner57e1f392006-01-13 02:03:13 +00005943<!-- _______________________________________________________________________ -->
5944<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005945 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005946</div>
5947
5948<div class="doc_text">
5949
5950<h5>Syntax:</h5>
5951<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005952 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005953</pre>
5954
5955<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005956<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5957 insert a prefetch instruction if supported; otherwise, it is a noop.
5958 Prefetches have no effect on the behavior of the program but can change its
5959 performance characteristics.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005960
5961<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005962<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5963 specifier determining if the fetch should be for a read (0) or write (1),
5964 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5965 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5966 and <tt>locality</tt> arguments must be constant integers.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005967
5968<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005969<p>This intrinsic does not modify the behavior of the program. In particular,
5970 prefetches cannot trap and do not produce a value. On targets that support
5971 this intrinsic, the prefetch can provide hints to the processor cache for
5972 better performance.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +00005973
5974</div>
5975
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005976<!-- _______________________________________________________________________ -->
5977<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00005978 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005979</div>
5980
5981<div class="doc_text">
5982
5983<h5>Syntax:</h5>
5984<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +00005985 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005986</pre>
5987
5988<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005989<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5990 Counter (PC) in a region of code to simulators and other tools. The method
5991 is target specific, but it is expected that the marker will use exported
5992 symbols to transmit the PC of the marker. The marker makes no guarantees
5993 that it will remain with any specific instruction after optimizations. It is
5994 possible that the presence of a marker will inhibit optimizations. The
5995 intended use is to be inserted after optimizations to allow correlations of
5996 simulation runs.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00005997
5998<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00005999<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00006000
6001<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006002<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006003 not support this intrinsic may ignore it.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +00006004
6005</div>
6006
Andrew Lenharth51b8d542005-11-11 16:47:30 +00006007<!-- _______________________________________________________________________ -->
6008<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006009 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
Andrew Lenharth51b8d542005-11-11 16:47:30 +00006010</div>
6011
6012<div class="doc_text">
6013
6014<h5>Syntax:</h5>
6015<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00006016 declare i64 @llvm.readcyclecounter()
Andrew Lenharth51b8d542005-11-11 16:47:30 +00006017</pre>
6018
6019<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006020<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6021 counter register (or similar low latency, high accuracy clocks) on those
6022 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6023 should map to RPCC. As the backing counters overflow quickly (on the order
6024 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +00006025
6026<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006027<p>When directly supported, reading the cycle counter should not modify any
6028 memory. Implementations are allowed to either return a application specific
6029 value or a system wide value. On backends without support, this is lowered
6030 to a constant 0.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +00006031
6032</div>
6033
Chris Lattner10610642004-02-14 04:08:35 +00006034<!-- ======================================================================= -->
6035<div class="doc_subsection">
Chris Lattner33aec9e2004-02-12 17:01:32 +00006036 <a name="int_libc">Standard C Library Intrinsics</a>
6037</div>
6038
6039<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006040
6041<p>LLVM provides intrinsics for a few important standard C library functions.
6042 These intrinsics allow source-language front-ends to pass information about
6043 the alignment of the pointer arguments to the code generator, providing
6044 opportunity for more efficient code generation.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006045
6046</div>
6047
6048<!-- _______________________________________________________________________ -->
6049<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006050 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006051</div>
6052
6053<div class="doc_text">
6054
6055<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006056<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
Mon P Wange88909b2010-04-07 06:35:53 +00006057 integer bit width and for different address spaces. Not all targets support
6058 all bit widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006059
Chris Lattner33aec9e2004-02-12 17:01:32 +00006060<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006061 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006062 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006063 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006064 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +00006065</pre>
6066
6067<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006068<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6069 source location to the destination location.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006070
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006071<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +00006072 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6073 and the pointers can be in specified address spaces.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006074
6075<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006076
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006077<p>The first argument is a pointer to the destination, the second is a pointer
6078 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +00006079 number of bytes to copy, the fourth argument is the alignment of the
6080 source and destination locations, and the fifth is a boolean indicating a
6081 volatile access.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006082
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006083<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006084 then the caller guarantees that both the source and destination pointers are
6085 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +00006086
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00006087<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6088 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6089 The detailed access behavior is not very cleanly specified and it is unwise
6090 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +00006091
Chris Lattner33aec9e2004-02-12 17:01:32 +00006092<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006093
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006094<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6095 source location to the destination location, which are not allowed to
6096 overlap. It copies "len" bytes of memory over. If the argument is known to
6097 be aligned to some boundary, this can be specified as the fourth argument,
6098 otherwise it should be set to 0 or 1.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +00006099
Chris Lattner33aec9e2004-02-12 17:01:32 +00006100</div>
6101
Chris Lattner0eb51b42004-02-12 18:10:10 +00006102<!-- _______________________________________________________________________ -->
6103<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006104 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006105</div>
6106
6107<div class="doc_text">
6108
6109<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +00006110<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Mon P Wange88909b2010-04-07 06:35:53 +00006111 width and for different address space. Not all targets support all bit
6112 widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006113
Chris Lattner0eb51b42004-02-12 18:10:10 +00006114<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006115 declare void @llvm.memmove.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006116 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006117 declare void @llvm.memmove.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +00006118 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +00006119</pre>
6120
6121<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006122<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6123 source location to the destination location. It is similar to the
6124 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6125 overlap.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006126
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006127<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +00006128 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6129 and the pointers can be in specified address spaces.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006130
6131<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006132
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006133<p>The first argument is a pointer to the destination, the second is a pointer
6134 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +00006135 number of bytes to copy, the fourth argument is the alignment of the
6136 source and destination locations, and the fifth is a boolean indicating a
6137 volatile access.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006138
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006139<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006140 then the caller guarantees that the source and destination pointers are
6141 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +00006142
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00006143<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6144 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6145 The detailed access behavior is not very cleanly specified and it is unwise
6146 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +00006147
Chris Lattner0eb51b42004-02-12 18:10:10 +00006148<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +00006149
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006150<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6151 source location to the destination location, which may overlap. It copies
6152 "len" bytes of memory over. If the argument is known to be aligned to some
6153 boundary, this can be specified as the fourth argument, otherwise it should
6154 be set to 0 or 1.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +00006155
Chris Lattner0eb51b42004-02-12 18:10:10 +00006156</div>
6157
Chris Lattner10610642004-02-14 04:08:35 +00006158<!-- _______________________________________________________________________ -->
