blob: 80bf1c8b3429f326d31131276f553a4d788385bb [file] [log] [blame]
Misha Brukmanc501f552004-03-01 17:47:27 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman76307852003-11-08 01:05:38 +00003<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencercb84e432004-08-26 20:44:00 +00006 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
Eric Christopher455c5772009-12-05 02:46:03 +00008 <meta name="description"
Reid Spencercb84e432004-08-26 20:44:00 +00009 content="LLVM Assembly Language Reference Manual.">
Misha Brukman76307852003-11-08 01:05:38 +000010 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattner757528b0b2004-05-23 21:06:01 +000012
Misha Brukman76307852003-11-08 01:05:38 +000013<body>
Chris Lattner757528b0b2004-05-23 21:06:01 +000014
Chris Lattner48b383b02003-11-25 01:02:51 +000015<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner2f7c9632001-06-06 20:29:01 +000016<ol>
Misha Brukman76307852003-11-08 01:05:38 +000017 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000020 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Bill Wendlinga3c6f6b2009-07-20 01:03:30 +000023 <li><a href="#linkage">Linkage Types</a>
24 <ol>
Bill Wendling8693ef82009-07-20 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 Wendling03bcd6e2010-07-01 21:55:59 +000027 <li><a href="#linkage_linker_private_weak">'<tt>linker_private_weak</tt>' Linkage</a></li>
Bill Wendling8693ef82009-07-20 02:41:50 +000028 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
29 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
30 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
31 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
32 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
33 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
34 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner80d73c72009-10-10 18:26:06 +000035 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendling8693ef82009-07-20 02:41:50 +000036 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
37 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendlinga3c6f6b2009-07-20 01:03:30 +000040 </ol>
41 </li>
Chris Lattner0132aff2005-05-06 22:57:40 +000042 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnerbc088212009-01-11 20:53:49 +000043 <li><a href="#namedtypes">Named Types</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000044 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +000045 <li><a href="#functionstructure">Functions</a></li>
Dan Gohmanef9462f2008-10-14 16:51:45 +000046 <li><a href="#aliasstructure">Aliases</a></li>
Devang Pateld1a89692010-01-11 19:35:55 +000047 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +000048 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel9eb525d2008-09-26 23:51:19 +000049 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen71183b62007-12-10 03:18:06 +000050 <li><a href="#gc">Garbage Collector Names</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +000051 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencer50c723a2007-02-19 23:54:10 +000052 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman6154a012009-07-27 18:07:55 +000053 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +000054 <li><a href="#volatile">Volatile Memory Accesses</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000055 </ol>
56 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000057 <li><a href="#typesystem">Type System</a>
58 <ol>
Chris Lattner7824d182008-01-04 04:32:38 +000059 <li><a href="#t_classifications">Type Classifications</a></li>
Eric Christopher455c5772009-12-05 02:46:03 +000060 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner48b383b02003-11-25 01:02:51 +000061 <ol>
Nick Lewycky84a1eeb2009-09-27 00:45:11 +000062 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner7824d182008-01-04 04:32:38 +000063 <li><a href="#t_floating">Floating Point Types</a></li>
64 <li><a href="#t_void">Void Type</a></li>
65 <li><a href="#t_label">Label Type</a></li>
Nick Lewyckyadbc2842009-05-30 05:06:04 +000066 <li><a href="#t_metadata">Metadata Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000067 </ol>
68 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000069 <li><a href="#t_derived">Derived Types</a>
70 <ol>
Chris Lattner392be582010-02-12 20:49:41 +000071 <li><a href="#t_aggregate">Aggregate Types</a>
72 <ol>
73 <li><a href="#t_array">Array Type</a></li>
74 <li><a href="#t_struct">Structure Type</a></li>
75 <li><a href="#t_pstruct">Packed Structure Type</a></li>
76 <li><a href="#t_union">Union Type</a></li>
77 <li><a href="#t_vector">Vector Type</a></li>
78 </ol>
79 </li>
Misha Brukman76307852003-11-08 01:05:38 +000080 <li><a href="#t_function">Function Type</a></li>
81 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner37b6b092005-04-25 17:34:15 +000082 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000083 </ol>
84 </li>
Chris Lattnercf7a5842009-02-02 07:32:36 +000085 <li><a href="#t_uprefs">Type Up-references</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000086 </ol>
87 </li>
Chris Lattner6af02f32004-12-09 16:11:40 +000088 <li><a href="#constants">Constants</a>
Chris Lattner74d3f822004-12-09 17:30:23 +000089 <ol>
Dan Gohmanef9462f2008-10-14 16:51:45 +000090 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner361bfcd2009-02-28 18:32:25 +000091 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohmanef9462f2008-10-14 16:51:45 +000092 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
93 <li><a href="#undefvalues">Undefined Values</a></li>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +000094 <li><a href="#trapvalues">Trap Values</a></li>
Chris Lattner2bfd3202009-10-27 21:19:13 +000095 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
Dan Gohmanef9462f2008-10-14 16:51:45 +000096 <li><a href="#constantexprs">Constant Expressions</a></li>
Chris Lattner74d3f822004-12-09 17:30:23 +000097 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +000098 </li>
Chris Lattner98f013c2006-01-25 23:47:57 +000099 <li><a href="#othervalues">Other Values</a>
100 <ol>
Dan Gohmanef9462f2008-10-14 16:51:45 +0000101 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Devang Pateld1a89692010-01-11 19:35:55 +0000102 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
Chris Lattner98f013c2006-01-25 23:47:57 +0000103 </ol>
104 </li>
Chris Lattnerae76db52009-07-20 05:55:19 +0000105 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
106 <ol>
107 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner58f9bb22009-07-20 06:14:25 +0000108 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
109 Global Variable</a></li>
Chris Lattnerae76db52009-07-20 05:55:19 +0000110 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
111 Global Variable</a></li>
112 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
113 Global Variable</a></li>
114 </ol>
115 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000116 <li><a href="#instref">Instruction Reference</a>
117 <ol>
118 <li><a href="#terminators">Terminator Instructions</a>
119 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000120 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
121 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000122 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +0000123 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000124 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000125 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000126 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000127 </ol>
128 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000129 <li><a href="#binaryops">Binary Operations</a>
130 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000131 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmana5b96452009-06-04 22:49:04 +0000132 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000133 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmana5b96452009-06-04 22:49:04 +0000134 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000135 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmana5b96452009-06-04 22:49:04 +0000136 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000137 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
138 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
139 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000140 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
141 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
142 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000143 </ol>
144 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000145 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
146 <ol>
Reid Spencer2ab01932007-02-02 13:57:07 +0000147 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
148 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
149 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000150 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000151 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000152 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000153 </ol>
154 </li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000155 <li><a href="#vectorops">Vector Operations</a>
156 <ol>
157 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
158 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
159 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000160 </ol>
161 </li>
Dan Gohmanb9d66602008-05-12 23:51:09 +0000162 <li><a href="#aggregateops">Aggregate Operations</a>
163 <ol>
164 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
165 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
166 </ol>
167 </li>
Chris Lattner6ab66722006-08-15 00:45:58 +0000168 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000169 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000170 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino820bc75b2006-02-17 21:18:08 +0000171 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
172 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
173 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000174 </ol>
175 </li>
Reid Spencer97c5fa42006-11-08 01:18:52 +0000176 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000177 <ol>
178 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
179 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
180 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
181 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
182 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencer51b07252006-11-09 23:03:26 +0000183 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
184 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
185 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
186 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000187 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
188 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5b950642006-11-11 23:08:07 +0000189 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000190 </ol>
Dan Gohmanef9462f2008-10-14 16:51:45 +0000191 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000192 <li><a href="#otherops">Other Operations</a>
193 <ol>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000194 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
195 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000196 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000197 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000198 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattner33337472006-01-13 23:26:01 +0000199 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000200 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000201 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000202 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000203 </li>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000204 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000205 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000206 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
207 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000208 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
209 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
210 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000211 </ol>
212 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000213 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
214 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000215 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
216 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
217 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000218 </ol>
219 </li>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000220 <li><a href="#int_codegen">Code Generator Intrinsics</a>
221 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000222 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
223 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
224 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
225 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
226 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
227 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohmane58f7b32010-05-26 21:56:15 +0000228 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswellaa1c3c12004-04-09 16:43:20 +0000229 </ol>
230 </li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000231 <li><a href="#int_libc">Standard C Library Intrinsics</a>
232 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000233 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
234 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
236 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
237 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohmanb6324c12007-10-15 20:30:11 +0000238 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
239 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000241 </ol>
242 </li>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000243 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000244 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000245 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattnerb748c672006-01-16 22:34:14 +0000246 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
247 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
248 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000249 </ol>
250 </li>
Bill Wendlingf4d70622009-02-08 01:40:31 +0000251 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
252 <ol>
Bill Wendlingfd2bd722009-02-08 04:04:40 +0000253 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
254 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
255 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
256 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
257 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingb9a73272009-02-08 23:00:09 +0000258 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingf4d70622009-02-08 01:40:31 +0000259 </ol>
260 </li>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +0000261 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
262 <ol>
Chris Lattnerbbd8bd32010-03-14 23:03:31 +0000263 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
264 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +0000265 </ol>
266 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000267 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskey2211f492007-03-14 19:31:19 +0000268 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands86e01192007-09-11 14:10:23 +0000269 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands644f9172007-07-27 12:58:54 +0000270 <ol>
271 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands644f9172007-07-27 12:58:54 +0000272 </ol>
273 </li>
Bill Wendlingf85850f2008-11-18 22:10:53 +0000274 <li><a href="#int_atomics">Atomic intrinsics</a>
275 <ol>
276 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
277 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
278 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
279 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
280 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
281 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
282 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
283 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
284 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
285 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
286 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
287 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
288 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
289 </ol>
290 </li>
Nick Lewycky6f7d8342009-10-13 07:03:23 +0000291 <li><a href="#int_memorymarkers">Memory Use Markers</a>
292 <ol>
293 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
294 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
295 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
296 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
297 </ol>
298 </li>
Reid Spencer5b2cb0f2007-07-20 19:59:11 +0000299 <li><a href="#int_general">General intrinsics</a>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000300 <ol>
Reid Spencer5b2cb0f2007-07-20 19:59:11 +0000301 <li><a href="#int_var_annotation">
Bill Wendling14313312008-11-19 05:56:17 +0000302 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattner293c0372007-09-21 22:59:12 +0000303 <li><a href="#int_annotation">
Bill Wendling14313312008-11-19 05:56:17 +0000304 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +0000305 <li><a href="#int_trap">
Bill Wendling14313312008-11-19 05:56:17 +0000306 '<tt>llvm.trap</tt>' Intrinsic</a></li>
307 <li><a href="#int_stackprotector">
308 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher73484322009-11-30 08:03:53 +0000309 <li><a href="#int_objectsize">
310 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Tanya Lattner293c0372007-09-21 22:59:12 +0000311 </ol>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000312 </li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000313 </ol>
314 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000315</ol>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000316
317<div class="doc_author">
318 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
319 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman76307852003-11-08 01:05:38 +0000320</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000321
Chris Lattner2f7c9632001-06-06 20:29:01 +0000322<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000323<div class="doc_section"> <a name="abstract">Abstract </a></div>
324<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000325
Misha Brukman76307852003-11-08 01:05:38 +0000326<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000327
328<p>This document is a reference manual for the LLVM assembly language. LLVM is
329 a Static Single Assignment (SSA) based representation that provides type
330 safety, low-level operations, flexibility, and the capability of representing
331 'all' high-level languages cleanly. It is the common code representation
332 used throughout all phases of the LLVM compilation strategy.</p>
333
Misha Brukman76307852003-11-08 01:05:38 +0000334</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000335
Chris Lattner2f7c9632001-06-06 20:29:01 +0000336<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000337<div class="doc_section"> <a name="introduction">Introduction</a> </div>
338<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000339
Misha Brukman76307852003-11-08 01:05:38 +0000340<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000341
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000342<p>The LLVM code representation is designed to be used in three different forms:
343 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
344 for fast loading by a Just-In-Time compiler), and as a human readable
345 assembly language representation. This allows LLVM to provide a powerful
346 intermediate representation for efficient compiler transformations and
347 analysis, while providing a natural means to debug and visualize the
348 transformations. The three different forms of LLVM are all equivalent. This
349 document describes the human readable representation and notation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000350
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000351<p>The LLVM representation aims to be light-weight and low-level while being
352 expressive, typed, and extensible at the same time. It aims to be a
353 "universal IR" of sorts, by being at a low enough level that high-level ideas
354 may be cleanly mapped to it (similar to how microprocessors are "universal
355 IR's", allowing many source languages to be mapped to them). By providing
356 type information, LLVM can be used as the target of optimizations: for
357 example, through pointer analysis, it can be proven that a C automatic
Bill Wendling7f4a3362009-11-02 00:24:16 +0000358 variable is never accessed outside of the current function, allowing it to
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000359 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000360
Misha Brukman76307852003-11-08 01:05:38 +0000361</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000362
Chris Lattner2f7c9632001-06-06 20:29:01 +0000363<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000364<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000365
Misha Brukman76307852003-11-08 01:05:38 +0000366<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000367
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000368<p>It is important to note that this document describes 'well formed' LLVM
369 assembly language. There is a difference between what the parser accepts and
370 what is considered 'well formed'. For example, the following instruction is
371 syntactically okay, but not well formed:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000372
Bill Wendling3716c5d2007-05-29 09:04:49 +0000373<div class="doc_code">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000374<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000375%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattner757528b0b2004-05-23 21:06:01 +0000376</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000377</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000378
Bill Wendling7f4a3362009-11-02 00:24:16 +0000379<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
380 LLVM infrastructure provides a verification pass that may be used to verify
381 that an LLVM module is well formed. This pass is automatically run by the
382 parser after parsing input assembly and by the optimizer before it outputs
383 bitcode. The violations pointed out by the verifier pass indicate bugs in
384 transformation passes or input to the parser.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000385
Bill Wendling3716c5d2007-05-29 09:04:49 +0000386</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000387
Chris Lattner87a3dbe2007-10-03 17:34:29 +0000388<!-- Describe the typesetting conventions here. -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000389
Chris Lattner2f7c9632001-06-06 20:29:01 +0000390<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000391<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000392<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000393
Misha Brukman76307852003-11-08 01:05:38 +0000394<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000395
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000396<p>LLVM identifiers come in two basic types: global and local. Global
397 identifiers (functions, global variables) begin with the <tt>'@'</tt>
398 character. Local identifiers (register names, types) begin with
399 the <tt>'%'</tt> character. Additionally, there are three different formats
400 for identifiers, for different purposes:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000401
Chris Lattner2f7c9632001-06-06 20:29:01 +0000402<ol>
Reid Spencerb23b65f2007-08-07 14:34:28 +0000403 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000404 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
405 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
406 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
407 other characters in their names can be surrounded with quotes. Special
408 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
409 ASCII code for the character in hexadecimal. In this way, any character
410 can be used in a name value, even quotes themselves.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000411
Reid Spencerb23b65f2007-08-07 14:34:28 +0000412 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000413 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000414
Reid Spencer8f08d802004-12-09 18:02:53 +0000415 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000416 constants</a>, below.</li>
Misha Brukman76307852003-11-08 01:05:38 +0000417</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000418
Reid Spencerb23b65f2007-08-07 14:34:28 +0000419<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000420 don't need to worry about name clashes with reserved words, and the set of
421 reserved words may be expanded in the future without penalty. Additionally,
422 unnamed identifiers allow a compiler to quickly come up with a temporary
423 variable without having to avoid symbol table conflicts.</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000424
Chris Lattner48b383b02003-11-25 01:02:51 +0000425<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000426 languages. There are keywords for different opcodes
427 ('<tt><a href="#i_add">add</a></tt>',
428 '<tt><a href="#i_bitcast">bitcast</a></tt>',
429 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
430 ('<tt><a href="#t_void">void</a></tt>',
431 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
432 reserved words cannot conflict with variable names, because none of them
433 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000434
435<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000436 '<tt>%X</tt>' by 8:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000437
Misha Brukman76307852003-11-08 01:05:38 +0000438<p>The easy way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000439
Bill Wendling3716c5d2007-05-29 09:04:49 +0000440<div class="doc_code">
Chris Lattnerd79749a2004-12-09 16:36:40 +0000441<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000442%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnerd79749a2004-12-09 16:36:40 +0000443</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000444</div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000445
Misha Brukman76307852003-11-08 01:05:38 +0000446<p>After strength reduction:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000447
Bill Wendling3716c5d2007-05-29 09:04:49 +0000448<div class="doc_code">
Chris Lattnerd79749a2004-12-09 16:36:40 +0000449<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000450%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnerd79749a2004-12-09 16:36:40 +0000451</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000452</div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000453
Misha Brukman76307852003-11-08 01:05:38 +0000454<p>And the hard way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000455
Bill Wendling3716c5d2007-05-29 09:04:49 +0000456<div class="doc_code">
Chris Lattnerd79749a2004-12-09 16:36:40 +0000457<pre>
Gabor Greifbd0328f2009-10-28 13:05:07 +0000458%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
459%1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000460%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnerd79749a2004-12-09 16:36:40 +0000461</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000462</div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000463
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000464<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
465 lexical features of LLVM:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000466
Chris Lattner2f7c9632001-06-06 20:29:01 +0000467<ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000468 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000469 line.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000470
471 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000472 assigned to a named value.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000473
Misha Brukman76307852003-11-08 01:05:38 +0000474 <li>Unnamed temporaries are numbered sequentially</li>
475</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000476
Bill Wendling7f4a3362009-11-02 00:24:16 +0000477<p>It also shows a convention that we follow in this document. When
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000478 demonstrating instructions, we will follow an instruction with a comment that
479 defines the type and name of value produced. Comments are shown in italic
480 text.</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000481
Misha Brukman76307852003-11-08 01:05:38 +0000482</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000483
484<!-- *********************************************************************** -->
485<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
486<!-- *********************************************************************** -->
487
488<!-- ======================================================================= -->
489<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
490</div>
491
492<div class="doc_text">
493
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000494<p>LLVM programs are composed of "Module"s, each of which is a translation unit
495 of the input programs. Each module consists of functions, global variables,
496 and symbol table entries. Modules may be combined together with the LLVM
497 linker, which merges function (and global variable) definitions, resolves
498 forward declarations, and merges symbol table entries. Here is an example of
499 the "hello world" module:</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000500
Bill Wendling3716c5d2007-05-29 09:04:49 +0000501<div class="doc_code">
Bill Wendling7f4a3362009-11-02 00:24:16 +0000502<pre>
503<i>; Declare the string constant as a global constant.</i>
504<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000505
506<i>; External declaration of the puts function</i>
Dan Gohmanaabfdb32010-05-28 17:13:49 +0000507<a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000508
509<i>; Definition of main function</i>
Bill Wendling7f4a3362009-11-02 00:24:16 +0000510define i32 @main() { <i>; i32()* </i>
511 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanaabfdb32010-05-28 17:13:49 +0000512 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000513
Bill Wendling7f4a3362009-11-02 00:24:16 +0000514 <i>; Call puts function to write out the string to stdout.</i>
Dan Gohmanaabfdb32010-05-28 17:13:49 +0000515 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
Devang Pateld1a89692010-01-11 19:35:55 +0000516 <a href="#i_ret">ret</a> i32 0<br>}
517
518<i>; Named metadata</i>
519!1 = metadata !{i32 41}
520!foo = !{!1, null}
Bill Wendling3716c5d2007-05-29 09:04:49 +0000521</pre>
522</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000523
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000524<p>This example is made up of a <a href="#globalvars">global variable</a> named
Devang Pateld1a89692010-01-11 19:35:55 +0000525 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000526 a <a href="#functionstructure">function definition</a> for
Devang Pateld1a89692010-01-11 19:35:55 +0000527 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
528 "<tt>foo"</tt>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000529
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000530<p>In general, a module is made up of a list of global values, where both
531 functions and global variables are global values. Global values are
532 represented by a pointer to a memory location (in this case, a pointer to an
533 array of char, and a pointer to a function), and have one of the
534 following <a href="#linkage">linkage types</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000535
Chris Lattnerd79749a2004-12-09 16:36:40 +0000536</div>
537
538<!-- ======================================================================= -->
539<div class="doc_subsection">
540 <a name="linkage">Linkage Types</a>
541</div>
542
543<div class="doc_text">
544
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000545<p>All Global Variables and Functions have one of the following types of
546 linkage:</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000547
548<dl>
Bill Wendling7f4a3362009-11-02 00:24:16 +0000549 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling03bcd6e2010-07-01 21:55:59 +0000550 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
551 by objects in the current module. In particular, linking code into a
552 module with an private global value may cause the private to be renamed as
553 necessary to avoid collisions. Because the symbol is private to the
554 module, all references can be updated. This doesn't show up in any symbol
555 table in the object file.</dd>
Rafael Espindola6de96a12009-01-15 20:18:42 +0000556
Bill Wendling7f4a3362009-11-02 00:24:16 +0000557 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling03bcd6e2010-07-01 21:55:59 +0000558 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
559 assembler and evaluated by the linker. Unlike normal strong symbols, they
560 are removed by the linker from the final linked image (executable or
561 dynamic library).</dd>
562
563 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
564 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
565 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
566 linker. The symbols are removed by the linker from the final linked image
567 (executable or dynamic library).</dd>
Bill Wendlinga3c6f6b2009-07-20 01:03:30 +0000568
Bill Wendling7f4a3362009-11-02 00:24:16 +0000569 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling36321712010-06-29 22:34:52 +0000570 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000571 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
572 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000573
Bill Wendling7f4a3362009-11-02 00:24:16 +0000574 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner184f1be2009-04-13 05:44:34 +0000575 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000576 into the object file corresponding to the LLVM module. They exist to
577 allow inlining and other optimizations to take place given knowledge of
578 the definition of the global, which is known to be somewhere outside the
579 module. Globals with <tt>available_externally</tt> linkage are allowed to
580 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
581 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner184f1be2009-04-13 05:44:34 +0000582
Bill Wendling7f4a3362009-11-02 00:24:16 +0000583 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattnere20b4702007-01-14 06:51:48 +0000584 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner0de4caa2010-01-09 19:15:14 +0000585 the same name when linkage occurs. This can be used to implement
586 some forms of inline functions, templates, or other code which must be
587 generated in each translation unit that uses it, but where the body may
588 be overridden with a more definitive definition later. Unreferenced
589 <tt>linkonce</tt> globals are allowed to be discarded. Note that
590 <tt>linkonce</tt> linkage does not actually allow the optimizer to
591 inline the body of this function into callers because it doesn't know if
592 this definition of the function is the definitive definition within the
593 program or whether it will be overridden by a stronger definition.
594 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
595 linkage.</dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000596
Bill Wendling7f4a3362009-11-02 00:24:16 +0000597 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattnerd0554882009-08-05 05:21:07 +0000598 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
599 <tt>linkonce</tt> linkage, except that unreferenced globals with
600 <tt>weak</tt> linkage may not be discarded. This is used for globals that
601 are declared "weak" in C source code.</dd>
602
Bill Wendling7f4a3362009-11-02 00:24:16 +0000603 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattnerd0554882009-08-05 05:21:07 +0000604 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
605 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
606 global scope.
607 Symbols with "<tt>common</tt>" linkage are merged in the same way as
608 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattner0aff0b22009-08-05 05:41:44 +0000609 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher455c5772009-12-05 02:46:03 +0000610 must have a zero initializer, and may not be marked '<a
Chris Lattner0aff0b22009-08-05 05:41:44 +0000611 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
612 have common linkage.</dd>
Chris Lattnerd0554882009-08-05 05:21:07 +0000613
Chris Lattnerd79749a2004-12-09 16:36:40 +0000614
Bill Wendling7f4a3362009-11-02 00:24:16 +0000615 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000616 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000617 pointer to array type. When two global variables with appending linkage
618 are linked together, the two global arrays are appended together. This is
619 the LLVM, typesafe, equivalent of having the system linker append together
620 "sections" with identical names when .o files are linked.</dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000621
Bill Wendling7f4a3362009-11-02 00:24:16 +0000622 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000623 <dd>The semantics of this linkage follow the ELF object file model: the symbol
624 is weak until linked, if not linked, the symbol becomes null instead of
625 being an undefined reference.</dd>
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000626
Bill Wendling7f4a3362009-11-02 00:24:16 +0000627 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
628 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000629 <dd>Some languages allow differing globals to be merged, such as two functions
630 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling03bcd6e2010-07-01 21:55:59 +0000631 that only equivalent globals are ever merged (the "one definition rule"
632 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000633 and <tt>weak_odr</tt> linkage types to indicate that the global will only
634 be merged with equivalent globals. These linkage types are otherwise the
635 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands12da8ce2009-03-07 15:45:40 +0000636
Chris Lattner6af02f32004-12-09 16:11:40 +0000637 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000638 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000639 visible, meaning that it participates in linkage and can be used to
640 resolve external symbol references.</dd>
Reid Spencer7972c472007-04-11 23:49:50 +0000641</dl>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000642
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000643<p>The next two types of linkage are targeted for Microsoft Windows platform
644 only. They are designed to support importing (exporting) symbols from (to)
645 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000646
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000647<dl>
Bill Wendling7f4a3362009-11-02 00:24:16 +0000648 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000649 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000650 or variable via a global pointer to a pointer that is set up by the DLL
651 exporting the symbol. On Microsoft Windows targets, the pointer name is
652 formed by combining <code>__imp_</code> and the function or variable
653 name.</dd>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000654
Bill Wendling7f4a3362009-11-02 00:24:16 +0000655 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000656 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000657 pointer to a pointer in a DLL, so that it can be referenced with the
658 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
659 name is formed by combining <code>__imp_</code> and the function or
660 variable name.</dd>
Chris Lattner6af02f32004-12-09 16:11:40 +0000661</dl>
662
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000663<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
664 another module defined a "<tt>.LC0</tt>" variable and was linked with this
665 one, one of the two would be renamed, preventing a collision. Since
666 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
667 declarations), they are accessible outside of the current module.</p>
668
669<p>It is illegal for a function <i>declaration</i> to have any linkage type
670 other than "externally visible", <tt>dllimport</tt>
671 or <tt>extern_weak</tt>.</p>
672
Duncan Sands12da8ce2009-03-07 15:45:40 +0000673<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000674 or <tt>weak_odr</tt> linkages.</p>
675
Chris Lattner6af02f32004-12-09 16:11:40 +0000676</div>
677
678<!-- ======================================================================= -->
679<div class="doc_subsection">
Chris Lattner0132aff2005-05-06 22:57:40 +0000680 <a name="callingconv">Calling Conventions</a>
681</div>
682
683<div class="doc_text">
684
685<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000686 and <a href="#i_invoke">invokes</a> can all have an optional calling
687 convention specified for the call. The calling convention of any pair of
688 dynamic caller/callee must match, or the behavior of the program is
689 undefined. The following calling conventions are supported by LLVM, and more
690 may be added in the future:</p>
Chris Lattner0132aff2005-05-06 22:57:40 +0000691
692<dl>
693 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000694 <dd>This calling convention (the default if no other calling convention is
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000695 specified) matches the target C calling conventions. This calling
696 convention supports varargs function calls and tolerates some mismatch in
697 the declared prototype and implemented declaration of the function (as
698 does normal C).</dd>
Chris Lattner0132aff2005-05-06 22:57:40 +0000699
700 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000701 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000702 (e.g. by passing things in registers). This calling convention allows the
703 target to use whatever tricks it wants to produce fast code for the
704 target, without having to conform to an externally specified ABI
Jeffrey Yasskinb8677462010-01-09 19:44:16 +0000705 (Application Binary Interface).
706 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattnera179e4d2010-03-11 00:22:57 +0000707 when this or the GHC convention is used.</a> This calling convention
708 does not support varargs and requires the prototype of all callees to
709 exactly match the prototype of the function definition.</dd>
Chris Lattner0132aff2005-05-06 22:57:40 +0000710
711 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000712 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000713 as possible under the assumption that the call is not commonly executed.
714 As such, these calls often preserve all registers so that the call does
715 not break any live ranges in the caller side. This calling convention
716 does not support varargs and requires the prototype of all callees to
717 exactly match the prototype of the function definition.</dd>
Chris Lattner0132aff2005-05-06 22:57:40 +0000718
Chris Lattnera179e4d2010-03-11 00:22:57 +0000719 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
720 <dd>This calling convention has been implemented specifically for use by the
721 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
722 It passes everything in registers, going to extremes to achieve this by
723 disabling callee save registers. This calling convention should not be
724 used lightly but only for specific situations such as an alternative to
725 the <em>register pinning</em> performance technique often used when
726 implementing functional programming languages.At the moment only X86
727 supports this convention and it has the following limitations:
728 <ul>
729 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
730 floating point types are supported.</li>
731 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
732 6 floating point parameters.</li>
733 </ul>
734 This calling convention supports
735 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
736 requires both the caller and callee are using it.
737 </dd>
738
Chris Lattner573f64e2005-05-07 01:46:40 +0000739 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000740 <dd>Any calling convention may be specified by number, allowing
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000741 target-specific calling conventions to be used. Target specific calling
742 conventions start at 64.</dd>
Chris Lattner573f64e2005-05-07 01:46:40 +0000743</dl>
Chris Lattner0132aff2005-05-06 22:57:40 +0000744
745<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000746 support Pascal conventions or any other well-known target-independent
747 convention.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +0000748
749</div>
750
751<!-- ======================================================================= -->
752<div class="doc_subsection">
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000753 <a name="visibility">Visibility Styles</a>
754</div>
755
756<div class="doc_text">
757
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000758<p>All Global Variables and Functions have one of the following visibility
759 styles:</p>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000760
761<dl>
762 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattner67c37d12008-08-05 18:29:16 +0000763 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000764 that the declaration is visible to other modules and, in shared libraries,
765 means that the declared entity may be overridden. On Darwin, default
766 visibility means that the declaration is visible to other modules. Default
767 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000768
769 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000770 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000771 object if they are in the same shared object. Usually, hidden visibility
772 indicates that the symbol will not be placed into the dynamic symbol
773 table, so no other module (executable or shared library) can reference it
774 directly.</dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000775
Anton Korobeynikov39f3cff2007-04-29 18:35:00 +0000776 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov39f3cff2007-04-29 18:35:00 +0000777 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000778 the dynamic symbol table, but that references within the defining module
779 will bind to the local symbol. That is, the symbol cannot be overridden by
780 another module.</dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000781</dl>
782
783</div>
784
785<!-- ======================================================================= -->
786<div class="doc_subsection">
Chris Lattnerbc088212009-01-11 20:53:49 +0000787 <a name="namedtypes">Named Types</a>
788</div>
789
790<div class="doc_text">
791
792<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000793 it easier to read the IR and make the IR more condensed (particularly when
794 recursive types are involved). An example of a name specification is:</p>
Chris Lattnerbc088212009-01-11 20:53:49 +0000795
796<div class="doc_code">
797<pre>
798%mytype = type { %mytype*, i32 }
799</pre>
800</div>
801
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000802<p>You may give a name to any <a href="#typesystem">type</a> except
803 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
804 is expected with the syntax "%mytype".</p>
Chris Lattnerbc088212009-01-11 20:53:49 +0000805
806<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000807 and that you can therefore specify multiple names for the same type. This
808 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
809 uses structural typing, the name is not part of the type. When printing out
810 LLVM IR, the printer will pick <em>one name</em> to render all types of a
811 particular shape. This means that if you have code where two different
812 source types end up having the same LLVM type, that the dumper will sometimes
813 print the "wrong" or unexpected type. This is an important design point and
814 isn't going to change.</p>
Chris Lattnerbc088212009-01-11 20:53:49 +0000815
816</div>
817
Chris Lattnerbc088212009-01-11 20:53:49 +0000818<!-- ======================================================================= -->
819<div class="doc_subsection">
Chris Lattner6af02f32004-12-09 16:11:40 +0000820 <a name="globalvars">Global Variables</a>
821</div>
822
823<div class="doc_text">
824
Chris Lattner5d5aede2005-02-12 19:30:21 +0000825<p>Global variables define regions of memory allocated at compilation time
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000826 instead of run-time. Global variables may optionally be initialized, may
827 have an explicit section to be placed in, and may have an optional explicit
828 alignment specified. A variable may be defined as "thread_local", which
829 means that it will not be shared by threads (each thread will have a
830 separated copy of the variable). A variable may be defined as a global
831 "constant," which indicates that the contents of the variable
832 will <b>never</b> be modified (enabling better optimization, allowing the
833 global data to be placed in the read-only section of an executable, etc).