6159<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006160 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner10610642004-02-14 04:08:35 +00006161</div>
6162
6163<div class="doc_text">
6164
6165<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +00006166<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
John Criswellcdcbbfc2010-07-30 16:30:28 +00006167 width and for different address spaces. However, not all targets support all
6168 bit widths.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006169
Chris Lattner10610642004-02-14 04:08:35 +00006170<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006171 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +00006172 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +00006173 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +00006174 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner10610642004-02-14 04:08:35 +00006175</pre>
6176
6177<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006178<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6179 particular byte value.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006180
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006181<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
John Criswellcdcbbfc2010-07-30 16:30:28 +00006182 intrinsic does not return a value and takes extra alignment/volatile
6183 arguments. Also, the destination can be in an arbitrary address space.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006184
6185<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006186<p>The first argument is a pointer to the destination to fill, the second is the
John Criswellcdcbbfc2010-07-30 16:30:28 +00006187 byte value with which to fill it, the third argument is an integer argument
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006188 specifying the number of bytes to fill, and the fourth argument is the known
John Criswellcdcbbfc2010-07-30 16:30:28 +00006189 alignment of the destination location.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006190
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00006191<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006192 then the caller guarantees that the destination pointer is aligned to that
6193 boundary.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006194
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +00006195<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6196 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6197 The detailed access behavior is not very cleanly specified and it is unwise
6198 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +00006199
Chris Lattner10610642004-02-14 04:08:35 +00006200<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006201<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6202 at the destination location. If the argument is known to be aligned to some
6203 boundary, this can be specified as the fourth argument, otherwise it should
6204 be set to 0 or 1.</p>
Chris Lattner10610642004-02-14 04:08:35 +00006205
Chris Lattner10610642004-02-14 04:08:35 +00006206</div>
6207
Chris Lattner32006282004-06-11 02:28:03 +00006208<!-- _______________________________________________________________________ -->
6209<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006210 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattnera4d74142005-07-21 01:29:16 +00006211</div>
6212
6213<div class="doc_text">
6214
6215<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006216<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6217 floating point or vector of floating point type. Not all targets support all
6218 types however.</p>
6219
Chris Lattnera4d74142005-07-21 01:29:16 +00006220<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +00006221 declare float @llvm.sqrt.f32(float %Val)
6222 declare double @llvm.sqrt.f64(double %Val)
6223 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6224 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6225 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +00006226</pre>
6227
6228<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006229<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6230 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6231 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6232 behavior for negative numbers other than -0.0 (which allows for better
6233 optimization, because there is no need to worry about errno being
6234 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +00006235
6236<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006237<p>The argument and return value are floating point numbers of the same
6238 type.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +00006239
6240<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006241<p>This function returns the sqrt of the specified operand if it is a
6242 nonnegative floating point number.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +00006243
Chris Lattnera4d74142005-07-21 01:29:16 +00006244</div>
6245
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006246<!-- _______________________________________________________________________ -->
6247<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006248 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006249</div>
6250
6251<div class="doc_text">
6252
6253<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006254<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6255 floating point or vector of floating point type. Not all targets support all
6256 types however.</p>
6257
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006258<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +00006259 declare float @llvm.powi.f32(float %Val, i32 %power)
6260 declare double @llvm.powi.f64(double %Val, i32 %power)
6261 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6262 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6263 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006264</pre>
6265
6266<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006267<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6268 specified (positive or negative) power. The order of evaluation of
6269 multiplications is not defined. When a vector of floating point type is
6270 used, the second argument remains a scalar integer value.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006271
6272<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006273<p>The second argument is an integer power, and the first is a value to raise to
6274 that power.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006275
6276<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006277<p>This function returns the first value raised to the second power with an
6278 unspecified sequence of rounding operations.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006279
Chris Lattnerf4d252d2006-09-08 06:34:02 +00006280</div>
6281
Dan Gohman91c284c2007-10-15 20:30:11 +00006282<!-- _______________________________________________________________________ -->
6283<div class="doc_subsubsection">
6284 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6285</div>
6286
6287<div class="doc_text">
6288
6289<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006290<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6291 floating point or vector of floating point type. Not all targets support all
6292 types however.</p>
6293
Dan Gohman91c284c2007-10-15 20:30:11 +00006294<pre>
6295 declare float @llvm.sin.f32(float %Val)
6296 declare double @llvm.sin.f64(double %Val)
6297 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6298 declare fp128 @llvm.sin.f128(fp128 %Val)
6299 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6300</pre>
6301
6302<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006303<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006304
6305<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006306<p>The argument and return value are floating point numbers of the same
6307 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006308
6309<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006310<p>This function returns the sine of the specified operand, returning the same
6311 values as the libm <tt>sin</tt> functions would, and handles error conditions
6312 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006313
Dan Gohman91c284c2007-10-15 20:30:11 +00006314</div>
6315
6316<!-- _______________________________________________________________________ -->
6317<div class="doc_subsubsection">
6318 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6319</div>
6320
6321<div class="doc_text">
6322
6323<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006324<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6325 floating point or vector of floating point type. Not all targets support all
6326 types however.</p>
6327
Dan Gohman91c284c2007-10-15 20:30:11 +00006328<pre>
6329 declare float @llvm.cos.f32(float %Val)
6330 declare double @llvm.cos.f64(double %Val)
6331 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6332 declare fp128 @llvm.cos.f128(fp128 %Val)
6333 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6334</pre>
6335
6336<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006337<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006338
6339<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006340<p>The argument and return value are floating point numbers of the same
6341 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006342
6343<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006344<p>This function returns the cosine of the specified operand, returning the same
6345 values as the libm <tt>cos</tt> functions would, and handles error conditions
6346 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006347
Dan Gohman91c284c2007-10-15 20:30:11 +00006348</div>
6349
6350<!-- _______________________________________________________________________ -->
6351<div class="doc_subsubsection">
6352 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6353</div>
6354
6355<div class="doc_text">
6356
6357<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006358<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6359 floating point or vector of floating point type. Not all targets support all
6360 types however.</p>
6361
Dan Gohman91c284c2007-10-15 20:30:11 +00006362<pre>
6363 declare float @llvm.pow.f32(float %Val, float %Power)
6364 declare double @llvm.pow.f64(double %Val, double %Power)
6365 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6366 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6367 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6368</pre>
6369
6370<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006371<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6372 specified (positive or negative) power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006373
6374<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006375<p>The second argument is a floating point power, and the first is a value to
6376 raise to that power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006377
6378<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006379<p>This function returns the first value raised to the second power, returning
6380 the same values as the libm <tt>pow</tt> functions would, and handles error
6381 conditions in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +00006382
Dan Gohman91c284c2007-10-15 20:30:11 +00006383</div>
6384
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006385<!-- ======================================================================= -->
6386<div class="doc_subsection">
Nate Begeman7e36c472006-01-13 23:26:38 +00006387 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006388</div>
6389
6390<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006391
6392<p>LLVM provides intrinsics for a few important bit manipulation operations.
6393 These allow efficient code generation for some algorithms.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006394
6395</div>
6396
6397<!-- _______________________________________________________________________ -->
6398<div class="doc_subsubsection">
Reid Spencera3e435f2007-04-04 02:42:35 +00006399 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
Nate Begeman7e36c472006-01-13 23:26:38 +00006400</div>
6401
6402<div class="doc_text">
6403
6404<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +00006405<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006406 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6407
Nate Begeman7e36c472006-01-13 23:26:38 +00006408<pre>
Chandler Carruth69940402007-08-04 01:51:18 +00006409 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
6410 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
6411 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +00006412</pre>
6413
6414<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006415<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6416 values with an even number of bytes (positive multiple of 16 bits). These
6417 are useful for performing operations on data that is not in the target's
6418 native byte order.</p>
Nate Begeman7e36c472006-01-13 23:26:38 +00006419
6420<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006421<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6422 and low byte of the input i16 swapped. Similarly,
6423 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6424 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6425 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6426 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6427 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6428 more, respectively).</p>
Nate Begeman7e36c472006-01-13 23:26:38 +00006429
6430</div>
6431
6432<!-- _______________________________________________________________________ -->
6433<div class="doc_subsubsection">
Reid Spencer0b118202006-01-16 21:12:35 +00006434 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006435</div>
6436
6437<div class="doc_text">
6438
6439<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +00006440<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006441 width. Not all targets support all bit widths however.</p>
6442
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006443<pre>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006444 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006445 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +00006446 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006447 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
6448 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006449</pre>
6450
6451<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006452<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6453 in a value.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006454
6455<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006456<p>The only argument is the value to be counted. The argument may be of any
6457 integer type. The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006458
6459<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006460<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006461
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006462</div>
6463
6464<!-- _______________________________________________________________________ -->
6465<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +00006466 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006467</div>
6468
6469<div class="doc_text">
6470
6471<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006472<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6473 integer bit width. Not all targets support all bit widths however.</p>
6474
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006475<pre>
Chandler Carruth69940402007-08-04 01:51:18 +00006476 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
6477 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +00006478 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006479 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
6480 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006481</pre>
6482
6483<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006484<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6485 leading zeros in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006486
6487<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006488<p>The only argument is the value to be counted. The argument may be of any
6489 integer type. The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006490
6491<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006492<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6493 zeros in a variable. If the src == 0 then the result is the size in bits of
6494 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006495
Andrew Lenharthec370fd2005-05-03 18:01:48 +00006496</div>
Chris Lattner32006282004-06-11 02:28:03 +00006497
Chris Lattnereff29ab2005-05-15 19:39:26 +00006498<!-- _______________________________________________________________________ -->
6499<div class="doc_subsubsection">
Chris Lattner8a886be2006-01-16 22:34:14 +00006500 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006501</div>
6502
6503<div class="doc_text">
6504
6505<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006506<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6507 integer bit width. Not all targets support all bit widths however.</p>
6508
Chris Lattnereff29ab2005-05-15 19:39:26 +00006509<pre>
Chandler Carruth69940402007-08-04 01:51:18 +00006510 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
6511 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +00006512 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +00006513 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
6514 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +00006515</pre>
6516
6517<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006518<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6519 trailing zeros.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006520
6521<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006522<p>The only argument is the value to be counted. The argument may be of any
6523 integer type. The return type must match the argument type.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006524
6525<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006526<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6527 zeros in a variable. If the src == 0 then the result is the size in bits of
6528 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +00006529
Chris Lattnereff29ab2005-05-15 19:39:26 +00006530</div>
6531
Bill Wendlingda01af72009-02-08 04:04:40 +00006532<!-- ======================================================================= -->
6533<div class="doc_subsection">
6534 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6535</div>
6536
6537<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006538
6539<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingda01af72009-02-08 04:04:40 +00006540
6541</div>
6542
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006543<!-- _______________________________________________________________________ -->
6544<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006545 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006546</div>
6547
6548<div class="doc_text">
6549
6550<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006551<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006552 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006553
6554<pre>
6555 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6556 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6557 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6558</pre>
6559
6560<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006561<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006562 a signed addition of the two arguments, and indicate whether an overflow
6563 occurred during the signed summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006564
6565<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006566<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006567 be of integer types of any bit width, but they must have the same bit
6568 width. The second element of the result structure must be of
6569 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6570 undergo signed addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006571
6572<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006573<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006574 a signed addition of the two variables. They return a structure &mdash; the
6575 first element of which is the signed summation, and the second element of
6576 which is a bit specifying if the signed summation resulted in an
6577 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006578
6579<h5>Examples:</h5>
6580<pre>
6581 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6582 %sum = extractvalue {i32, i1} %res, 0
6583 %obit = extractvalue {i32, i1} %res, 1
6584 br i1 %obit, label %overflow, label %normal
6585</pre>
6586
6587</div>
6588
6589<!-- _______________________________________________________________________ -->
6590<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006591 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006592</div>
6593
6594<div class="doc_text">
6595
6596<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006597<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006598 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006599
6600<pre>
6601 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6602 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6603 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6604</pre>
6605
6606<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006607<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006608 an unsigned addition of the two arguments, and indicate whether a carry
6609 occurred during the unsigned summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006610
6611<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006612<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006613 be of integer types of any bit width, but they must have the same bit
6614 width. The second element of the result structure must be of
6615 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6616 undergo unsigned addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006617
6618<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006619<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006620 an unsigned addition of the two arguments. They return a structure &mdash;
6621 the first element of which is the sum, and the second element of which is a
6622 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006623
6624<h5>Examples:</h5>
6625<pre>
6626 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6627 %sum = extractvalue {i32, i1} %res, 0
6628 %obit = extractvalue {i32, i1} %res, 1
6629 br i1 %obit, label %carry, label %normal
6630</pre>
6631
6632</div>
6633
6634<!-- _______________________________________________________________________ -->
6635<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006636 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006637</div>
6638
6639<div class="doc_text">
6640
6641<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006642<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006643 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006644
6645<pre>
6646 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6647 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6648 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6649</pre>