834 Note that variables that need runtime initialization cannot be marked
835 "constant" as there is a store to the variable.</p>
Chris Lattner5d5aede2005-02-12 19:30:21 +0000836
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000837<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
838 constant, even if the final definition of the global is not. This capability
839 can be used to enable slightly better optimization of the program, but
840 requires the language definition to guarantee that optimizations based on the
841 'constantness' are valid for the translation units that do not include the
842 definition.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000843
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000844<p>As SSA values, global variables define pointer values that are in scope
845 (i.e. they dominate) all basic blocks in the program. Global variables
846 always define a pointer to their "content" type because they describe a
847 region of memory, and all memory objects in LLVM are accessed through
848 pointers.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000849
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000850<p>A global variable may be declared to reside in a target-specific numbered
851 address space. For targets that support them, address spaces may affect how
852 optimizations are performed and/or what target instructions are used to
853 access the variable. The default address space is zero. The address space
854 qualifier must precede any other attributes.</p>
Christopher Lamb308121c2007-12-11 09:31:00 +0000855
Chris Lattner662c8722005-11-12 00:45:07 +0000856<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000857 supports it, it will emit globals to the section specified.</p>
Chris Lattner662c8722005-11-12 00:45:07 +0000858
Chris Lattner78e00bc2010-04-28 00:13:42 +0000859<p>An explicit alignment may be specified for a global, which must be a power
860 of 2. If not present, or if the alignment is set to zero, the alignment of
861 the global is set by the target to whatever it feels convenient. If an
862 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner4bd85e42010-04-28 00:31:12 +0000863 alignment. Targets and optimizers are not allowed to over-align the global
864 if the global has an assigned section. In this case, the extra alignment
865 could be observable: for example, code could assume that the globals are
866 densely packed in their section and try to iterate over them as an array,
867 alignment padding would break this iteration.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000868
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000869<p>For example, the following defines a global in a numbered address space with
870 an initializer, section, and alignment:</p>
Chris Lattner5760c502007-01-14 00:27:09 +0000871
Bill Wendling3716c5d2007-05-29 09:04:49 +0000872<div class="doc_code">
Chris Lattner5760c502007-01-14 00:27:09 +0000873<pre>
Dan Gohmanaaa679b2009-01-11 00:40:00 +0000874@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner5760c502007-01-14 00:27:09 +0000875</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000876</div>
Chris Lattner5760c502007-01-14 00:27:09 +0000877
Chris Lattner6af02f32004-12-09 16:11:40 +0000878</div>
879
880
881<!-- ======================================================================= -->
882<div class="doc_subsection">
883 <a name="functionstructure">Functions</a>
884</div>
885
886<div class="doc_text">
887
Dan Gohmana269a0a2010-03-01 17:41:39 +0000888<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000889 optional <a href="#linkage">linkage type</a>, an optional
890 <a href="#visibility">visibility style</a>, an optional
891 <a href="#callingconv">calling convention</a>, a return type, an optional
892 <a href="#paramattrs">parameter attribute</a> for the return type, a function
893 name, a (possibly empty) argument list (each with optional
894 <a href="#paramattrs">parameter attributes</a>), optional
895 <a href="#fnattrs">function attributes</a>, an optional section, an optional
896 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
897 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000898
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000899<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
900 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher455c5772009-12-05 02:46:03 +0000901 <a href="#visibility">visibility style</a>, an optional
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000902 <a href="#callingconv">calling convention</a>, a return type, an optional
903 <a href="#paramattrs">parameter attribute</a> for the return type, a function
904 name, a possibly empty list of arguments, an optional alignment, and an
905 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000906
Chris Lattner67c37d12008-08-05 18:29:16 +0000907<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000908 (Control Flow Graph) for the function. Each basic block may optionally start
909 with a label (giving the basic block a symbol table entry), contains a list
910 of instructions, and ends with a <a href="#terminators">terminator</a>
911 instruction (such as a branch or function return).</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000912
Chris Lattnera59fb102007-06-08 16:52:14 +0000913<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000914 executed on entrance to the function, and it is not allowed to have
915 predecessor basic blocks (i.e. there can not be any branches to the entry
916 block of a function). Because the block can have no predecessors, it also
917 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000918
Chris Lattner662c8722005-11-12 00:45:07 +0000919<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000920 supports it, it will emit functions to the section specified.</p>
Chris Lattner662c8722005-11-12 00:45:07 +0000921
Chris Lattner54611b42005-11-06 08:02:57 +0000922<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000923 the alignment is set to zero, the alignment of the function is set by the
924 target to whatever it feels convenient. If an explicit alignment is
925 specified, the function is forced to have at least that much alignment. All
926 alignments must be a power of 2.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000927
Bill Wendling30235112009-07-20 02:39:26 +0000928<h5>Syntax:</h5>
Devang Patel02256232008-10-07 17:48:33 +0000929<div class="doc_code">
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000930<pre>
Chris Lattner0ae02092008-10-13 16:55:18 +0000931define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000932 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
933 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
934 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
935 [<a href="#gc">gc</a>] { ... }
936</pre>
Devang Patel02256232008-10-07 17:48:33 +0000937</div>
938
Chris Lattner6af02f32004-12-09 16:11:40 +0000939</div>
940
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000941<!-- ======================================================================= -->
942<div class="doc_subsection">
943 <a name="aliasstructure">Aliases</a>
944</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000945
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000946<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000947
948<p>Aliases act as "second name" for the aliasee value (which can be either
949 function, global variable, another alias or bitcast of global value). Aliases
950 may have an optional <a href="#linkage">linkage type</a>, and an
951 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000952
Bill Wendling30235112009-07-20 02:39:26 +0000953<h5>Syntax:</h5>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000954<div class="doc_code">
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000955<pre>
Duncan Sands7e99a942008-09-12 20:48:21 +0000956@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000957</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000958</div>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000959
960</div>
961
Chris Lattner91c15c42006-01-23 23:23:47 +0000962<!-- ======================================================================= -->
Devang Pateld1a89692010-01-11 19:35:55 +0000963<div class="doc_subsection">
964 <a name="namedmetadatastructure">Named Metadata</a>
965</div>
966
967<div class="doc_text">
968
Chris Lattnerc2f8f162010-01-15 21:50:19 +0000969<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
970 nodes</a> (but not metadata strings) and null are the only valid operands for
971 a named metadata.</p>
Devang Pateld1a89692010-01-11 19:35:55 +0000972
973<h5>Syntax:</h5>
974<div class="doc_code">
975<pre>
976!1 = metadata !{metadata !"one"}
977!name = !{null, !1}
978</pre>
979</div>
980
981</div>
982
983<!-- ======================================================================= -->
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000984<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000985
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000986<div class="doc_text">
987
988<p>The return type and each parameter of a function type may have a set of
989 <i>parameter attributes</i> associated with them. Parameter attributes are
990 used to communicate additional information about the result or parameters of
991 a function. Parameter attributes are considered to be part of the function,
992 not of the function type, so functions with different parameter attributes
993 can have the same function type.</p>
994
995<p>Parameter attributes are simple keywords that follow the type specified. If
996 multiple parameter attributes are needed, they are space separated. For
997 example:</p>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000998
999<div class="doc_code">
1000<pre>
Nick Lewyckydac78d82009-02-15 23:06:14 +00001001declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerd2597d72008-10-04 18:33:34 +00001002declare i32 @atoi(i8 zeroext)
1003declare signext i8 @returns_signed_char()
Bill Wendling3716c5d2007-05-29 09:04:49 +00001004</pre>
1005</div>
1006
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001007<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1008 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001009
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001010<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001011
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001012<dl>
Bill Wendling7f4a3362009-11-02 00:24:16 +00001013 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001014 <dd>This indicates to the code generator that the parameter or return value
1015 should be zero-extended to a 32-bit value by the caller (for a parameter)
1016 or the callee (for a return value).</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001017
Bill Wendling7f4a3362009-11-02 00:24:16 +00001018 <dt><tt><b>signext</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001019 <dd>This indicates to the code generator that the parameter or return value
1020 should be sign-extended to a 32-bit value by the caller (for a parameter)
1021 or the callee (for a return value).</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001022
Bill Wendling7f4a3362009-11-02 00:24:16 +00001023 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001024 <dd>This indicates that this parameter or return value should be treated in a
1025 special target-dependent fashion during while emitting code for a function
1026 call or return (usually, by putting it in a register as opposed to memory,
1027 though some targets use it to distinguish between two different kinds of
1028 registers). Use of this attribute is target-specific.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001029
Bill Wendling7f4a3362009-11-02 00:24:16 +00001030 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001031 <dd>This indicates that the pointer parameter should really be passed by value
1032 to the function. The attribute implies that a hidden copy of the pointee
1033 is made between the caller and the callee, so the callee is unable to
1034 modify the value in the callee. This attribute is only valid on LLVM
1035 pointer arguments. It is generally used to pass structs and arrays by
1036 value, but is also valid on pointers to scalars. The copy is considered
1037 to belong to the caller not the callee (for example,
1038 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1039 <tt>byval</tt> parameters). This is not a valid attribute for return
1040 values. The byval attribute also supports specifying an alignment with
1041 the align attribute. This has a target-specific effect on the code
1042 generator that usually indicates a desired alignment for the synthesized
1043 stack slot.</dd>
1044
Dan Gohman3770af52010-07-02 23:18:08 +00001045 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001046 <dd>This indicates that the pointer parameter specifies the address of a
1047 structure that is the return value of the function in the source program.
1048 This pointer must be guaranteed by the caller to be valid: loads and
1049 stores to the structure may be assumed by the callee to not to trap. This
1050 may only be applied to the first parameter. This is not a valid attribute
1051 for return values. </dd>
1052
Dan Gohman3770af52010-07-02 23:18:08 +00001053 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohmandf12d082010-07-02 18:41:32 +00001054 <dd>This indicates that pointer values
1055 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmande256292010-07-02 23:46:54 +00001056 value do not alias pointer values which are not <i>based</i> on it,
1057 ignoring certain "irrelevant" dependencies.
1058 For a call to the parent function, dependencies between memory
1059 references from before or after the call and from those during the call
1060 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1061 return value used in that call.
Dan Gohmandf12d082010-07-02 18:41:32 +00001062 The caller shares the responsibility with the callee for ensuring that
1063 these requirements are met.
1064 For further details, please see the discussion of the NoAlias response in
Dan Gohman6c858db2010-07-06 15:26:33 +00001065 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1066<br>
John McCall72ed8902010-07-06 21:07:14 +00001067 Note that this definition of <tt>noalias</tt> is intentionally
1068 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner5eff9ca2010-07-06 20:51:35 +00001069 arguments, though it is slightly weaker.
Dan Gohman6c858db2010-07-06 15:26:33 +00001070<br>
1071 For function return values, C99's <tt>restrict</tt> is not meaningful,
1072 while LLVM's <tt>noalias</tt> is.
1073 </dd>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001074
Dan Gohman3770af52010-07-02 23:18:08 +00001075 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001076 <dd>This indicates that the callee does not make any copies of the pointer
1077 that outlive the callee itself. This is not a valid attribute for return
1078 values.</dd>
1079
Dan Gohman3770af52010-07-02 23:18:08 +00001080 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001081 <dd>This indicates that the pointer parameter can be excised using the
1082 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1083 attribute for return values.</dd>
1084</dl>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001085
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001086</div>
1087
1088<!-- ======================================================================= -->
Chris Lattner91c15c42006-01-23 23:23:47 +00001089<div class="doc_subsection">
Gordon Henriksen71183b62007-12-10 03:18:06 +00001090 <a name="gc">Garbage Collector Names</a>
1091</div>
1092
1093<div class="doc_text">
Gordon Henriksen71183b62007-12-10 03:18:06 +00001094
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001095<p>Each function may specify a garbage collector name, which is simply a
1096 string:</p>
1097
1098<div class="doc_code">
1099<pre>
Bill Wendling7f4a3362009-11-02 00:24:16 +00001100define void @f() gc "name" { ... }
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001101</pre>
1102</div>
Gordon Henriksen71183b62007-12-10 03:18:06 +00001103
1104<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001105 collector which will cause the compiler to alter its output in order to
1106 support the named garbage collection algorithm.</p>
1107
Gordon Henriksen71183b62007-12-10 03:18:06 +00001108</div>
1109
1110<!-- ======================================================================= -->
1111<div class="doc_subsection">
Devang Patel9eb525d2008-09-26 23:51:19 +00001112 <a name="fnattrs">Function Attributes</a>
Devang Patelcaacdba2008-09-04 23:05:13 +00001113</div>
1114
1115<div class="doc_text">
Devang Patel9eb525d2008-09-26 23:51:19 +00001116
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001117<p>Function attributes are set to communicate additional information about a
1118 function. Function attributes are considered to be part of the function, not
1119 of the function type, so functions with different parameter attributes can
1120 have the same function type.</p>
Devang Patel9eb525d2008-09-26 23:51:19 +00001121
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001122<p>Function attributes are simple keywords that follow the type specified. If
1123 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelcaacdba2008-09-04 23:05:13 +00001124
1125<div class="doc_code">
Bill Wendlingb175fa42008-09-07 10:26:33 +00001126<pre>
Devang Patel9eb525d2008-09-26 23:51:19 +00001127define void @f() noinline { ... }
1128define void @f() alwaysinline { ... }
1129define void @f() alwaysinline optsize { ... }
Bill Wendling7f4a3362009-11-02 00:24:16 +00001130define void @f() optsize { ... }
Bill Wendlingb175fa42008-09-07 10:26:33 +00001131</pre>
Devang Patelcaacdba2008-09-04 23:05:13 +00001132</div>
1133
Bill Wendlingb175fa42008-09-07 10:26:33 +00001134<dl>
Charles Davisbe5557e2010-02-12 00:31:15 +00001135 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1136 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1137 the backend should forcibly align the stack pointer. Specify the
1138 desired alignment, which must be a power of two, in parentheses.
1139
Bill Wendling7f4a3362009-11-02 00:24:16 +00001140 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001141 <dd>This attribute indicates that the inliner should attempt to inline this
1142 function into callers whenever possible, ignoring any active inlining size
1143 threshold for this caller.</dd>
Bill Wendlingb175fa42008-09-07 10:26:33 +00001144
Jakob Stoklund Olesen74bb06c2010-02-06 01:16:28 +00001145 <dt><tt><b>inlinehint</b></tt></dt>
1146 <dd>This attribute indicates that the source code contained a hint that inlining
1147 this function is desirable (such as the "inline" keyword in C/C++). It
1148 is just a hint; it imposes no requirements on the inliner.</dd>
1149
Nick Lewycky14b58da2010-07-06 18:24:09 +00001150 <dt><tt><b>naked</b></tt></dt>
1151 <dd>This attribute disables prologue / epilogue emission for the function.
1152 This can have very system-specific consequences.</dd>
1153
1154 <dt><tt><b>noimplicitfloat</b></tt></dt>
1155 <dd>This attributes disables implicit floating point instructions.</dd>
1156
Bill Wendling7f4a3362009-11-02 00:24:16 +00001157 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001158 <dd>This attribute indicates that the inliner should never inline this
1159 function in any situation. This attribute may not be used together with
1160 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel9eb525d2008-09-26 23:51:19 +00001161
Nick Lewycky14b58da2010-07-06 18:24:09 +00001162 <dt><tt><b>noredzone</b></tt></dt>
1163 <dd>This attribute indicates that the code generator should not use a red
1164 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel9eb525d2008-09-26 23:51:19 +00001165
Bill Wendling7f4a3362009-11-02 00:24:16 +00001166 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001167 <dd>This function attribute indicates that the function never returns
1168 normally. This produces undefined behavior at runtime if the function
1169 ever does dynamically return.</dd>
Bill Wendlinga8130172008-11-13 01:02:51 +00001170
Bill Wendling7f4a3362009-11-02 00:24:16 +00001171 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001172 <dd>This function attribute indicates that the function never returns with an
1173 unwind or exceptional control flow. If the function does unwind, its
1174 runtime behavior is undefined.</dd>
Bill Wendling0f5541e2008-11-26 19:07:40 +00001175
Nick Lewycky14b58da2010-07-06 18:24:09 +00001176 <dt><tt><b>optsize</b></tt></dt>
1177 <dd>This attribute suggests that optimization passes and code generator passes
1178 make choices that keep the code size of this function low, and otherwise
1179 do optimizations specifically to reduce code size.</dd>
1180
Bill Wendling7f4a3362009-11-02 00:24:16 +00001181 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001182 <dd>This attribute indicates that the function computes its result (or decides
1183 to unwind an exception) based strictly on its arguments, without
1184 dereferencing any pointer arguments or otherwise accessing any mutable
1185 state (e.g. memory, control registers, etc) visible to caller functions.