6650
6651<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006652<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006653 a signed subtraction of the two arguments, and indicate whether an overflow
6654 occurred during the signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006655
6656<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006657<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006658 be of integer types of any bit width, but they must have the same bit
6659 width. The second element of the result structure must be of
6660 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6661 undergo signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006662
6663<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006664<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006665 a signed subtraction of the two arguments. They return a structure &mdash;
6666 the first element of which is the subtraction, and the second element of
6667 which is a bit specifying if the signed subtraction resulted in an
6668 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006669
6670<h5>Examples:</h5>
6671<pre>
6672 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6673 %sum = extractvalue {i32, i1} %res, 0
6674 %obit = extractvalue {i32, i1} %res, 1
6675 br i1 %obit, label %overflow, label %normal
6676</pre>
6677
6678</div>
6679
6680<!-- _______________________________________________________________________ -->
6681<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006682 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006683</div>
6684
6685<div class="doc_text">
6686
6687<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006688<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006689 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006690
6691<pre>
6692 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6693 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6694 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6695</pre>
6696
6697<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006698<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006699 an unsigned subtraction of the two arguments, and indicate whether an
6700 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006701
6702<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006703<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006704 be of integer types of any bit width, but they must have the same bit
6705 width. The second element of the result structure must be of
6706 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6707 undergo unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006708
6709<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006710<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006711 an unsigned subtraction of the two arguments. They return a structure &mdash;
6712 the first element of which is the subtraction, and the second element of
6713 which is a bit specifying if the unsigned subtraction resulted in an
6714 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006715
6716<h5>Examples:</h5>
6717<pre>
6718 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6719 %sum = extractvalue {i32, i1} %res, 0
6720 %obit = extractvalue {i32, i1} %res, 1
6721 br i1 %obit, label %overflow, label %normal
6722</pre>
6723
6724</div>
6725
6726<!-- _______________________________________________________________________ -->
6727<div class="doc_subsubsection">
Bill Wendlingda01af72009-02-08 04:04:40 +00006728 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006729</div>
6730
6731<div class="doc_text">
6732
6733<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006734<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006735 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006736
6737<pre>
6738 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6739 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6740 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6741</pre>
6742
6743<h5>Overview:</h5>
6744
6745<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006746 a signed multiplication of the two arguments, and indicate whether an
6747 overflow occurred during the signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006748
6749<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006750<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006751 be of integer types of any bit width, but they must have the same bit
6752 width. The second element of the result structure must be of
6753 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6754 undergo signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006755
6756<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006757<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006758 a signed multiplication of the two arguments. They return a structure &mdash;
6759 the first element of which is the multiplication, and the second element of
6760 which is a bit specifying if the signed multiplication resulted in an
6761 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +00006762
6763<h5>Examples:</h5>
6764<pre>
6765 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6766 %sum = extractvalue {i32, i1} %res, 0
6767 %obit = extractvalue {i32, i1} %res, 1
6768 br i1 %obit, label %overflow, label %normal
6769</pre>
6770
Reid Spencerf86037f2007-04-11 23:23:49 +00006771</div>
6772
Bill Wendling41b485c2009-02-08 23:00:09 +00006773<!-- _______________________________________________________________________ -->
6774<div class="doc_subsubsection">
6775 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6776</div>
6777
6778<div class="doc_text">
6779
6780<h5>Syntax:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006781<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006782 on any integer bit width.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006783
6784<pre>
6785 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6786 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6787 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6788</pre>
6789
6790<h5>Overview:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006791<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006792 a unsigned multiplication of the two arguments, and indicate whether an
6793 overflow occurred during the unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006794
6795<h5>Arguments:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006796<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006797 be of integer types of any bit width, but they must have the same bit
6798 width. The second element of the result structure must be of
6799 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6800 undergo unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006801
6802<h5>Semantics:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +00006803<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006804 an unsigned multiplication of the two arguments. They return a structure
6805 &mdash; the first element of which is the multiplication, and the second
6806 element of which is a bit specifying if the unsigned multiplication resulted
6807 in an overflow.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +00006808
6809<h5>Examples:</h5>
6810<pre>
6811 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6812 %sum = extractvalue {i32, i1} %res, 0
6813 %obit = extractvalue {i32, i1} %res, 1
6814 br i1 %obit, label %overflow, label %normal
6815</pre>
6816
6817</div>
6818
Chris Lattner8ff75902004-01-06 05:31:32 +00006819<!-- ======================================================================= -->
6820<div class="doc_subsection">
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006821 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
6822</div>
6823
6824<div class="doc_text">
6825
Chris Lattner0cec9c82010-03-15 04:12:21 +00006826<p>Half precision floating point is a storage-only format. This means that it is
6827 a dense encoding (in memory) but does not support computation in the
6828 format.</p>
Chris Lattner82c3dc62010-03-14 23:03:31 +00006829
Chris Lattner0cec9c82010-03-15 04:12:21 +00006830<p>This means that code must first load the half-precision floating point
Chris Lattner82c3dc62010-03-14 23:03:31 +00006831 value as an i16, then convert it to float with <a
6832 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
6833 Computation can then be performed on the float value (including extending to
Chris Lattner0cec9c82010-03-15 04:12:21 +00006834 double etc). To store the value back to memory, it is first converted to
6835 float if needed, then converted to i16 with
Chris Lattner82c3dc62010-03-14 23:03:31 +00006836 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
6837 storing as an i16 value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006838</div>
6839
6840<!-- _______________________________________________________________________ -->
6841<div class="doc_subsubsection">
Chris Lattner82c3dc62010-03-14 23:03:31 +00006842 <a name="int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006843</div>
6844
6845<div class="doc_text">
6846
6847<h5>Syntax:</h5>
6848<pre>
6849 declare i16 @llvm.convert.to.fp16(f32 %a)
6850</pre>
6851
6852<h5>Overview:</h5>
6853<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6854 a conversion from single precision floating point format to half precision
6855 floating point format.</p>
6856
6857<h5>Arguments:</h5>
6858<p>The intrinsic function contains single argument - the value to be
6859 converted.</p>
6860
6861<h5>Semantics:</h5>
6862<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6863 a conversion from single precision floating point format to half precision
Chris Lattner0cec9c82010-03-15 04:12:21 +00006864 floating point format. The return value is an <tt>i16</tt> which
Chris Lattner82c3dc62010-03-14 23:03:31 +00006865 contains the converted number.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006866
6867<h5>Examples:</h5>
6868<pre>
6869 %res = call i16 @llvm.convert.to.fp16(f32 %a)
6870 store i16 %res, i16* @x, align 2
6871</pre>
6872
6873</div>
6874
6875<!-- _______________________________________________________________________ -->
6876<div class="doc_subsubsection">
Chris Lattner82c3dc62010-03-14 23:03:31 +00006877 <a name="int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006878</div>
6879
6880<div class="doc_text">
6881
6882<h5>Syntax:</h5>
6883<pre>
6884 declare f32 @llvm.convert.from.fp16(i16 %a)
6885</pre>
6886
6887<h5>Overview:</h5>
6888<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
6889 a conversion from half precision floating point format to single precision
6890 floating point format.</p>
6891
6892<h5>Arguments:</h5>
6893<p>The intrinsic function contains single argument - the value to be
6894 converted.</p>
6895
6896<h5>Semantics:</h5>
6897<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner0cec9c82010-03-15 04:12:21 +00006898 conversion from half single precision floating point format to single
Chris Lattner82c3dc62010-03-14 23:03:31 +00006899 precision floating point format. The input half-float value is represented by
6900 an <tt>i16</tt> value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +00006901
6902<h5>Examples:</h5>
6903<pre>
6904 %a = load i16* @x, align 2
6905 %res = call f32 @llvm.convert.from.fp16(i16 %a)
6906</pre>
6907
6908</div>
6909
6910<!-- ======================================================================= -->
6911<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +00006912 <a name="int_debugger">Debugger Intrinsics</a>
6913</div>
6914
6915<div class="doc_text">
Chris Lattner8ff75902004-01-06 05:31:32 +00006916