1186 It does not write through any pointer arguments
1187 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1188 changes any state visible to callers. This means that it cannot unwind
1189 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1190 could use the <tt>unwind</tt> instruction.</dd>
Devang Patel310fd4a2009-06-12 19:45:19 +00001191
Bill Wendling7f4a3362009-11-02 00:24:16 +00001192 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001193 <dd>This attribute indicates that the function does not write through any
1194 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1195 arguments) or otherwise modify any state (e.g. memory, control registers,
1196 etc) visible to caller functions. It may dereference pointer arguments
1197 and read state that may be set in the caller. A readonly function always
1198 returns the same value (or unwinds an exception identically) when called
1199 with the same set of arguments and global state. It cannot unwind an
1200 exception by calling the <tt>C++</tt> exception throwing methods, but may
1201 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovc8ce7b082009-07-17 18:07:26 +00001202
Bill Wendling7f4a3362009-11-02 00:24:16 +00001203 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001204 <dd>This attribute indicates that the function should emit a stack smashing
1205 protector. It is in the form of a "canary"&mdash;a random value placed on
1206 the stack before the local variables that's checked upon return from the
1207 function to see if it has been overwritten. A heuristic is used to
1208 determine if a function needs stack protectors or not.<br>
1209<br>
1210 If a function that has an <tt>ssp</tt> attribute is inlined into a
1211 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1212 function will have an <tt>ssp</tt> attribute.</dd>
1213
Bill Wendling7f4a3362009-11-02 00:24:16 +00001214 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001215 <dd>This attribute indicates that the function should <em>always</em> emit a
1216 stack smashing protector. This overrides
Bill Wendling30235112009-07-20 02:39:26 +00001217 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1218<br>
1219 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1220 function that doesn't have an <tt>sspreq</tt> attribute or which has
1221 an <tt>ssp</tt> attribute, then the resulting function will have
1222 an <tt>sspreq</tt> attribute.</dd>
Bill Wendlingb175fa42008-09-07 10:26:33 +00001223</dl>
1224
Devang Patelcaacdba2008-09-04 23:05:13 +00001225</div>
1226
1227<!-- ======================================================================= -->
1228<div class="doc_subsection">
Chris Lattner93564892006-04-08 04:40:53 +00001229 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner91c15c42006-01-23 23:23:47 +00001230</div>
1231
1232<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001233
1234<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1235 the GCC "file scope inline asm" blocks. These blocks are internally
1236 concatenated by LLVM and treated as a single unit, but may be separated in
1237 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner91c15c42006-01-23 23:23:47 +00001238
Bill Wendling3716c5d2007-05-29 09:04:49 +00001239<div class="doc_code">
1240<pre>
1241module asm "inline asm code goes here"
1242module asm "more can go here"
1243</pre>
1244</div>
Chris Lattner91c15c42006-01-23 23:23:47 +00001245
1246<p>The strings can contain any character by escaping non-printable characters.
1247 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001248 for the number.</p>
Chris Lattner91c15c42006-01-23 23:23:47 +00001249
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001250<p>The inline asm code is simply printed to the machine code .s file when
1251 assembly code is generated.</p>
1252
Chris Lattner91c15c42006-01-23 23:23:47 +00001253</div>
Chris Lattner6af02f32004-12-09 16:11:40 +00001254
Reid Spencer50c723a2007-02-19 23:54:10 +00001255<!-- ======================================================================= -->
1256<div class="doc_subsection">
1257 <a name="datalayout">Data Layout</a>
1258</div>
1259
1260<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001261
Reid Spencer50c723a2007-02-19 23:54:10 +00001262<p>A module may specify a target specific data layout string that specifies how
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001263 data is to be laid out in memory. The syntax for the data layout is
1264 simply:</p>
1265
1266<div class="doc_code">
1267<pre>
1268target datalayout = "<i>layout specification</i>"
1269</pre>
1270</div>
1271
1272<p>The <i>layout specification</i> consists of a list of specifications
1273 separated by the minus sign character ('-'). Each specification starts with
1274 a letter and may include other information after the letter to define some
1275 aspect of the data layout. The specifications accepted are as follows:</p>
1276
Reid Spencer50c723a2007-02-19 23:54:10 +00001277<dl>
1278 <dt><tt>E</tt></dt>
1279 <dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001280 bits with the most significance have the lowest address location.</dd>
1281
Reid Spencer50c723a2007-02-19 23:54:10 +00001282 <dt><tt>e</tt></dt>
Chris Lattner67c37d12008-08-05 18:29:16 +00001283 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001284 the bits with the least significance have the lowest address
1285 location.</dd>
1286
Reid Spencer50c723a2007-02-19 23:54:10 +00001287 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher455c5772009-12-05 02:46:03 +00001288 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001289 <i>preferred</i> alignments. All sizes are in bits. Specifying
1290 the <i>pref</i> alignment is optional. If omitted, the
1291 preceding <tt>:</tt> should be omitted too.</dd>
1292
Reid Spencer50c723a2007-02-19 23:54:10 +00001293 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1294 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001295 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1296
Reid Spencer50c723a2007-02-19 23:54:10 +00001297 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher455c5772009-12-05 02:46:03 +00001298 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001299 <i>size</i>.</dd>
1300
Reid Spencer50c723a2007-02-19 23:54:10 +00001301 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher455c5772009-12-05 02:46:03 +00001302 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesence522852010-05-28 18:54:47 +00001303 <i>size</i>. Only values of <i>size</i> that are supported by the target
1304 will work. 32 (float) and 64 (double) are supported on all targets;
1305 80 or 128 (different flavors of long double) are also supported on some
1306 targets.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001307
Reid Spencer50c723a2007-02-19 23:54:10 +00001308 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1309 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001310 <i>size</i>.</dd>
1311
Daniel Dunbar7921a592009-06-08 22:17:53 +00001312 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1313 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001314 <i>size</i>.</dd>
Chris Lattnera381eff2009-11-07 09:35:34 +00001315
1316 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1317 <dd>This specifies a set of native integer widths for the target CPU
1318 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1319 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christopher455c5772009-12-05 02:46:03 +00001320 this set are considered to support most general arithmetic
Chris Lattnera381eff2009-11-07 09:35:34 +00001321 operations efficiently.</dd>
Reid Spencer50c723a2007-02-19 23:54:10 +00001322</dl>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001323
Reid Spencer50c723a2007-02-19 23:54:10 +00001324<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman61110ae2010-04-28 00:36:01 +00001325 default set of specifications which are then (possibly) overridden by the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001326 specifications in the <tt>datalayout</tt> keyword. The default specifications
1327 are given in this list:</p>
1328
Reid Spencer50c723a2007-02-19 23:54:10 +00001329<ul>
1330 <li><tt>E</tt> - big endian</li>
Dan Gohman8ad777d2010-02-23 02:44:03 +00001331 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
Reid Spencer50c723a2007-02-19 23:54:10 +00001332 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1333 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1334 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1335 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner67c37d12008-08-05 18:29:16 +00001336 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Reid Spencer50c723a2007-02-19 23:54:10 +00001337 alignment of 64-bits</li>
1338 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1339 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1340 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1341 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1342 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar7921a592009-06-08 22:17:53 +00001343 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencer50c723a2007-02-19 23:54:10 +00001344</ul>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001345
1346<p>When LLVM is determining the alignment for a given type, it uses the
1347 following rules:</p>
1348
Reid Spencer50c723a2007-02-19 23:54:10 +00001349<ol>
1350 <li>If the type sought is an exact match for one of the specifications, that
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001351 specification is used.</li>
1352
Reid Spencer50c723a2007-02-19 23:54:10 +00001353 <li>If no match is found, and the type sought is an integer type, then the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001354 smallest integer type that is larger than the bitwidth of the sought type
1355 is used. If none of the specifications are larger than the bitwidth then
1356 the the largest integer type is used. For example, given the default
1357 specifications above, the i7 type will use the alignment of i8 (next
1358 largest) while both i65 and i256 will use the alignment of i64 (largest
1359 specified).</li>
1360
Reid Spencer50c723a2007-02-19 23:54:10 +00001361 <li>If no match is found, and the type sought is a vector type, then the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001362 largest vector type that is smaller than the sought vector type will be
1363 used as a fall back. This happens because &lt;128 x double&gt; can be
1364 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencer50c723a2007-02-19 23:54:10 +00001365</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001366
Reid Spencer50c723a2007-02-19 23:54:10 +00001367</div>
Chris Lattner6af02f32004-12-09 16:11:40 +00001368
Dan Gohman6154a012009-07-27 18:07:55 +00001369<!-- ======================================================================= -->
1370<div class="doc_subsection">
1371 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1372</div>
1373
1374<div class="doc_text">
1375
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001376<p>Any memory access must be done through a pointer value associated
Andreas Bolkae39f0332009-07-27 20:37:10 +00001377with an address range of the memory access, otherwise the behavior
Dan Gohman6154a012009-07-27 18:07:55 +00001378is undefined. Pointer values are associated with address ranges
1379according to the following rules:</p>
1380
1381<ul>
Dan Gohmandf12d082010-07-02 18:41:32 +00001382 <li>A pointer value is associated with the addresses associated with
1383 any value it is <i>based</i> on.
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001384 <li>An address of a global variable is associated with the address
Dan Gohman6154a012009-07-27 18:07:55 +00001385 range of the variable's storage.</li>
1386 <li>The result value of an allocation instruction is associated with
1387 the address range of the allocated storage.</li>
1388 <li>A null pointer in the default address-space is associated with
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001389 no address.</li>
Dan Gohman6154a012009-07-27 18:07:55 +00001390 <li>An integer constant other than zero or a pointer value returned
1391 from a function not defined within LLVM may be associated with address
1392 ranges allocated through mechanisms other than those provided by
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001393 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman6154a012009-07-27 18:07:55 +00001394 allocated by mechanisms provided by LLVM.</li>
Dan Gohmandf12d082010-07-02 18:41:32 +00001395</ul>
1396
1397<p>A pointer value is <i>based</i> on another pointer value according
1398 to the following rules:</p>
1399
1400<ul>
1401 <li>A pointer value formed from a
1402 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1403 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1404 <li>The result value of a
1405 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1406 of the <tt>bitcast</tt>.</li>
1407 <li>A pointer value formed by an
1408 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1409 pointer values that contribute (directly or indirectly) to the
1410 computation of the pointer's value.</li>
1411 <li>The "<i>based</i> on" relationship is transitive.</li>
1412</ul>
1413
1414<p>Note that this definition of <i>"based"</i> is intentionally
1415 similar to the definition of <i>"based"</i> in C99, though it is
1416 slightly weaker.</p>
Dan Gohman6154a012009-07-27 18:07:55 +00001417
1418<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001419<tt><a href="#i_load">load</a></tt> merely indicates the size and
1420alignment of the memory from which to load, as well as the
Dan Gohman4eb47192010-06-17 19:23:50 +00001421interpretation of the value. The first operand type of a
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001422<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1423and alignment of the store.</p>
Dan Gohman6154a012009-07-27 18:07:55 +00001424
1425<p>Consequently, type-based alias analysis, aka TBAA, aka
1426<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1427LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1428additional information which specialized optimization passes may use
1429to implement type-based alias analysis.</p>
1430
1431</div>
1432
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00001433<!-- ======================================================================= -->
1434<div class="doc_subsection">
1435 <a name="volatile">Volatile Memory Accesses</a>
1436</div>
1437
1438<div class="doc_text">
1439
1440<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1441href="#i_store"><tt>store</tt></a>s, and <a
1442href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1443The optimizers must not change the number of volatile operations or change their
1444order of execution relative to other volatile operations. The optimizers
1445<i>may</i> change the order of volatile operations relative to non-volatile
1446operations. This is not Java's "volatile" and has no cross-thread
1447synchronization behavior.</p>
1448
1449</div>
1450
Chris Lattner2f7c9632001-06-06 20:29:01 +00001451<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001452<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1453<!-- *********************************************************************** -->
Chris Lattner6af02f32004-12-09 16:11:40 +00001454
Misha Brukman76307852003-11-08 01:05:38 +00001455<div class="doc_text">
Chris Lattner6af02f32004-12-09 16:11:40 +00001456
Misha Brukman76307852003-11-08 01:05:38 +00001457<p>The LLVM type system is one of the most important features of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001458 intermediate representation. Being typed enables a number of optimizations
1459 to be performed on the intermediate representation directly, without having
1460 to do extra analyses on the side before the transformation. A strong type
1461 system makes it easier to read the generated code and enables novel analyses
1462 and transformations that are not feasible to perform on normal three address
1463 code representations.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +00001464
1465</div>
1466
Chris Lattner2f7c9632001-06-06 20:29:01 +00001467<!-- ======================================================================= -->
Chris Lattner7824d182008-01-04 04:32:38 +00001468<div class="doc_subsection"> <a name="t_classifications">Type
Chris Lattner48b383b02003-11-25 01:02:51 +00001469Classifications</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001470
Misha Brukman76307852003-11-08 01:05:38 +00001471<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001472
1473<p>The types fall into a few useful classifications:</p>
Misha Brukmanc501f552004-03-01 17:47:27 +00001474
1475<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +00001476 <tbody>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001477 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner48b383b02003-11-25 01:02:51 +00001478 <tr>
Chris Lattner7824d182008-01-04 04:32:38 +00001479 <td><a href="#t_integer">integer</a></td>
Reid Spencer138249b2007-05-16 18:44:01 +00001480 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +00001481 </tr>
1482 <tr>
Chris Lattner7824d182008-01-04 04:32:38 +00001483 <td><a href="#t_floating">floating point</a></td>
1484 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +00001485 </tr>
1486 <tr>
1487 <td><a name="t_firstclass">first class</a></td>
Chris Lattner7824d182008-01-04 04:32:38 +00001488 <td><a href="#t_integer">integer</a>,
1489 <a href="#t_floating">floating point</a>,
1490 <a href="#t_pointer">pointer</a>,
Dan Gohman08783a882008-06-18 18:42:13 +00001491 <a href="#t_vector">vector</a>,
Dan Gohmanb9d66602008-05-12 23:51:09 +00001492 <a href="#t_struct">structure</a>,
Chris Lattner392be582010-02-12 20:49:41 +00001493 <a href="#t_union">union</a>,
Dan Gohmanb9d66602008-05-12 23:51:09 +00001494 <a href="#t_array">array</a>,
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001495 <a href="#t_label">label</a>,
1496 <a href="#t_metadata">metadata</a>.
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001497 </td>
Chris Lattner48b383b02003-11-25 01:02:51 +00001498 </tr>
Chris Lattner7824d182008-01-04 04:32:38 +00001499 <tr>
1500 <td><a href="#t_primitive">primitive</a></td>
1501 <td><a href="#t_label">label</a>,
1502 <a href="#t_void">void</a>,
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001503 <a href="#t_floating">floating point</a>,
1504 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner7824d182008-01-04 04:32:38 +00001505 </tr>
1506 <tr>
1507 <td><a href="#t_derived">derived</a></td>
Chris Lattner392be582010-02-12 20:49:41 +00001508 <td><a href="#t_array">array</a>,
Chris Lattner7824d182008-01-04 04:32:38 +00001509 <a href="#t_function">function</a>,
1510 <a href="#t_pointer">pointer</a>,
1511 <a href="#t_struct">structure</a>,
1512 <a href="#t_pstruct">packed structure</a>,
Chris Lattner392be582010-02-12 20:49:41 +00001513 <a href="#t_union">union</a>,
Chris Lattner7824d182008-01-04 04:32:38 +00001514 <a href="#t_vector">vector</a>,
1515 <a href="#t_opaque">opaque</a>.
Dan Gohman93bf60d2008-10-14 16:32:04 +00001516 </td>
Chris Lattner7824d182008-01-04 04:32:38 +00001517 </tr>
Chris Lattner48b383b02003-11-25 01:02:51 +00001518 </tbody>
Misha Brukman76307852003-11-08 01:05:38 +00001519</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00001520
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001521<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1522 important. Values of these types are the only ones which can be produced by
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001523 instructions.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001524
Misha Brukman76307852003-11-08 01:05:38 +00001525</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001526
Chris Lattner2f7c9632001-06-06 20:29:01 +00001527<!-- ======================================================================= -->
Chris Lattner7824d182008-01-04 04:32:38 +00001528<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner43542b32008-01-04 04:34:14 +00001529
Chris Lattner7824d182008-01-04 04:32:38 +00001530<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001531
Chris Lattner7824d182008-01-04 04:32:38 +00001532<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001533 system.</p>
Chris Lattner7824d182008-01-04 04:32:38 +00001534
Chris Lattner43542b32008-01-04 04:34:14 +00001535</div>
1536
Chris Lattner7824d182008-01-04 04:32:38 +00001537<!-- _______________________________________________________________________ -->
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001538<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1539
1540<div class="doc_text">
1541
1542<h5>Overview:</h5>
1543<p>The integer type is a very simple type that simply specifies an arbitrary
1544 bit width for the integer type desired. Any bit width from 1 bit to
1545 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1546
1547<h5>Syntax:</h5>
1548<pre>
1549 iN
1550</pre>
1551
1552<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1553 value.</p>
1554
1555<h5>Examples:</h5>
1556<table class="layout">
1557 <tr class="layout">
1558 <td class="left"><tt>i1</tt></td>
1559 <td class="left">a single-bit integer.</td>
1560 </tr>
1561 <tr class="layout">
1562 <td class="left"><tt>i32</tt></td>
1563 <td class="left">a 32-bit integer.</td>
1564 </tr>
1565 <tr class="layout">
1566 <td class="left"><tt>i1942652</tt></td>
1567 <td class="left">a really big integer of over 1 million bits.</td>
1568 </tr>
1569</table>
1570
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001571</div>
1572
1573<!-- _______________________________________________________________________ -->
Chris Lattner7824d182008-01-04 04:32:38 +00001574<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1575
1576<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001577
1578<table>
1579 <tbody>
1580 <tr><th>Type</th><th>Description</th></tr>
1581 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1582 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1583 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1584 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1585 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1586 </tbody>
1587</table>
1588
Chris Lattner7824d182008-01-04 04:32:38 +00001589</div>
1590
1591<!-- _______________________________________________________________________ -->
1592<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1593
1594<div class="doc_text">
Bill Wendling30235112009-07-20 02:39:26 +00001595
Chris Lattner7824d182008-01-04 04:32:38 +00001596<h5>Overview:</h5>
1597<p>The void type does not represent any value and has no size.</p>
1598
1599<h5>Syntax:</h5>
Chris Lattner7824d182008-01-04 04:32:38 +00001600<pre>
1601 void
1602</pre>
Bill Wendling30235112009-07-20 02:39:26 +00001603
Chris Lattner7824d182008-01-04 04:32:38 +00001604</div>
1605
1606<!-- _______________________________________________________________________ -->
1607<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1608
1609<div class="doc_text">
Bill Wendling30235112009-07-20 02:39:26 +00001610
Chris Lattner7824d182008-01-04 04:32:38 +00001611<h5>Overview:</h5>
1612<p>The label type represents code labels.</p>
1613
1614<h5>Syntax:</h5>
Chris Lattner7824d182008-01-04 04:32:38 +00001615<pre>
1616 label
1617</pre>
Bill Wendling30235112009-07-20 02:39:26 +00001618
Chris Lattner7824d182008-01-04 04:32:38 +00001619</div>
1620
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001621<!-- _______________________________________________________________________ -->
1622<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1623
1624<div class="doc_text">
Bill Wendling30235112009-07-20 02:39:26 +00001625
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001626<h5>Overview:</h5>
Nick Lewycky93e06a52009-09-27 23:27:42 +00001627<p>The metadata type represents embedded metadata. No derived types may be
1628 created from metadata except for <a href="#t_function">function</a>
1629 arguments.
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001630
1631<h5>Syntax:</h5>
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001632<pre>
1633 metadata
1634</pre>
Bill Wendling30235112009-07-20 02:39:26 +00001635
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001636</div>
1637
Chris Lattner7824d182008-01-04 04:32:38 +00001638
1639<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001640<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001641
Misha Brukman76307852003-11-08 01:05:38 +00001642<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +00001643
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001644<p>The real power in LLVM comes from the derived types in the system. This is
1645 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001646 useful types. Each of these types contain one or more element types which
1647 may be a primitive type, or another derived type. For example, it is
1648 possible to have a two dimensional array, using an array as the element type
1649 of another array.</p>
Dan Gohman142ccc02009-01-24 15:58:40 +00001650
Chris Lattner392be582010-02-12 20:49:41 +00001651
1652</div>
1653
1654<!-- _______________________________________________________________________ -->
1655<div class="doc_subsubsection"> <a name="t_aggregate">Aggregate Types</a> </div>
1656
1657<div class="doc_text">
1658
1659<p>Aggregate Types are a subset of derived types that can contain multiple
1660 member types. <a href="#t_array">Arrays</a>,
1661 <a href="#t_struct">structs</a>, <a href="#t_vector">vectors</a> and
1662 <a href="#t_union">unions</a> are aggregate types.</p>
1663
1664</div>
1665
Bill Wendling3716c5d2007-05-29 09:04:49 +00001666</div>
Reid Spencer138249b2007-05-16 18:44:01 +00001667
1668<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001669<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001670
Misha Brukman76307852003-11-08 01:05:38 +00001671<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +00001672
Chris Lattner2f7c9632001-06-06 20:29:01 +00001673<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001674<p>The array type is a very simple derived type that arranges elements
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001675 sequentially in memory. The array type requires a size (number of elements)
1676 and an underlying data type.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001677
Chris Lattner590645f2002-04-14 06:13:44 +00001678<h5>Syntax:</h5>
Chris Lattner74d3f822004-12-09 17:30:23 +00001679<pre>
1680 [&lt;# elements&gt; x &lt;elementtype&gt;]
1681</pre>
1682
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001683<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1684 be any type with a size.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001685
Chris Lattner590645f2002-04-14 06:13:44 +00001686<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001687<table class="layout">
1688 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00001689 <td class="left"><tt>[40 x i32]</tt></td>
1690 <td class="left">Array of 40 32-bit integer values.</td>
1691 </tr>
1692 <tr class="layout">
1693 <td class="left"><tt>[41 x i32]</tt></td>
1694 <td class="left">Array of 41 32-bit integer values.</td>
1695 </tr>
1696 <tr class="layout">
1697 <td class="left"><tt>[4 x i8]</tt></td>
1698 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001699 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001700</table>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001701<p>Here are some examples of multidimensional arrays:</p>
1702<table class="layout">
1703 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00001704 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1705 <td class="left">3x4 array of 32-bit integer values.</td>
1706 </tr>
1707 <tr class="layout">
1708 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1709 <td class="left">12x10 array of single precision floating point values.</td>
1710 </tr>
1711 <tr class="layout">
1712 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1713 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001714 </tr>
1715</table>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00001716
Dan Gohmanc74bc282009-11-09 19:01:53 +00001717<p>There is no restriction on indexing beyond the end of the array implied by
1718 a static type (though there are restrictions on indexing beyond the bounds
1719 of an allocated object in some cases). This means that single-dimension
1720 'variable sized array' addressing can be implemented in LLVM with a zero
1721 length array type. An implementation of 'pascal style arrays' in LLVM could
1722 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00001723
Misha Brukman76307852003-11-08 01:05:38 +00001724</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001725
Chris Lattner2f7c9632001-06-06 20:29:01 +00001726<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001727<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001728
Misha Brukman76307852003-11-08 01:05:38 +00001729<div class="doc_text">
Chris Lattnerda508ac2008-04-23 04:59:35 +00001730
Chris Lattner2f7c9632001-06-06 20:29:01 +00001731<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001732<p>The function type can be thought of as a function signature. It consists of
1733 a return type and a list of formal parameter types. The return type of a
Chris Lattner392be582010-02-12 20:49:41 +00001734 function type is a scalar type, a void type, a struct type, or a union
1735 type. If the return type is a struct type then all struct elements must be
1736 of first class types, and the struct must have at least one element.</p>
Devang Pateld6cff512008-03-10 20:49:15 +00001737
Chris Lattner2f7c9632001-06-06 20:29:01 +00001738<h5>Syntax:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +00001739<pre>
Nick Lewycky14d1ccc2009-09-27 07:55:32 +00001740 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerda508ac2008-04-23 04:59:35 +00001741</pre>
1742
John Criswell4c0cf7f2005-10-24 16:17:18 +00001743<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001744 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1745 which indicates that the function takes a variable number of arguments.
1746 Variable argument functions can access their arguments with
1747 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner47f2a832010-03-02 06:36:51 +00001748 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewycky93e06a52009-09-27 23:27:42 +00001749 <a href="#t_label">label</a>.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00001750
Chris Lattner2f7c9632001-06-06 20:29:01 +00001751<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001752<table class="layout">
1753 <tr class="layout">
Reid Spencer58c08712006-12-31 07:18:34 +00001754 <td class="left"><tt>i32 (i32)</tt></td>
1755 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001756 </td>
Reid Spencer58c08712006-12-31 07:18:34 +00001757 </tr><tr class="layout">
Chris Lattner47f2a832010-03-02 06:36:51 +00001758 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencer655dcc62006-12-31 07:20:23 +00001759 </tt></td>
Eric Christopher455c5772009-12-05 02:46:03 +00001760 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner47f2a832010-03-02 06:36:51 +00001761 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1762 returning <tt>float</tt>.
Reid Spencer58c08712006-12-31 07:18:34 +00001763 </td>
1764 </tr><tr class="layout">
1765 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher455c5772009-12-05 02:46:03 +00001766 <td class="left">A vararg function that takes at least one
1767 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1768 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer58c08712006-12-31 07:18:34 +00001769 LLVM.
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001770 </td>
Devang Patele3dfc1c2008-03-24 05:35:41 +00001771 </tr><tr class="layout">
1772 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky14d1ccc2009-09-27 07:55:32 +00001773 <td class="left">A function taking an <tt>i32</tt>, returning a
1774 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patele3dfc1c2008-03-24 05:35:41 +00001775 </td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001776 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001777</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00001778
Misha Brukman76307852003-11-08 01:05:38 +00001779</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001780
Chris Lattner2f7c9632001-06-06 20:29:01 +00001781<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001782<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001783
Misha Brukman76307852003-11-08 01:05:38 +00001784<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001785
Chris Lattner2f7c9632001-06-06 20:29:01 +00001786<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001787<p>The structure type is used to represent a collection of data members together
1788 in memory. The packing of the field types is defined to match the ABI of the
1789 underlying processor. The elements of a structure may be any type that has a
1790 size.</p>
1791
Jeffrey Yasskinf991bbb2010-01-11 19:19:26 +00001792<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1793 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1794 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1795 Structures in registers are accessed using the
1796 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1797 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001798<h5>Syntax:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00001799<pre>
1800 { &lt;type list&gt; }
1801</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001802
Chris Lattner2f7c9632001-06-06 20:29:01 +00001803<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001804<table class="layout">
1805 <tr class="layout">
Jeff Cohen5819f182007-04-22 01:17:39 +00001806 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1807 <td class="left">A triple of three <tt>i32</tt> values</td>
1808 </tr><tr class="layout">
1809 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1810 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1811 second element is a <a href="#t_pointer">pointer</a> to a
1812 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1813 an <tt>i32</tt>.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001814 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001815</table>
Dan Gohman142ccc02009-01-24 15:58:40 +00001816
Misha Brukman76307852003-11-08 01:05:38 +00001817</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001818
Chris Lattner2f7c9632001-06-06 20:29:01 +00001819<!-- _______________________________________________________________________ -->
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001820<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1821</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001822
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001823<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001824
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001825<h5>Overview:</h5>
1826<p>The packed structure type is used to represent a collection of data members
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001827 together in memory. There is no padding between fields. Further, the
1828 alignment of a packed structure is 1 byte. The elements of a packed
1829 structure may be any type that has a size.</p>
1830
1831<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1832 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1833 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1834
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001835<h5>Syntax:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00001836<pre>
1837 &lt; { &lt;type list&gt; } &gt;
1838</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001839
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001840<h5>Examples:</h5>
1841<table class="layout">
1842 <tr class="layout">
Jeff Cohen5819f182007-04-22 01:17:39 +00001843 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1844 <td class="left">A triple of three <tt>i32</tt> values</td>
1845 </tr><tr class="layout">
Bill Wendlingb175fa42008-09-07 10:26:33 +00001846 <td class="left">
1847<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Jeff Cohen5819f182007-04-22 01:17:39 +00001848 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1849 second element is a <a href="#t_pointer">pointer</a> to a
1850 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1851 an <tt>i32</tt>.</td>
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001852 </tr>
1853</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001854
Andrew Lenharth8df88e22006-12-08 17:13:00 +00001855</div>
1856
1857<!-- _______________________________________________________________________ -->
Chris Lattner392be582010-02-12 20:49:41 +00001858<div class="doc_subsubsection"> <a name="t_union">Union Type</a> </div>
1859
1860<div class="doc_text">
1861
1862<h5>Overview:</h5>
1863<p>A union type describes an object with size and alignment suitable for
1864 an object of any one of a given set of types (also known as an "untagged"
1865 union). It is similar in concept and usage to a
1866 <a href="#t_struct">struct</a>, except that all members of the union
1867 have an offset of zero. The elements of a union may be any type that has a
1868 size. Unions must have at least one member - empty unions are not allowed.
1869 </p>
1870
1871<p>The size of the union as a whole will be the size of its largest member,
1872 and the alignment requirements of the union as a whole will be the largest
1873 alignment requirement of any member.</p>
1874
Dan Gohman1ad14992010-02-25 16:51:31 +00001875<p>Union members are accessed using '<tt><a href="#i_load">load</a></tt> and
Chris Lattner392be582010-02-12 20:49:41 +00001876 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1877 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1878 Since all members are at offset zero, the getelementptr instruction does
1879 not affect the address, only the type of the resulting pointer.</p>
1880
1881<h5>Syntax:</h5>
1882<pre>
1883 union { &lt;type list&gt; }
1884</pre>
1885
1886<h5>Examples:</h5>
1887<table class="layout">
1888 <tr class="layout">
1889 <td class="left"><tt>union { i32, i32*, float }</tt></td>
1890 <td class="left">A union of three types: an <tt>i32</tt>, a pointer to
1891 an <tt>i32</tt>, and a <tt>float</tt>.</td>
1892 </tr><tr class="layout">
1893 <td class="left">
1894 <tt>union {&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1895 <td class="left">A union, where the first element is a <tt>float</tt> and the
1896 second element is a <a href="#t_pointer">pointer</a> to a
1897 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1898 an <tt>i32</tt>.</td>
1899 </tr>
1900</table>
1901
1902</div>
1903
1904<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00001905<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Chris Lattner4a67c912009-02-08 19:53:29 +00001906
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001907<div class="doc_text">
1908
1909<h5>Overview:</h5>
Dan Gohman88481112010-02-25 16:50:07 +00001910<p>The pointer type is used to specify memory locations.
1911 Pointers are commonly used to reference objects in memory.</p>
1912
1913<p>Pointer types may have an optional address space attribute defining the
1914 numbered address space where the pointed-to object resides. The default
1915 address space is number zero. The semantics of non-zero address
1916 spaces are target-specific.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001917
1918<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1919 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner4a67c912009-02-08 19:53:29 +00001920
Chris Lattner590645f2002-04-14 06:13:44 +00001921<h5>Syntax:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00001922<pre>
1923 &lt;type&gt; *
1924</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001925
Chris Lattner590645f2002-04-14 06:13:44 +00001926<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001927<table class="layout">
1928 <tr class="layout">
Dan Gohman623806e2009-01-04 23:44:43 +00001929 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner747359f2007-12-19 05:04:11 +00001930 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1931 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1932 </tr>
1933 <tr class="layout">
Dan Gohmanaabfdb32010-05-28 17:13:49 +00001934 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner747359f2007-12-19 05:04:11 +00001935 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001936 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner747359f2007-12-19 05:04:11 +00001937 <tt>i32</tt>.</td>
1938 </tr>
1939 <tr class="layout">
1940 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1941 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1942 that resides in address space #5.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001943 </tr>
Misha Brukman76307852003-11-08 01:05:38 +00001944</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001945
Misha Brukman76307852003-11-08 01:05:38 +00001946</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001947
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001948<!-- _______________________________________________________________________ -->
Reid Spencer404a3252007-02-15 03:07:05 +00001949<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001950
Misha Brukman76307852003-11-08 01:05:38 +00001951<div class="doc_text">
Chris Lattner37b6b092005-04-25 17:34:15 +00001952
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001953<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001954<p>A vector type is a simple derived type that represents a vector of elements.