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006917<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6918 prefix), are described in
6919 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6920 Level Debugging</a> document.</p>
6921
6922</div>
Chris Lattner8ff75902004-01-06 05:31:32 +00006923
Jim Laskeydd4ef1b2007-03-14 19:31:19 +00006924<!-- ======================================================================= -->
6925<div class="doc_subsection">
6926 <a name="int_eh">Exception Handling Intrinsics</a>
6927</div>
6928
6929<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006930
6931<p>The LLVM exception handling intrinsics (which all start with
6932 <tt>llvm.eh.</tt> prefix), are described in
6933 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6934 Handling</a> document.</p>
6935
Jim Laskeydd4ef1b2007-03-14 19:31:19 +00006936</div>
6937
Tanya Lattner6d806e92007-06-15 20:50:54 +00006938<!-- ======================================================================= -->
6939<div class="doc_subsection">
Duncan Sandsf7331b32007-09-11 14:10:23 +00006940 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands36397f52007-07-27 12:58:54 +00006941</div>
6942
6943<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006944
6945<p>This intrinsic makes it possible to excise one parameter, marked with
Dan Gohmanff235352010-07-02 23:18:08 +00006946 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
6947 The result is a callable
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006948 function pointer lacking the nest parameter - the caller does not need to
6949 provide a value for it. Instead, the value to use is stored in advance in a
6950 "trampoline", a block of memory usually allocated on the stack, which also
6951 contains code to splice the nest value into the argument list. This is used
6952 to implement the GCC nested function address extension.</p>
6953
6954<p>For example, if the function is
6955 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6956 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6957 follows:</p>
6958
Benjamin Kramer26fe25f2010-07-13 12:26:09 +00006959<pre class="doc_code">
Duncan Sandsf7331b32007-09-11 14:10:23 +00006960 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6961 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00006962 %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 +00006963 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands36397f52007-07-27 12:58:54 +00006964</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006965
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00006966<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
6967 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006968
Duncan Sands36397f52007-07-27 12:58:54 +00006969</div>
6970
6971<!-- _______________________________________________________________________ -->
6972<div class="doc_subsubsection">
6973 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6974</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006975
Duncan Sands36397f52007-07-27 12:58:54 +00006976<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006977
Duncan Sands36397f52007-07-27 12:58:54 +00006978<h5>Syntax:</h5>
6979<pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006980 declare i8* @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands36397f52007-07-27 12:58:54 +00006981</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006982
Duncan Sands36397f52007-07-27 12:58:54 +00006983<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006984<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6985 function pointer suitable for executing it.</p>
6986
Duncan Sands36397f52007-07-27 12:58:54 +00006987<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006988<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6989 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6990 sufficiently aligned block of memory; this memory is written to by the
6991 intrinsic. Note that the size and the alignment are target-specific - LLVM
6992 currently provides no portable way of determining them, so a front-end that
6993 generates this intrinsic needs to have some target-specific knowledge.
6994 The <tt>func</tt> argument must hold a function bitcast to
6995 an <tt>i8*</tt>.</p>
6996
Duncan Sands36397f52007-07-27 12:58:54 +00006997<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00006998<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6999 dependent code, turning it into a function. A pointer to this function is
7000 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
7001 function pointer type</a> before being called. The new function's signature
7002 is the same as that of <tt>func</tt> with any arguments marked with
7003 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
7004 is allowed, and it must be of pointer type. Calling the new function is
7005 equivalent to calling <tt>func</tt> with the same argument list, but
7006 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
7007 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
7008 by <tt>tramp</tt> is modified, then the effect of any later call to the
7009 returned function pointer is undefined.</p>
7010
Duncan Sands36397f52007-07-27 12:58:54 +00007011</div>
7012
7013<!-- ======================================================================= -->
7014<div class="doc_subsection">
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007015 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
7016</div>
7017
7018<div class="doc_text">
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007019
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007020<p>These intrinsic functions expand the "universal IR" of LLVM to represent
7021 hardware constructs for atomic operations and memory synchronization. This
7022 provides an interface to the hardware, not an interface to the programmer. It
7023 is aimed at a low enough level to allow any programming models or APIs
7024 (Application Programming Interfaces) which need atomic behaviors to map
7025 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
7026 hardware provides a "universal IR" for source languages, it also provides a
7027 starting point for developing a "universal" atomic operation and
7028 synchronization IR.</p>
7029
7030<p>These do <em>not</em> form an API such as high-level threading libraries,
7031 software transaction memory systems, atomic primitives, and intrinsic
7032 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
7033 application libraries. The hardware interface provided by LLVM should allow
7034 a clean implementation of all of these APIs and parallel programming models.
7035 No one model or paradigm should be selected above others unless the hardware
7036 itself ubiquitously does so.</p>
7037
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007038</div>
7039
7040<!-- _______________________________________________________________________ -->
7041<div class="doc_subsubsection">
7042 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
7043</div>
7044<div class="doc_text">
7045<h5>Syntax:</h5>
7046<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007047 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 +00007048</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007049
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007050<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007051<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7052 specific pairs of memory access types.</p>
7053
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007054<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007055<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7056 The first four arguments enables a specific barrier as listed below. The
Dan Gohmanb55a1ee2010-03-01 17:41:39 +00007057 fifth argument specifies that the barrier applies to io or device or uncached
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007058 memory.</p>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007059
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007060<ul>
7061 <li><tt>ll</tt>: load-load barrier</li>
7062 <li><tt>ls</tt>: load-store barrier</li>
7063 <li><tt>sl</tt>: store-load barrier</li>
7064 <li><tt>ss</tt>: store-store barrier</li>
7065 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7066</ul>
7067
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007068<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007069<p>This intrinsic causes the system to enforce some ordering constraints upon
7070 the loads and stores of the program. This barrier does not
7071 indicate <em>when</em> any events will occur, it only enforces
7072 an <em>order</em> in which they occur. For any of the specified pairs of load
7073 and store operations (f.ex. load-load, or store-load), all of the first
7074 operations preceding the barrier will complete before any of the second
7075 operations succeeding the barrier begin. Specifically the semantics for each
7076 pairing is as follows:</p>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007077
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007078<ul>
7079 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7080 after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007081 <li><tt>ls</tt>: All loads before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007082 store after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007083 <li><tt>ss</tt>: All stores before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007084 store after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007085 <li><tt>sl</tt>: All stores before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007086 load after the barrier begins.</li>
7087</ul>
7088
7089<p>These semantics are applied with a logical "and" behavior when more than one
7090 is enabled in a single memory barrier intrinsic.</p>
7091
7092<p>Backends may implement stronger barriers than those requested when they do
7093 not support as fine grained a barrier as requested. Some architectures do
7094 not need all types of barriers and on such architectures, these become
7095 noops.</p>
7096
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007097<h5>Example:</h5>
7098<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007099%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7100%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007101 store i32 4, %ptr
7102
7103%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007104 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false)
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007105 <i>; guarantee the above finishes</i>
7106 store i32 8, %ptr <i>; before this begins</i>
7107</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007108
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007109</div>
7110
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007111<!-- _______________________________________________________________________ -->
7112<div class="doc_subsubsection">
Mon P Wang28873102008-06-25 08:15:39 +00007113 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007114</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007115
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007116<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007117
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007118<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007119<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7120 any integer bit width and for different address spaces. Not all targets
7121 support all bit widths however.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007122
7123<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007124 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt;)