1955 Vector types are used when multiple primitive data are operated in parallel
1956 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sands31c0e0e2009-11-27 13:38:03 +00001957 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001958 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +00001959
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001960<h5>Syntax:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +00001961<pre>
1962 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1963</pre>
1964
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001965<p>The number of elements is a constant integer value; elementtype may be any
1966 integer or floating point type.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +00001967
Chris Lattnerc8cb6952004-08-12 19:12:28 +00001968<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001969<table class="layout">
1970 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00001971 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1972 <td class="left">Vector of 4 32-bit integer values.</td>
1973 </tr>
1974 <tr class="layout">
1975 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1976 <td class="left">Vector of 8 32-bit floating-point values.</td>
1977 </tr>
1978 <tr class="layout">
1979 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1980 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001981 </tr>
1982</table>
Dan Gohman142ccc02009-01-24 15:58:40 +00001983
Misha Brukman76307852003-11-08 01:05:38 +00001984</div>
1985
Chris Lattner37b6b092005-04-25 17:34:15 +00001986<!-- _______________________________________________________________________ -->
1987<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1988<div class="doc_text">
1989
1990<h5>Overview:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +00001991<p>Opaque types are used to represent unknown types in the system. This
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001992 corresponds (for example) to the C notion of a forward declared structure
1993 type. In LLVM, opaque types can eventually be resolved to any type (not just
1994 a structure type).</p>
Chris Lattner37b6b092005-04-25 17:34:15 +00001995
1996<h5>Syntax:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +00001997<pre>
1998 opaque
1999</pre>
2000
2001<h5>Examples:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +00002002<table class="layout">
2003 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00002004 <td class="left"><tt>opaque</tt></td>
2005 <td class="left">An opaque type.</td>
Chris Lattner37b6b092005-04-25 17:34:15 +00002006 </tr>
2007</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002008
Chris Lattner37b6b092005-04-25 17:34:15 +00002009</div>
2010
Chris Lattnercf7a5842009-02-02 07:32:36 +00002011<!-- ======================================================================= -->
2012<div class="doc_subsection">
2013 <a name="t_uprefs">Type Up-references</a>
2014</div>
2015
2016<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002017
Chris Lattnercf7a5842009-02-02 07:32:36 +00002018<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002019<p>An "up reference" allows you to refer to a lexically enclosing type without
2020 requiring it to have a name. For instance, a structure declaration may
2021 contain a pointer to any of the types it is lexically a member of. Example
2022 of up references (with their equivalent as named type declarations)
2023 include:</p>
Chris Lattnercf7a5842009-02-02 07:32:36 +00002024
2025<pre>
Chris Lattnerbf1d5452009-02-09 10:00:56 +00002026 { \2 * } %x = type { %x* }
Chris Lattnercf7a5842009-02-02 07:32:36 +00002027 { \2 }* %y = type { %y }*
2028 \1* %z = type %z*
2029</pre>
2030
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002031<p>An up reference is needed by the asmprinter for printing out cyclic types
2032 when there is no declared name for a type in the cycle. Because the
2033 asmprinter does not want to print out an infinite type string, it needs a
2034 syntax to handle recursive types that have no names (all names are optional
2035 in llvm IR).</p>
Chris Lattnercf7a5842009-02-02 07:32:36 +00002036
2037<h5>Syntax:</h5>
2038<pre>
2039 \&lt;level&gt;
2040</pre>
2041
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002042<p>The level is the count of the lexical type that is being referred to.</p>
Chris Lattnercf7a5842009-02-02 07:32:36 +00002043
2044<h5>Examples:</h5>
Chris Lattnercf7a5842009-02-02 07:32:36 +00002045<table class="layout">
2046 <tr class="layout">
2047 <td class="left"><tt>\1*</tt></td>
2048 <td class="left">Self-referential pointer.</td>
2049 </tr>
2050 <tr class="layout">
2051 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
2052 <td class="left">Recursive structure where the upref refers to the out-most
2053 structure.</td>
2054 </tr>
2055</table>
Chris Lattnercf7a5842009-02-02 07:32:36 +00002056
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002057</div>
Chris Lattner37b6b092005-04-25 17:34:15 +00002058
Chris Lattner74d3f822004-12-09 17:30:23 +00002059<!-- *********************************************************************** -->
2060<div class="doc_section"> <a name="constants">Constants</a> </div>
2061<!-- *********************************************************************** -->
2062
2063<div class="doc_text">
2064
2065<p>LLVM has several different basic types of constants. This section describes
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002066 them all and their syntax.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002067
2068</div>
2069
2070<!-- ======================================================================= -->
Reid Spencer8f08d802004-12-09 18:02:53 +00002071<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002072
2073<div class="doc_text">
2074
2075<dl>
2076 <dt><b>Boolean constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002077 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00002078 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002079
2080 <dt><b>Integer constants</b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002081 <dd>Standard integers (such as '4') are constants of
2082 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2083 with integer types.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002084
2085 <dt><b>Floating point constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002086 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002087 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2088 notation (see below). The assembler requires the exact decimal value of a
2089 floating-point constant. For example, the assembler accepts 1.25 but
2090 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2091 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002092
2093 <dt><b>Null pointer constants</b></dt>
John Criswelldfe6a862004-12-10 15:51:16 +00002094 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002095 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002096</dl>
2097
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002098<p>The one non-intuitive notation for constants is the hexadecimal form of
2099 floating point constants. For example, the form '<tt>double
2100 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2101 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2102 constants are required (and the only time that they are generated by the
2103 disassembler) is when a floating point constant must be emitted but it cannot
2104 be represented as a decimal floating point number in a reasonable number of
2105 digits. For example, NaN's, infinities, and other special values are
2106 represented in their IEEE hexadecimal format so that assembly and disassembly
2107 do not cause any bits to change in the constants.</p>
2108
Dale Johannesencd4a3012009-02-11 22:14:51 +00002109<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002110 represented using the 16-digit form shown above (which matches the IEEE754
2111 representation for double); float values must, however, be exactly
2112 representable as IEE754 single precision. Hexadecimal format is always used
2113 for long double, and there are three forms of long double. The 80-bit format
2114 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2115 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2116 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2117 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2118 currently supported target uses this format. Long doubles will only work if
2119 they match the long double format on your target. All hexadecimal formats
2120 are big-endian (sign bit at the left).</p>
2121
Chris Lattner74d3f822004-12-09 17:30:23 +00002122</div>
2123
2124<!-- ======================================================================= -->
Chris Lattner361bfcd2009-02-28 18:32:25 +00002125<div class="doc_subsection">
Bill Wendling972b7202009-07-20 02:32:41 +00002126<a name="aggregateconstants"></a> <!-- old anchor -->
2127<a name="complexconstants">Complex Constants</a>
Chris Lattner74d3f822004-12-09 17:30:23 +00002128</div>
2129
2130<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002131
Chris Lattner361bfcd2009-02-28 18:32:25 +00002132<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002133 constants and smaller complex constants.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002134
2135<dl>
2136 <dt><b>Structure constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002137 <dd>Structure constants are represented with notation similar to structure
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002138 type definitions (a comma separated list of elements, surrounded by braces
2139 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2140 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2141 Structure constants must have <a href="#t_struct">structure type</a>, and
2142 the number and types of elements must match those specified by the
2143 type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002144
Chris Lattner392be582010-02-12 20:49:41 +00002145 <dt><b>Union constants</b></dt>
2146 <dd>Union constants are represented with notation similar to a structure with
2147 a single element - that is, a single typed element surrounded
2148 by braces (<tt>{}</tt>)). For example: "<tt>{ i32 4 }</tt>". The
2149 <a href="#t_union">union type</a> can be initialized with a single-element
2150 struct as long as the type of the struct element matches the type of
2151 one of the union members.</dd>
2152
Chris Lattner74d3f822004-12-09 17:30:23 +00002153 <dt><b>Array constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002154 <dd>Array constants are represented with notation similar to array type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002155 definitions (a comma separated list of elements, surrounded by square
2156 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2157 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2158 the number and types of elements must match those specified by the
2159 type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002160
Reid Spencer404a3252007-02-15 03:07:05 +00002161 <dt><b>Vector constants</b></dt>
Reid Spencer404a3252007-02-15 03:07:05 +00002162 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002163 definitions (a comma separated list of elements, surrounded by
2164 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2165 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2166 have <a href="#t_vector">vector type</a>, and the number and types of
2167 elements must match those specified by the type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002168
2169 <dt><b>Zero initialization</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002170 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattner392be582010-02-12 20:49:41 +00002171 value to zero of <em>any</em> type, including scalar and
2172 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002173 This is often used to avoid having to print large zero initializers
2174 (e.g. for large arrays) and is always exactly equivalent to using explicit
2175 zero initializers.</dd>
Nick Lewycky49f89192009-04-04 07:22:01 +00002176
2177 <dt><b>Metadata node</b></dt>
Nick Lewycky8e2c4f42009-05-30 16:08:30 +00002178 <dd>A metadata node is a structure-like constant with
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002179 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2180 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2181 be interpreted as part of the instruction stream, metadata is a place to
2182 attach additional information such as debug info.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002183</dl>
2184
2185</div>
2186
2187<!-- ======================================================================= -->
2188<div class="doc_subsection">
2189 <a name="globalconstants">Global Variable and Function Addresses</a>
2190</div>
2191
2192<div class="doc_text">
2193
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002194<p>The addresses of <a href="#globalvars">global variables</a>
2195 and <a href="#functionstructure">functions</a> are always implicitly valid
2196 (link-time) constants. These constants are explicitly referenced when
2197 the <a href="#identifiers">identifier for the global</a> is used and always
2198 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2199 legal LLVM file:</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002200
Bill Wendling3716c5d2007-05-29 09:04:49 +00002201<div class="doc_code">
Chris Lattner74d3f822004-12-09 17:30:23 +00002202<pre>
Chris Lattner00538a12007-06-06 18:28:13 +00002203@X = global i32 17
2204@Y = global i32 42
2205@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattner74d3f822004-12-09 17:30:23 +00002206</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002207</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002208
2209</div>
2210
2211<!-- ======================================================================= -->
Reid Spencer641f5c92004-12-09 18:13:12 +00002212<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002213<div class="doc_text">
2214
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002215<p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
Benjamin Kramer0f420382009-10-12 14:46:08 +00002216 indicates that the user of the value may receive an unspecified bit-pattern.
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002217 Undefined values may be of any type (other than label or void) and be used
2218 anywhere a constant is permitted.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002219
Chris Lattner92ada5d2009-09-11 01:49:31 +00002220<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002221 program is well defined no matter what value is used. This gives the
2222 compiler more freedom to optimize. Here are some examples of (potentially
2223 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002224
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002225
2226<div class="doc_code">
2227<pre>
2228 %A = add %X, undef
2229 %B = sub %X, undef
2230 %C = xor %X, undef
2231Safe:
2232 %A = undef
2233 %B = undef
2234 %C = undef
2235</pre>
2236</div>
2237
2238<p>This is safe because all of the output bits are affected by the undef bits.
2239Any output bit can have a zero or one depending on the input bits.</p>
2240
2241<div class="doc_code">
2242<pre>
2243 %A = or %X, undef
2244 %B = and %X, undef
2245Safe:
2246 %A = -1
2247 %B = 0
2248Unsafe:
2249 %A = undef
2250 %B = undef
2251</pre>
2252</div>
2253
2254<p>These logical operations have bits that are not always affected by the input.
2255For example, if "%X" has a zero bit, then the output of the 'and' operation will
2256always be a zero, no matter what the corresponding bit from the undef is. As
Chris Lattner92ada5d2009-09-11 01:49:31 +00002257such, it is unsafe to optimize or assume that the result of the and is undef.
Eric Christopher455c5772009-12-05 02:46:03 +00002258However, it is safe to assume that all bits of the undef could be 0, and
2259optimize the and to 0. Likewise, it is safe to assume that all the bits of
2260the undef operand to the or could be set, allowing the or to be folded to
Chris Lattner92ada5d2009-09-11 01:49:31 +00002261-1.</p>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002262
2263<div class="doc_code">
2264<pre>
2265 %A = select undef, %X, %Y
2266 %B = select undef, 42, %Y
2267 %C = select %X, %Y, undef
2268Safe:
2269 %A = %X (or %Y)
2270 %B = 42 (or %Y)
2271 %C = %Y
2272Unsafe:
2273 %A = undef
2274 %B = undef
2275 %C = undef
2276</pre>
2277</div>
2278
2279<p>This set of examples show that undefined select (and conditional branch)
2280conditions can go "either way" but they have to come from one of the two
2281operands. In the %A example, if %X and %Y were both known to have a clear low
2282bit, then %A would have to have a cleared low bit. However, in the %C example,
2283the optimizer is allowed to assume that the undef operand could be the same as
2284%Y, allowing the whole select to be eliminated.</p>
2285
2286
2287<div class="doc_code">
2288<pre>
2289 %A = xor undef, undef
Eric Christopher455c5772009-12-05 02:46:03 +00002290
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002291 %B = undef
2292 %C = xor %B, %B
2293
2294 %D = undef
2295 %E = icmp lt %D, 4
2296 %F = icmp gte %D, 4
2297
2298Safe:
2299 %A = undef
2300 %B = undef
2301 %C = undef
2302 %D = undef
2303 %E = undef
2304 %F = undef
2305</pre>
2306</div>
2307
2308<p>This example points out that two undef operands are not necessarily the same.
2309This can be surprising to people (and also matches C semantics) where they
2310assume that "X^X" is always zero, even if X is undef. This isn't true for a
2311number of reasons, but the short answer is that an undef "variable" can
2312arbitrarily change its value over its "live range". This is true because the
2313"variable" doesn't actually <em>have a live range</em>. Instead, the value is
2314logically read from arbitrary registers that happen to be around when needed,
Benjamin Kramer0f420382009-10-12 14:46:08 +00002315so the value is not necessarily consistent over time. In fact, %A and %C need
Chris Lattner6760e542009-09-08 15:13:16 +00002316to have the same semantics or the core LLVM "replace all uses with" concept
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002317would not hold.</p>
Chris Lattnera34a7182009-09-07 23:33:52 +00002318
2319<div class="doc_code">
2320<pre>
2321 %A = fdiv undef, %X
2322 %B = fdiv %X, undef
2323Safe:
2324 %A = undef
2325b: unreachable
2326</pre>
2327</div>
2328
2329<p>These examples show the crucial difference between an <em>undefined
2330value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2331allowed to have an arbitrary bit-pattern. This means that the %A operation
2332can be constant folded to undef because the undef could be an SNaN, and fdiv is
2333not (currently) defined on SNaN's. However, in the second example, we can make
2334a more aggressive assumption: because the undef is allowed to be an arbitrary
2335value, we are allowed to assume that it could be zero. Since a divide by zero
Chris Lattner10ff0c12009-09-08 19:45:34 +00002336has <em>undefined behavior</em>, we are allowed to assume that the operation
Chris Lattnera34a7182009-09-07 23:33:52 +00002337does not execute at all. This allows us to delete the divide and all code after
2338it: since the undefined operation "can't happen", the optimizer can assume that
2339it occurs in dead code.
2340</p>
Eric Christopher455c5772009-12-05 02:46:03 +00002341
Chris Lattnera34a7182009-09-07 23:33:52 +00002342<div class="doc_code">
2343<pre>
2344a: store undef -> %X
2345b: store %X -> undef
2346Safe:
2347a: &lt;deleted&gt;
2348b: unreachable
2349</pre>
2350</div>
2351
2352<p>These examples reiterate the fdiv example: a store "of" an undefined value
Eric Christopher455c5772009-12-05 02:46:03 +00002353can be assumed to not have any effect: we can assume that the value is
Chris Lattnera34a7182009-09-07 23:33:52 +00002354overwritten with bits that happen to match what was already there. However, a
2355store "to" an undefined location could clobber arbitrary memory, therefore, it
2356has undefined behavior.</p>
2357
Chris Lattner74d3f822004-12-09 17:30:23 +00002358</div>
2359
2360<!-- ======================================================================= -->
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002361<div class="doc_subsection"><a name="trapvalues">Trap Values</a></div>
2362<div class="doc_text">
2363
Dan Gohmanb2a709b2010-04-26 20:21:21 +00002364<p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002365 instead of representing an unspecified bit pattern, they represent the
2366 fact that an instruction or constant expression which cannot evoke side
2367 effects has nevertheless detected a condition which results in undefined
Dan Gohmanb2a709b2010-04-26 20:21:21 +00002368 behavior.</p>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002369
Dan Gohman2f1ae062010-04-28 00:49:41 +00002370<p>There is currently no way of representing a trap value in the IR; they
Dan Gohmanac355aa2010-05-03 14:51:43 +00002371 only exist when produced by operations such as
Dan Gohman2f1ae062010-04-28 00:49:41 +00002372 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002373
Dan Gohman2f1ae062010-04-28 00:49:41 +00002374<p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002375
Dan Gohman2f1ae062010-04-28 00:49:41 +00002376<p>
2377<ul>
2378<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2379 their operands.</li>
2380
2381<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2382 to their dynamic predecessor basic block.</li>
2383
2384<li>Function arguments depend on the corresponding actual argument values in
2385 the dynamic callers of their functions.</li>
2386
2387<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2388 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2389 control back to them.</li>
2390
Dan Gohman7292a752010-05-03 14:55:22 +00002391<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2392 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2393 or exception-throwing call instructions that dynamically transfer control
2394 back to them.</li>
2395
Dan Gohman2f1ae062010-04-28 00:49:41 +00002396<li>Non-volatile loads and stores depend on the most recent stores to all of the
2397 referenced memory addresses, following the order in the IR
2398 (including loads and stores implied by intrinsics such as
2399 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2400
Dan Gohman3513ea52010-05-03 14:59:34 +00002401<!-- TODO: In the case of multiple threads, this only applies if the store
2402 "happens-before" the load or store. -->
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002403
Dan Gohman2f1ae062010-04-28 00:49:41 +00002404<!-- TODO: floating-point exception state -->
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002405
Dan Gohman2f1ae062010-04-28 00:49:41 +00002406<li>An instruction with externally visible side effects depends on the most
2407 recent preceding instruction with externally visible side effects, following
Dan Gohman6c858db2010-07-06 15:26:33 +00002408 the order in the IR. (This includes
2409 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002410
Dan Gohman7292a752010-05-03 14:55:22 +00002411<li>An instruction <i>control-depends</i> on a
2412 <a href="#terminators">terminator instruction</a>
2413 if the terminator instruction has multiple successors and the instruction
2414 is always executed when control transfers to one of the successors, and
2415 may not be executed when control is transfered to another.</li>
Dan Gohman2f1ae062010-04-28 00:49:41 +00002416
2417<li>Dependence is transitive.</li>
2418
2419</ul>
2420</p>
2421
2422<p>Whenever a trap value is generated, all values which depend on it evaluate
2423 to trap. If they have side effects, the evoke their side effects as if each
2424 operand with a trap value were undef. If they have externally-visible side
2425 effects, the behavior is undefined.</p>
2426
2427<p>Here are some examples:</p>
Dan Gohman48a25882010-04-26 20:54:53 +00002428
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002429<div class="doc_code">
2430<pre>
2431entry:
2432 %trap = sub nuw i32 0, 1 ; Results in a trap value.
Dan Gohman2f1ae062010-04-28 00:49:41 +00002433 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2434 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2435 store i32 0, i32* %trap_yet_again ; undefined behavior
2436
2437 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2438 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2439
2440 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2441
2442 %narrowaddr = bitcast i32* @g to i16*
2443 %wideaddr = bitcast i32* @g to i64*
2444 %trap3 = load 16* %narrowaddr ; Returns a trap value.
2445 %trap4 = load i64* %widaddr ; Returns a trap value.
2446
2447 %cmp = icmp i32 slt %trap, 0 ; Returns a trap value.
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002448 %br i1 %cmp, %true, %end ; Branch to either destination.
2449
2450true:
Dan Gohman2f1ae062010-04-28 00:49:41 +00002451 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2452 ; it has undefined behavior.
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002453 br label %end
2454
2455end:
2456 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2457 ; Both edges into this PHI are
2458 ; control-dependent on %cmp, so this
Dan Gohman2f1ae062010-04-28 00:49:41 +00002459 ; always results in a trap value.
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002460
2461 volatile store i32 0, i32* @g ; %end is control-equivalent to %entry
2462 ; so this is defined (ignoring earlier
2463 ; undefined behavior in this example).
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002464</pre>
2465</div>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002466
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002467</div>
2468
2469<!-- ======================================================================= -->
Chris Lattner2bfd3202009-10-27 21:19:13 +00002470<div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2471 Blocks</a></div>
Chris Lattnere4801f72009-10-27 21:01:34 +00002472<div class="doc_text">
2473
Chris Lattneraa99c942009-11-01 01:27:45 +00002474<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnere4801f72009-10-27 21:01:34 +00002475
2476<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner5c5f0ac2009-10-27 21:49:40 +00002477 basic block in the specified function, and always has an i8* type. Taking
Chris Lattneraa99c942009-11-01 01:27:45 +00002478 the address of the entry block is illegal.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00002479
Chris Lattnere4801f72009-10-27 21:01:34 +00002480<p>This value only has defined behavior when used as an operand to the
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00002481 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
Chris Lattnere4801f72009-10-27 21:01:34 +00002482 against null. Pointer equality tests between labels addresses is undefined
2483 behavior - though, again, comparison against null is ok, and no label is
Chris Lattner2bfd3202009-10-27 21:19:13 +00002484 equal to the null pointer. This may also be passed around as an opaque
2485 pointer sized value as long as the bits are not inspected. This allows
Chris Lattnerda37b302009-10-27 21:44:20 +00002486 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00002487 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00002488
Chris Lattner2bfd3202009-10-27 21:19:13 +00002489<p>Finally, some targets may provide defined semantics when
Chris Lattnere4801f72009-10-27 21:01:34 +00002490 using the value as the operand to an inline assembly, but that is target
2491 specific.
2492 </p>
2493
2494</div>
2495
2496
2497<!-- ======================================================================= -->
Chris Lattner74d3f822004-12-09 17:30:23 +00002498<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2499</div>
2500
2501<div class="doc_text">
2502
2503<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002504 to be used as constants. Constant expressions may be of
2505 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2506 operation that does not have side effects (e.g. load and call are not
2507 supported). The following is the syntax for constant expressions:</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002508
2509<dl>
Dan Gohmand6a6f612010-05-28 17:07:41 +00002510 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002511 <dd>Truncate a constant to another type. The bit size of CST must be larger
2512 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002513
Dan Gohmand6a6f612010-05-28 17:07:41 +00002514 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002515 <dd>Zero extend a constant to another type. The bit size of CST must be
Duncan Sandsa522e562010-07-13 12:06:14 +00002516 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002517
Dan Gohmand6a6f612010-05-28 17:07:41 +00002518 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002519 <dd>Sign extend a constant to another type. The bit size of CST must be
Duncan Sandsa522e562010-07-13 12:06:14 +00002520 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002521
Dan Gohmand6a6f612010-05-28 17:07:41 +00002522 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002523 <dd>Truncate a floating point constant to another floating point type. The
2524 size of CST must be larger than the size of TYPE. Both types must be
2525 floating point.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002526
Dan Gohmand6a6f612010-05-28 17:07:41 +00002527 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002528 <dd>Floating point extend a constant to another type. The size of CST must be
2529 smaller or equal to the size of TYPE. Both types must be floating
2530 point.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002531
Dan Gohmand6a6f612010-05-28 17:07:41 +00002532 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002533 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002534 constant. TYPE must be a scalar or vector integer type. CST must be of
2535 scalar or vector floating point type. Both CST and TYPE must be scalars,
2536 or vectors of the same number of elements. If the value won't fit in the
2537 integer type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002538
Dan Gohmand6a6f612010-05-28 17:07:41 +00002539 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002540 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002541 constant. TYPE must be a scalar or vector integer type. CST must be of
2542 scalar or vector floating point type. Both CST and TYPE must be scalars,
2543 or vectors of the same number of elements. If the value won't fit in the
2544 integer type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002545
Dan Gohmand6a6f612010-05-28 17:07:41 +00002546 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002547 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002548 constant. TYPE must be a scalar or vector floating point type. CST must be
2549 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2550 vectors of the same number of elements. If the value won't fit in the
2551 floating point type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002552
Dan Gohmand6a6f612010-05-28 17:07:41 +00002553 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002554 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002555 constant. TYPE must be a scalar or vector floating point type. CST must be
2556 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2557 vectors of the same number of elements. If the value won't fit in the
2558 floating point type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002559
Dan Gohmand6a6f612010-05-28 17:07:41 +00002560 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5b950642006-11-11 23:08:07 +00002561 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002562 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2563 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2564 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5b950642006-11-11 23:08:07 +00002565
Dan Gohmand6a6f612010-05-28 17:07:41 +00002566 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002567 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2568 type. CST must be of integer type. The CST value is zero extended,
2569 truncated, or unchanged to make it fit in a pointer size. This one is
2570 <i>really</i> dangerous!</dd>
Reid Spencer5b950642006-11-11 23:08:07 +00002571
Dan Gohmand6a6f612010-05-28 17:07:41 +00002572 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner789dee32009-02-28 18:27:03 +00002573 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2574 are the same as those for the <a href="#i_bitcast">bitcast
2575 instruction</a>.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002576
Dan Gohmand6a6f612010-05-28 17:07:41 +00002577 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2578 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002579 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002580 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2581 instruction, the index list may have zero or more indexes, which are
2582 required to make sense for the type of "CSTPTR".</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002583
Dan Gohmand6a6f612010-05-28 17:07:41 +00002584 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002585 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer9965ee72006-12-04 19:23:19 +00002586
Dan Gohmand6a6f612010-05-28 17:07:41 +00002587 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer9965ee72006-12-04 19:23:19 +00002588 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2589
Dan Gohmand6a6f612010-05-28 17:07:41 +00002590 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer9965ee72006-12-04 19:23:19 +00002591 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino7e97a6d2006-01-10 19:31:34 +00002592
Dan Gohmand6a6f612010-05-28 17:07:41 +00002593 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002594 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2595 constants.</dd>
Robert Bocchino7e97a6d2006-01-10 19:31:34 +00002596
Dan Gohmand6a6f612010-05-28 17:07:41 +00002597 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002598 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2599 constants.</dd>
Chris Lattner016a0e52006-04-08 00:13:41 +00002600
Dan Gohmand6a6f612010-05-28 17:07:41 +00002601 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002602 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2603 constants.</dd>
Chris Lattner016a0e52006-04-08 00:13:41 +00002604
Nick Lewycky9ab9a7f2010-05-29 06:44:15 +00002605 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2606 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2607 constants. The index list is interpreted in a similar manner as indices in
2608 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2609 index value must be specified.</dd>
2610
2611 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2612 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2613 constants. The index list is interpreted in a similar manner as indices in
2614 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2615 index value must be specified.</dd>
2616
Dan Gohmand6a6f612010-05-28 17:07:41 +00002617 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002618 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2619 be any of the <a href="#binaryops">binary</a>
2620 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2621 on operands are the same as those for the corresponding instruction
2622 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002623</dl>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002624
Chris Lattner74d3f822004-12-09 17:30:23 +00002625</div>
Chris Lattnerb1652612004-03-08 16:49:10 +00002626
Chris Lattner2f7c9632001-06-06 20:29:01 +00002627<!-- *********************************************************************** -->
Chris Lattner98f013c2006-01-25 23:47:57 +00002628<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2629<!-- *********************************************************************** -->
2630
2631<!-- ======================================================================= -->
2632<div class="doc_subsection">
2633<a name="inlineasm">Inline Assembler Expressions</a>
2634</div>
2635
2636<div class="doc_text">
2637
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002638<p>LLVM supports inline assembler expressions (as opposed
2639 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2640 a special value. This value represents the inline assembler as a string
2641 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002642 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen1cfb9582009-10-21 23:28:00 +00002643 expression has side effects, and a flag indicating whether the function
2644 containing the asm needs to align its stack conservatively. An example
2645 inline assembler expression is:</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002646
Bill Wendling3716c5d2007-05-29 09:04:49 +00002647<div class="doc_code">
Chris Lattner98f013c2006-01-25 23:47:57 +00002648<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002649i32 (i32) asm "bswap $0", "=r,r"
Chris Lattner98f013c2006-01-25 23:47:57 +00002650</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002651</div>
Chris Lattner98f013c2006-01-25 23:47:57 +00002652
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002653<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2654 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2655 have:</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002656
Bill Wendling3716c5d2007-05-29 09:04:49 +00002657<div class="doc_code">
Chris Lattner98f013c2006-01-25 23:47:57 +00002658<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002659%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattner98f013c2006-01-25 23:47:57 +00002660</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002661</div>
Chris Lattner98f013c2006-01-25 23:47:57 +00002662
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002663<p>Inline asms with side effects not visible in the constraint list must be
2664 marked as having side effects. This is done through the use of the
2665 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002666
Bill Wendling3716c5d2007-05-29 09:04:49 +00002667<div class="doc_code">
Chris Lattner98f013c2006-01-25 23:47:57 +00002668<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002669call void asm sideeffect "eieio", ""()
Chris Lattner98f013c2006-01-25 23:47:57 +00002670</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00002671</div>
Chris Lattner98f013c2006-01-25 23:47:57 +00002672
Dale Johannesen1cfb9582009-10-21 23:28:00 +00002673<p>In some cases inline asms will contain code that will not work unless the
2674 stack is aligned in some way, such as calls or SSE instructions on x86,
2675 yet will not contain code that does that alignment within the asm.
2676 The compiler should make conservative assumptions about what the asm might
2677 contain and should generate its usual stack alignment code in the prologue
2678 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002679
2680<div class="doc_code">
2681<pre>
Dale Johannesen1cfb9582009-10-21 23:28:00 +00002682call void asm alignstack "eieio", ""()
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002683</pre>
2684</div>
2685
2686<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2687 first.</p>
2688
Chris Lattner98f013c2006-01-25 23:47:57 +00002689<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002690 documented here. Constraints on what can be done (e.g. duplication, moving,
2691 etc need to be documented). This is probably best done by reference to
2692 another document that covers inline asm from a holistic perspective.</p>
Chris Lattner51065562010-04-07 05:38:05 +00002693</div>
2694
2695<div class="doc_subsubsection">
2696<a name="inlineasm_md">Inline Asm Metadata</a>
2697</div>
2698
2699<div class="doc_text">
2700
2701<p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2702 attached to it that contains a constant integer. If present, the code
2703 generator will use the integer as the location cookie value when report
2704 errors through the LLVMContext error reporting mechanisms. This allows a
Dan Gohman61110ae2010-04-28 00:36:01 +00002705 front-end to correlate backend errors that occur with inline asm back to the
Chris Lattner51065562010-04-07 05:38:05 +00002706 source code that produced it. For example:</p>
2707
2708<div class="doc_code">
2709<pre>
2710call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2711...
2712!42 = !{ i32 1234567 }
2713</pre>
2714</div>
2715
2716<p>It is up to the front-end to make sense of the magic numbers it places in the
2717 IR.</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002718
2719</div>
2720
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002721<!-- ======================================================================= -->
2722<div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2723 Strings</a>
2724</div>
2725
2726<div class="doc_text">
2727
2728<p>LLVM IR allows metadata to be attached to instructions in the program that
2729 can convey extra information about the code to the optimizers and code
2730 generator. One example application of metadata is source-level debug
2731 information. There are two metadata primitives: strings and nodes. All
2732 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2733 preceding exclamation point ('<tt>!</tt>').</p>
2734
2735<p>A metadata string is a string surrounded by double quotes. It can contain
2736 any character by escaping non-printable characters with "\xx" where "xx" is
2737 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2738
2739<p>Metadata nodes are represented with notation similar to structure constants
2740 (a comma separated list of elements, surrounded by braces and preceded by an
2741 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2742 10}</tt>". Metadata nodes can have any values as their operand.</p>
2743
2744<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2745 metadata nodes, which can be looked up in the module symbol table. For
2746 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2747
Devang Patel9984bd62010-03-04 23:44:48 +00002748<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2749 function is using two metadata arguments.
2750
2751 <div class="doc_code">
2752 <pre>
2753 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2754 </pre>
2755 </div></p>
2756
2757<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2758 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.