7125 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt;)
7126 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt;)
7127 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 +00007128</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007129
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007130<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007131<p>This loads a value in memory and compares it to a given value. If they are
7132 equal, it stores a new value into the memory.</p>
7133
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007134<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007135<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7136 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7137 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7138 this integer type. While any bit width integer may be used, targets may only
7139 lower representations they support in hardware.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007140
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007141<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007142<p>This entire intrinsic must be executed atomically. It first loads the value
7143 in memory pointed to by <tt>ptr</tt> and compares it with the
7144 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7145 memory. The loaded value is yielded in all cases. This provides the
7146 equivalent of an atomic compare-and-swap operation within the SSA
7147 framework.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007148
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007149<h5>Examples:</h5>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007150<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007151%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7152%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007153 store i32 4, %ptr
7154
7155%val1 = add i32 4, 4
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007156%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007157 <i>; yields {i32}:result1 = 4</i>
7158%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7159%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7160
7161%val2 = add i32 1, 1
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007162%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007163 <i>; yields {i32}:result2 = 8</i>
7164%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7165
7166%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7167</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007168
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007169</div>
7170
7171<!-- _______________________________________________________________________ -->
7172<div class="doc_subsubsection">
7173 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7174</div>
7175<div class="doc_text">
7176<h5>Syntax:</h5>
7177
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007178<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7179 integer bit width. Not all targets support all bit widths however.</p>
7180
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007181<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007182 declare i8 @llvm.atomic.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;val&gt;)
7183 declare i16 @llvm.atomic.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;val&gt;)
7184 declare i32 @llvm.atomic.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;val&gt;)
7185 declare i64 @llvm.atomic.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;val&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007186</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007187
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007188<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007189<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7190 the value from memory. It then stores the value in <tt>val</tt> in the memory
7191 at <tt>ptr</tt>.</p>
7192
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007193<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007194<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7195 the <tt>val</tt> argument and the result must be integers of the same bit
7196 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7197 integer type. The targets may only lower integer representations they
7198 support.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007199
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007200<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007201<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7202 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7203 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007204
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007205<h5>Examples:</h5>
7206<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007207%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7208%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007209 store i32 4, %ptr
7210
7211%val1 = add i32 4, 4
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007212%result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007213 <i>; yields {i32}:result1 = 4</i>
7214%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7215%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7216
7217%val2 = add i32 1, 1
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007218%result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007219 <i>; yields {i32}:result2 = 8</i>
7220
7221%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7222%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7223</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007224
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007225</div>
7226
7227<!-- _______________________________________________________________________ -->
7228<div class="doc_subsubsection">
Mon P Wang28873102008-06-25 08:15:39 +00007229 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007230
7231</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007232
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007233<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007234
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007235<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007236<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7237 any integer bit width. Not all targets support all bit widths however.</p>
7238
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007239<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007240 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7241 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7242 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7243 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007244</pre>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007245
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007246<h5>Overview:</h5>
7247<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7248 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7249
7250<h5>Arguments:</h5>
7251<p>The intrinsic takes two arguments, the first a pointer to an integer value
7252 and the second an integer value. The result is also an integer value. These
7253 integer types can have any bit width, but they must all have the same bit
7254 width. The targets may only lower integer representations they support.</p>
7255
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007256<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007257<p>This intrinsic does a series of operations atomically. It first loads the
7258 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7259 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007260
7261<h5>Examples:</h5>
7262<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007263%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7264%ptr = bitcast i8* %mallocP to i32*
7265 store i32 4, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007266%result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007267 <i>; yields {i32}:result1 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007268%result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007269 <i>; yields {i32}:result2 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007270%result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007271 <i>; yields {i32}:result3 = 10</i>
Mon P Wang28873102008-06-25 08:15:39 +00007272%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007273</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007274
Andrew Lenharthab0b9492008-02-21 06:45:13 +00007275</div>
7276
Mon P Wang28873102008-06-25 08:15:39 +00007277<!-- _______________________________________________________________________ -->
7278<div class="doc_subsubsection">
7279 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
7280
7281</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007282
Mon P Wang28873102008-06-25 08:15:39 +00007283<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007284
Mon P Wang28873102008-06-25 08:15:39 +00007285<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007286<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
7287 any integer bit width and for different address spaces. Not all targets
7288 support all bit widths however.</p>
7289
Mon P Wang28873102008-06-25 08:15:39 +00007290<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007291 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7292 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7293 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7294 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007295</pre>
Mon P Wang28873102008-06-25 08:15:39 +00007296
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007297<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007298<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007299 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7300
7301<h5>Arguments:</h5>
7302<p>The intrinsic takes two arguments, the first a pointer to an integer value
7303 and the second an integer value. The result is also an integer value. These
7304 integer types can have any bit width, but they must all have the same bit
7305 width. The targets may only lower integer representations they support.</p>
7306
Mon P Wang28873102008-06-25 08:15:39 +00007307<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007308<p>This intrinsic does a series of operations atomically. It first loads the
7309 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
7310 result to <tt>ptr</tt>. It yields the original value stored
7311 at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007312
7313<h5>Examples:</h5>
7314<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007315%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7316%ptr = bitcast i8* %mallocP to i32*
7317 store i32 8, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007318%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
Mon P Wang28873102008-06-25 08:15:39 +00007319 <i>; yields {i32}:result1 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007320%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
Mon P Wang28873102008-06-25 08:15:39 +00007321 <i>; yields {i32}:result2 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007322%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
Mon P Wang28873102008-06-25 08:15:39 +00007323 <i>; yields {i32}:result3 = 2</i>
7324%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
7325</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007326
Mon P Wang28873102008-06-25 08:15:39 +00007327</div>
7328
7329<!-- _______________________________________________________________________ -->
7330<div class="doc_subsubsection">
7331 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
7332 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
7333 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
7334 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
Mon P Wang28873102008-06-25 08:15:39 +00007335</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007336
Mon P Wang28873102008-06-25 08:15:39 +00007337<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007338
Mon P Wang28873102008-06-25 08:15:39 +00007339<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007340<p>These are overloaded intrinsics. You can
7341 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
7342 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
7343 bit width and for different address spaces. Not all targets support all bit
7344 widths however.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007345
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007346<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007347 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7348 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7349 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7350 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007351</pre>
7352
7353<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007354 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7355 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7356 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7357 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007358</pre>
7359
7360<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007361 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7362 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7363 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7364 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007365</pre>
7366
7367<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007368 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7369 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7370 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7371 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007372</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007373
Mon P Wang28873102008-06-25 08:15:39 +00007374<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007375<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
7376 the value stored in memory at <tt>ptr</tt>. It yields the original value
7377 at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007378
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007379<h5>Arguments:</h5>
7380<p>These intrinsics take two arguments, the first a pointer to an integer value
7381 and the second an integer value. The result is also an integer value. These
7382 integer types can have any bit width, but they must all have the same bit
7383 width. The targets may only lower integer representations they support.</p>
7384
Mon P Wang28873102008-06-25 08:15:39 +00007385<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007386<p>These intrinsics does a series of operations atomically. They first load the
7387 value stored at <tt>ptr</tt>. They then do the bitwise
7388 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
7389 original value stored at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007390
7391<h5>Examples:</h5>
7392<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007393%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7394%ptr = bitcast i8* %mallocP to i32*
7395 store i32 0x0F0F, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007396%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang28873102008-06-25 08:15:39 +00007397 <i>; yields {i32}:result0 = 0x0F0F</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007398%result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang28873102008-06-25 08:15:39 +00007399 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007400%result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang28873102008-06-25 08:15:39 +00007401 <i>; yields {i32}:result2 = 0xF0</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007402%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang28873102008-06-25 08:15:39 +00007403 <i>; yields {i32}:result3 = FF</i>
7404%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
7405</pre>
Mon P Wang28873102008-06-25 08:15:39 +00007406
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007407</div>
Mon P Wang28873102008-06-25 08:15:39 +00007408
7409<!-- _______________________________________________________________________ -->
7410<div class="doc_subsubsection">
7411 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
7412 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
7413 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
7414 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
Mon P Wang28873102008-06-25 08:15:39 +00007415</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007416
Mon P Wang28873102008-06-25 08:15:39 +00007417<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007418
Mon P Wang28873102008-06-25 08:15:39 +00007419<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007420<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
7421 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
7422 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
7423 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007424
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007425<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007426 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7427 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7428 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7429 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007430</pre>
7431
7432<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007433 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7434 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7435 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7436 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007437</pre>
7438
7439<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007440 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7441 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7442 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7443 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007444</pre>
7445
7446<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007447 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7448 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7449 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7450 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +00007451</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007452
Mon P Wang28873102008-06-25 08:15:39 +00007453<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +00007454<p>These intrinsics takes the signed or unsigned minimum or maximum of
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007455 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7456 original value at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007457
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007458<h5>Arguments:</h5>
7459<p>These intrinsics take two arguments, the first a pointer to an integer value
7460 and the second an integer value. The result is also an integer value. These
7461 integer types can have any bit width, but they must all have the same bit
7462 width. The targets may only lower integer representations they support.</p>
7463
Mon P Wang28873102008-06-25 08:15:39 +00007464<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007465<p>These intrinsics does a series of operations atomically. They first load the
7466 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7467 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7468 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +00007469
7470<h5>Examples:</h5>
7471<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +00007472%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7473%ptr = bitcast i8* %mallocP to i32*
7474 store i32 7, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007475%result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
Mon P Wang28873102008-06-25 08:15:39 +00007476 <i>; yields {i32}:result0 = 7</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007477%result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
Mon P Wang28873102008-06-25 08:15:39 +00007478 <i>; yields {i32}:result1 = -2</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007479%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
Mon P Wang28873102008-06-25 08:15:39 +00007480 <i>; yields {i32}:result2 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007481%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
Mon P Wang28873102008-06-25 08:15:39 +00007482 <i>; yields {i32}:result3 = 8</i>
7483%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7484</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007485
Mon P Wang28873102008-06-25 08:15:39 +00007486</div>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007487
Nick Lewyckycc271862009-10-13 07:03:23 +00007488
7489<!-- ======================================================================= -->
7490<div class="doc_subsection">
7491 <a name="int_memorymarkers">Memory Use Markers</a>
7492</div>
7493
7494<div class="doc_text">
7495
7496<p>This class of intrinsics exists to information about the lifetime of memory
7497 objects and ranges where variables are immutable.</p>
7498
7499</div>
7500
7501<!-- _______________________________________________________________________ -->
7502<div class="doc_subsubsection">
7503 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7504</div>
7505
7506<div class="doc_text">
7507
7508<h5>Syntax:</h5>
7509<pre>
7510 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7511</pre>
7512
7513<h5>Overview:</h5>
7514<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7515 object's lifetime.</p>
7516
7517<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +00007518<p>The first argument is a constant integer representing the size of the
7519 object, or -1 if it is variable sized. The second argument is a pointer to
7520 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007521
7522<h5>Semantics:</h5>
7523<p>This intrinsic indicates that before this point in the code, the value of the
7524 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
Nick Lewycky8d336592009-10-27 16:56:58 +00007525 never be used and has an undefined value. A load from the pointer that
7526 precedes this intrinsic can be replaced with
Nick Lewyckycc271862009-10-13 07:03:23 +00007527 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7528
7529</div>
7530
7531<!-- _______________________________________________________________________ -->
7532<div class="doc_subsubsection">
7533 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7534</div>
7535
7536<div class="doc_text">
7537
7538<h5>Syntax:</h5>
7539<pre>
7540 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7541</pre>
7542
7543<h5>Overview:</h5>
7544<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7545 object's lifetime.</p>
7546
7547<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +00007548<p>The first argument is a constant integer representing the size of the
7549 object, or -1 if it is variable sized. The second argument is a pointer to
7550 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007551
7552<h5>Semantics:</h5>
7553<p>This intrinsic indicates that after this point in the code, the value of the
7554 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7555 never be used and has an undefined value. Any stores into the memory object
7556 following this intrinsic may be removed as dead.
7557
7558</div>
7559
7560<!-- _______________________________________________________________________ -->
7561<div class="doc_subsubsection">
7562 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7563</div>
7564
7565<div class="doc_text">
7566
7567<h5>Syntax:</h5>
7568<pre>
Nick Lewycky29b6cb42010-11-30 04:13:41 +00007569 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
Nick Lewyckycc271862009-10-13 07:03:23 +00007570</pre>
7571
7572<h5>Overview:</h5>
7573<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7574 a memory object will not change.</p>
7575
7576<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +00007577<p>The first argument is a constant integer representing the size of the
7578 object, or -1 if it is variable sized. The second argument is a pointer to
7579 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007580
7581<h5>Semantics:</h5>
7582<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7583 the return value, the referenced memory location is constant and
7584 unchanging.</p>
7585
7586</div>
7587
7588<!-- _______________________________________________________________________ -->
7589<div class="doc_subsubsection">
7590 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7591</div>
7592
7593<div class="doc_text">
7594
7595<h5>Syntax:</h5>
7596<pre>
7597 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7598</pre>
7599
7600<h5>Overview:</h5>
7601<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7602 a memory object are mutable.</p>
7603
7604<h5>Arguments:</h5>
7605<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky321333e2009-10-13 07:57:33 +00007606 The second argument is a constant integer representing the size of the
7607 object, or -1 if it is variable sized and the third argument is a pointer
7608 to the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +00007609
7610<h5>Semantics:</h5>
7611<p>This intrinsic indicates that the memory is mutable again.</p>
7612
7613</div>
7614
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +00007615<!-- ======================================================================= -->
7616<div class="doc_subsection">
Tanya Lattner6d806e92007-06-15 20:50:54 +00007617 <a name="int_general">General Intrinsics</a>
7618</div>
7619
7620<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007621
7622<p>This class of intrinsics is designed to be generic and has no specific
7623 purpose.</p>
7624
Tanya Lattner6d806e92007-06-15 20:50:54 +00007625</div>
7626
7627<!-- _______________________________________________________________________ -->
7628<div class="doc_subsubsection">
7629 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7630</div>
7631
7632<div class="doc_text">
7633
7634<h5>Syntax:</h5>
7635<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007636 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 +00007637</pre>
7638
7639<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007640<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +00007641
7642<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007643<p>The first argument is a pointer to a value, the second is a pointer to a
7644 global string, the third is a pointer to a global string which is the source
7645 file name, and the last argument is the line number.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +00007646
7647<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007648<p>This intrinsic allows annotation of local variables with arbitrary strings.
7649 This can be useful for special purpose optimizations that want to look for
7650 these annotations. These have no other defined use, they are ignored by code
7651 generation and optimization.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +00007652
Tanya Lattner6d806e92007-06-15 20:50:54 +00007653</div>
7654
Tanya Lattnerb6367882007-09-21 22:59:12 +00007655<!-- _______________________________________________________________________ -->
7656<div class="doc_subsubsection">
Tanya Lattnere1a8da02007-09-21 23:57:59 +00007657 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007658</div>
7659
7660<div class="doc_text">
7661
7662<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007663<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7664 any integer bit width.</p>
7665
Tanya Lattnerb6367882007-09-21 22:59:12 +00007666<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007667 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7668 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7669 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7670 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7671 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 +00007672</pre>
7673
7674<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007675<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007676
7677<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007678<p>The first argument is an integer value (result of some expression), the
7679 second is a pointer to a global string, the third is a pointer to a global
7680 string which is the source file name, and the last argument is the line
7681 number. It returns the value of the first argument.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007682
7683<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007684<p>This intrinsic allows annotations to be put on arbitrary expressions with
7685 arbitrary strings. This can be useful for special purpose optimizations that
7686 want to look for these annotations. These have no other defined use, they
7687 are ignored by code generation and optimization.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +00007688
Tanya Lattnerb6367882007-09-21 22:59:12 +00007689</div>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +00007690
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007691<!-- _______________________________________________________________________ -->
7692<div class="doc_subsubsection">
7693 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7694</div>
7695
7696<div class="doc_text">
7697
7698<h5>Syntax:</h5>
7699<pre>
7700 declare void @llvm.trap()
7701</pre>
7702
7703<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007704<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007705
7706<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007707<p>None.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007708
7709<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007710<p>This intrinsics is lowered to the target dependent trap instruction. If the
7711 target does not have a trap instruction, this intrinsic will be lowered to
7712 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007713
Anton Korobeynikov4cb86182008-01-15 22:31:34 +00007714</div>
7715
Bill Wendling69e4adb2008-11-19 05:56:17 +00007716<!-- _______________________________________________________________________ -->
7717<div class="doc_subsubsection">
Misha Brukmandccb0252008-11-22 23:55:29 +00007718 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendling69e4adb2008-11-19 05:56:17 +00007719</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007720
Bill Wendling69e4adb2008-11-19 05:56:17 +00007721<div class="doc_text">
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007722
Bill Wendling69e4adb2008-11-19 05:56:17 +00007723<h5>Syntax:</h5>
7724<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007725 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling69e4adb2008-11-19 05:56:17 +00007726</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007727
Bill Wendling69e4adb2008-11-19 05:56:17 +00007728<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007729<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7730 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7731 ensure that it is placed on the stack before local variables.</p>
7732
Bill Wendling69e4adb2008-11-19 05:56:17 +00007733<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007734<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7735 arguments. The first argument is the value loaded from the stack
7736 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7737 that has enough space to hold the value of the guard.</p>
7738
Bill Wendling69e4adb2008-11-19 05:56:17 +00007739<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007740<p>This intrinsic causes the prologue/epilogue inserter to force the position of
7741 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7742 stack. This is to ensure that if a local variable on the stack is
7743 overwritten, it will destroy the value of the guard. When the function exits,
Bill Wendling1b383ba2010-10-27 01:07:41 +00007744 the guard on the stack is checked against the original guard. If they are
Bill Wendlinge910b4c2009-07-20 02:29:24 +00007745 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7746 function.</p>
7747
Bill Wendling69e4adb2008-11-19 05:56:17 +00007748</div>
7749
Eric Christopher0e671492009-11-30 08:03:53 +00007750<!-- _______________________________________________________________________ -->
7751<div class="doc_subsubsection">
7752 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7753</div>
7754
7755<div class="doc_text">
7756
7757<h5>Syntax:</h5>
7758<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +00007759 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;type&gt;)
7760 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;type&gt;)
Eric Christopher0e671492009-11-30 08:03:53 +00007761</pre>
7762
7763<h5>Overview:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +00007764<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
7765 the optimizers to determine at compile time whether a) an operation (like
7766 memcpy) will overflow a buffer that corresponds to an object, or b) that a
7767 runtime check for overflow isn't necessary. An object in this context means
7768 an allocation of a specific class, structure, array, or other object.</p>
Eric Christopher0e671492009-11-30 08:03:53 +00007769
7770<h5>Arguments:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +00007771<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher8295a0a2009-12-23 00:29:49 +00007772 argument is a pointer to or into the <tt>object</tt>. The second argument
Bill Wendling1b383ba2010-10-27 01:07:41 +00007773 is a boolean 0 or 1. This argument determines whether you want the
7774 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
Eric Christopher8295a0a2009-12-23 00:29:49 +00007775 1, variables are not allowed.</p>
7776
Eric Christopher0e671492009-11-30 08:03:53 +00007777<h5>Semantics:</h5>
7778<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Bill Wendling1b383ba2010-10-27 01:07:41 +00007779 representing the size of the object concerned, or <tt>i32/i64 -1 or 0</tt>,
7780 depending on the <tt>type</tt> argument, if the size cannot be determined at
7781 compile time.</p>
Eric Christopher0e671492009-11-30 08:03:53 +00007782
7783</div>
7784
Chris Lattner00950542001-06-06 20:29:01 +00007785<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +00007786<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007787<address>
7788 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +00007789 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007790 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +00007791 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007792
7793 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +00007794 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +00007795 Last modified: $Date$
7796</address>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +00007797
Misha Brukman9d0919f2003-11-08 01:05:38 +00007798</body>
7799</html>