2759
2760 <div class="doc_code">
2761 <pre>
2762 %indvar.next = add i64 %indvar, 1, !dbg !21
2763 </pre>
2764 </div></p>
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002765</div>
2766
Chris Lattnerae76db52009-07-20 05:55:19 +00002767
2768<!-- *********************************************************************** -->
2769<div class="doc_section">
2770 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2771</div>
2772<!-- *********************************************************************** -->
2773
2774<p>LLVM has a number of "magic" global variables that contain data that affect
2775code generation or other IR semantics. These are documented here. All globals
Chris Lattner58f9bb22009-07-20 06:14:25 +00002776of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2777section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2778by LLVM.</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00002779
2780<!-- ======================================================================= -->
2781<div class="doc_subsection">
2782<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2783</div>
2784
2785<div class="doc_text">
2786
2787<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2788href="#linkage_appending">appending linkage</a>. This array contains a list of
2789pointers to global variables and functions which may optionally have a pointer
2790cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2791
2792<pre>
2793 @X = global i8 4
2794 @Y = global i32 123
2795
2796 @llvm.used = appending global [2 x i8*] [
2797 i8* @X,
2798 i8* bitcast (i32* @Y to i8*)
2799 ], section "llvm.metadata"
2800</pre>
2801
2802<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2803compiler, assembler, and linker are required to treat the symbol as if there is
2804a reference to the global that it cannot see. For example, if a variable has
2805internal linkage and no references other than that from the <tt>@llvm.used</tt>
2806list, it cannot be deleted. This is commonly used to represent references from
2807inline asms and other things the compiler cannot "see", and corresponds to
2808"attribute((used))" in GNU C.</p>
2809
2810<p>On some targets, the code generator must emit a directive to the assembler or
2811object file to prevent the assembler and linker from molesting the symbol.</p>
2812
2813</div>
2814
2815<!-- ======================================================================= -->
2816<div class="doc_subsection">
Chris Lattner58f9bb22009-07-20 06:14:25 +00002817<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2818</div>
2819
2820<div class="doc_text">
2821
2822<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2823<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2824touching the symbol. On targets that support it, this allows an intelligent
2825linker to optimize references to the symbol without being impeded as it would be
2826by <tt>@llvm.used</tt>.</p>
2827
2828<p>This is a rare construct that should only be used in rare circumstances, and
2829should not be exposed to source languages.</p>
2830
2831</div>
2832
2833<!-- ======================================================================= -->
2834<div class="doc_subsection">
Chris Lattnerae76db52009-07-20 05:55:19 +00002835<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2836</div>
2837
2838<div class="doc_text">
David Chisnalla9d4a6f2010-04-30 19:23:49 +00002839<pre>
2840%0 = type { i32, void ()* }
David Chisnallb492b812010-04-30 19:27:35 +00002841@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalla9d4a6f2010-04-30 19:23:49 +00002842</pre>
2843<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.
2844</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00002845
2846</div>
2847
2848<!-- ======================================================================= -->
2849<div class="doc_subsection">
2850<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2851</div>
2852
2853<div class="doc_text">
David Chisnalla9d4a6f2010-04-30 19:23:49 +00002854<pre>
2855%0 = type { i32, void ()* }
David Chisnallb492b812010-04-30 19:27:35 +00002856@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalla9d4a6f2010-04-30 19:23:49 +00002857</pre>
Chris Lattnerae76db52009-07-20 05:55:19 +00002858
David Chisnalla9d4a6f2010-04-30 19:23:49 +00002859<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.
2860</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00002861
2862</div>
2863
2864
Chris Lattner98f013c2006-01-25 23:47:57 +00002865<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002866<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2867<!-- *********************************************************************** -->
Chris Lattner74d3f822004-12-09 17:30:23 +00002868
Misha Brukman76307852003-11-08 01:05:38 +00002869<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +00002870
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002871<p>The LLVM instruction set consists of several different classifications of
2872 instructions: <a href="#terminators">terminator
2873 instructions</a>, <a href="#binaryops">binary instructions</a>,
2874 <a href="#bitwiseops">bitwise binary instructions</a>,
2875 <a href="#memoryops">memory instructions</a>, and
2876 <a href="#otherops">other instructions</a>.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002877
Misha Brukman76307852003-11-08 01:05:38 +00002878</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002879
Chris Lattner2f7c9632001-06-06 20:29:01 +00002880<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002881<div class="doc_subsection"> <a name="terminators">Terminator
2882Instructions</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002883
Misha Brukman76307852003-11-08 01:05:38 +00002884<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +00002885
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002886<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2887 in a program ends with a "Terminator" instruction, which indicates which
2888 block should be executed after the current block is finished. These
2889 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2890 control flow, not values (the one exception being the
2891 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2892
Duncan Sands626b0242010-04-15 20:35:54 +00002893<p>There are seven different terminator instructions: the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002894 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2895 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2896 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
Bill Wendling33fef7e2009-11-02 00:25:26 +00002897 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002898 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2899 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2900 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002901
Misha Brukman76307852003-11-08 01:05:38 +00002902</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002903
Chris Lattner2f7c9632001-06-06 20:29:01 +00002904<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002905<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2906Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002907
Misha Brukman76307852003-11-08 01:05:38 +00002908<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002909
Chris Lattner2f7c9632001-06-06 20:29:01 +00002910<h5>Syntax:</h5>
Dan Gohmancc3132e2008-10-04 19:00:07 +00002911<pre>
2912 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner590645f2002-04-14 06:13:44 +00002913 ret void <i>; Return from void function</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002914</pre>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002915
Chris Lattner2f7c9632001-06-06 20:29:01 +00002916<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002917<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2918 a value) from a function back to the caller.</p>
2919
2920<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2921 value and then causes control flow, and one that just causes control flow to
2922 occur.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002923
Chris Lattner2f7c9632001-06-06 20:29:01 +00002924<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002925<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2926 return value. The type of the return value must be a
2927 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmancc3132e2008-10-04 19:00:07 +00002928
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002929<p>A function is not <a href="#wellformed">well formed</a> if it it has a
2930 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2931 value or a return value with a type that does not match its type, or if it
2932 has a void return type and contains a '<tt>ret</tt>' instruction with a
2933 return value.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002934
Chris Lattner2f7c9632001-06-06 20:29:01 +00002935<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002936<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2937 the calling function's context. If the caller is a
2938 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2939 instruction after the call. If the caller was an
2940 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2941 the beginning of the "normal" destination block. If the instruction returns
2942 a value, that value shall set the call or invoke instruction's return
2943 value.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002944
Chris Lattner2f7c9632001-06-06 20:29:01 +00002945<h5>Example:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002946<pre>
2947 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner590645f2002-04-14 06:13:44 +00002948 ret void <i>; Return from a void function</i>
Bill Wendling050ee8f2009-02-28 22:12:54 +00002949 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002950</pre>
Dan Gohman3065b612009-01-12 23:12:39 +00002951
Misha Brukman76307852003-11-08 01:05:38 +00002952</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002953<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00002954<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002955
Misha Brukman76307852003-11-08 01:05:38 +00002956<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002957
Chris Lattner2f7c9632001-06-06 20:29:01 +00002958<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002959<pre>
2960 br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002961</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002962
Chris Lattner2f7c9632001-06-06 20:29:01 +00002963<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002964<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2965 different basic block in the current function. There are two forms of this
2966 instruction, corresponding to a conditional branch and an unconditional
2967 branch.</p>
2968
Chris Lattner2f7c9632001-06-06 20:29:01 +00002969<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002970<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2971 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2972 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2973 target.</p>
2974
Chris Lattner2f7c9632001-06-06 20:29:01 +00002975<h5>Semantics:</h5>
Reid Spencer36a15422007-01-12 03:35:51 +00002976<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002977 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2978 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2979 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2980
Chris Lattner2f7c9632001-06-06 20:29:01 +00002981<h5>Example:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00002982<pre>
2983Test:
2984 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2985 br i1 %cond, label %IfEqual, label %IfUnequal
2986IfEqual:
2987 <a href="#i_ret">ret</a> i32 1
2988IfUnequal:
2989 <a href="#i_ret">ret</a> i32 0
2990</pre>
2991
Misha Brukman76307852003-11-08 01:05:38 +00002992</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002993
Chris Lattner2f7c9632001-06-06 20:29:01 +00002994<!-- _______________________________________________________________________ -->
Chris Lattnercf96c6c2004-02-24 04:54:45 +00002995<div class="doc_subsubsection">
2996 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2997</div>
2998
Misha Brukman76307852003-11-08 01:05:38 +00002999<div class="doc_text">
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003000
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003001<h5>Syntax:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003002<pre>
3003 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
3004</pre>
3005
Chris Lattner2f7c9632001-06-06 20:29:01 +00003006<h5>Overview:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003007<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003008 several different places. It is a generalization of the '<tt>br</tt>'
3009 instruction, allowing a branch to occur to one of many possible
3010 destinations.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003011
Chris Lattner2f7c9632001-06-06 20:29:01 +00003012<h5>Arguments:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003013<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003014 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3015 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3016 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003017
Chris Lattner2f7c9632001-06-06 20:29:01 +00003018<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00003019<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003020 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3021 is searched for the given value. If the value is found, control flow is
Benjamin Kramer0f420382009-10-12 14:46:08 +00003022 transferred to the corresponding destination; otherwise, control flow is
3023 transferred to the default destination.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003024
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003025<h5>Implementation:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003026<p>Depending on properties of the target machine and the particular
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003027 <tt>switch</tt> instruction, this instruction may be code generated in
3028 different ways. For example, it could be generated as a series of chained
3029 conditional branches or with a lookup table.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003030
3031<h5>Example:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003032<pre>
3033 <i>; Emulate a conditional br instruction</i>
Reid Spencer36a15422007-01-12 03:35:51 +00003034 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman623806e2009-01-04 23:44:43 +00003035 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003036
3037 <i>; Emulate an unconditional br instruction</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00003038 switch i32 0, label %dest [ ]
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003039
3040 <i>; Implement a jump table:</i>
Dan Gohman623806e2009-01-04 23:44:43 +00003041 switch i32 %val, label %otherwise [ i32 0, label %onzero
3042 i32 1, label %onone
3043 i32 2, label %ontwo ]
Chris Lattner2f7c9632001-06-06 20:29:01 +00003044</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003045
Misha Brukman76307852003-11-08 01:05:38 +00003046</div>
Chris Lattner0132aff2005-05-06 22:57:40 +00003047
Chris Lattner3ed871f2009-10-27 19:13:16 +00003048
3049<!-- _______________________________________________________________________ -->
3050<div class="doc_subsubsection">
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003051 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
Chris Lattner3ed871f2009-10-27 19:13:16 +00003052</div>
3053
3054<div class="doc_text">
3055
3056<h5>Syntax:</h5>
3057<pre>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003058 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattner3ed871f2009-10-27 19:13:16 +00003059</pre>
3060
3061<h5>Overview:</h5>
3062
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003063<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattner3ed871f2009-10-27 19:13:16 +00003064 within the current function, whose address is specified by
Chris Lattnere4801f72009-10-27 21:01:34 +00003065 "<tt>address</tt>". Address must be derived from a <a
3066 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattner3ed871f2009-10-27 19:13:16 +00003067
3068<h5>Arguments:</h5>
3069
3070<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3071 rest of the arguments indicate the full set of possible destinations that the
3072 address may point to. Blocks are allowed to occur multiple times in the
3073 destination list, though this isn't particularly useful.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00003074
Chris Lattner3ed871f2009-10-27 19:13:16 +00003075<p>This destination list is required so that dataflow analysis has an accurate
3076 understanding of the CFG.</p>
3077
3078<h5>Semantics:</h5>
3079
3080<p>Control transfers to the block specified in the address argument. All
3081 possible destination blocks must be listed in the label list, otherwise this
3082 instruction has undefined behavior. This implies that jumps to labels
3083 defined in other functions have undefined behavior as well.</p>
3084
3085<h5>Implementation:</h5>
3086
3087<p>This is typically implemented with a jump through a register.</p>
3088
3089<h5>Example:</h5>
3090<pre>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003091 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattner3ed871f2009-10-27 19:13:16 +00003092</pre>
3093
3094</div>
3095
3096
Chris Lattner2f7c9632001-06-06 20:29:01 +00003097<!-- _______________________________________________________________________ -->
Chris Lattner0132aff2005-05-06 22:57:40 +00003098<div class="doc_subsubsection">
3099 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3100</div>
3101
Misha Brukman76307852003-11-08 01:05:38 +00003102<div class="doc_text">
Chris Lattner0132aff2005-05-06 22:57:40 +00003103
Chris Lattner2f7c9632001-06-06 20:29:01 +00003104<h5>Syntax:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00003105<pre>
Devang Patel02256232008-10-07 17:48:33 +00003106 &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 Lattner6b7a0082006-05-14 18:23:06 +00003107 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattner0132aff2005-05-06 22:57:40 +00003108</pre>
3109
Chris Lattnera8292f32002-05-06 22:08:29 +00003110<h5>Overview:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00003111<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003112 function, with the possibility of control flow transfer to either the
3113 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3114 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3115 control flow will return to the "normal" label. If the callee (or any
3116 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3117 instruction, control is interrupted and continued at the dynamically nearest
3118 "exception" label.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003119
Chris Lattner2f7c9632001-06-06 20:29:01 +00003120<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003121<p>This instruction requires several arguments:</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003122
Chris Lattner2f7c9632001-06-06 20:29:01 +00003123<ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003124 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3125 convention</a> the call should use. If none is specified, the call
3126 defaults to using C calling conventions.</li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00003127
3128 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003129 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3130 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00003131
Chris Lattner0132aff2005-05-06 22:57:40 +00003132 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003133 function value being invoked. In most cases, this is a direct function
3134 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3135 off an arbitrary pointer to function value.</li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003136
3137 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003138 function to be invoked. </li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003139
3140 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner47f2a832010-03-02 06:36:51 +00003141 signature argument types and parameter attributes. All arguments must be
3142 of <a href="#t_firstclass">first class</a> type. If the function
3143 signature indicates the function accepts a variable number of arguments,
3144 the extra arguments can be specified.</li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003145
3146 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003147 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003148
3149 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003150 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003151
Devang Patel02256232008-10-07 17:48:33 +00003152 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003153 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3154 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003155</ol>
Chris Lattner0132aff2005-05-06 22:57:40 +00003156
Chris Lattner2f7c9632001-06-06 20:29:01 +00003157<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003158<p>This instruction is designed to operate as a standard
3159 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3160 primary difference is that it establishes an association with a label, which
3161 is used by the runtime library to unwind the stack.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003162
3163<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003164 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3165 exception. Additionally, this is important for implementation of
3166 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003167
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003168<p>For the purposes of the SSA form, the definition of the value returned by the
3169 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3170 block to the "normal" label. If the callee unwinds then no return value is
3171 available.</p>
Dan Gohman9069d892009-05-22 21:47:08 +00003172
Chris Lattner97257f82010-01-15 18:08:37 +00003173<p>Note that the code generator does not yet completely support unwind, and
3174that the invoke/unwind semantics are likely to change in future versions.</p>
3175
Chris Lattner2f7c9632001-06-06 20:29:01 +00003176<h5>Example:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00003177<pre>
Nick Lewycky084ab472008-03-16 07:18:12 +00003178 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen5819f182007-04-22 01:17:39 +00003179 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewycky084ab472008-03-16 07:18:12 +00003180 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen5819f182007-04-22 01:17:39 +00003181 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003182</pre>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003183
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003184</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003185
Chris Lattner5ed60612003-09-03 00:41:47 +00003186<!-- _______________________________________________________________________ -->
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003187
Chris Lattner48b383b02003-11-25 01:02:51 +00003188<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
3189Instruction</a> </div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003190
Misha Brukman76307852003-11-08 01:05:38 +00003191<div class="doc_text">
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003192
Chris Lattner5ed60612003-09-03 00:41:47 +00003193<h5>Syntax:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003194<pre>
3195 unwind
3196</pre>
3197
Chris Lattner5ed60612003-09-03 00:41:47 +00003198<h5>Overview:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003199<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003200 at the first callee in the dynamic call stack which used
3201 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3202 This is primarily used to implement exception handling.</p>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003203
Chris Lattner5ed60612003-09-03 00:41:47 +00003204<h5>Semantics:</h5>
Chris Lattnerfe8519c2008-04-19 21:01:16 +00003205<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003206 immediately halt. The dynamic call stack is then searched for the
3207 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3208 Once found, execution continues at the "exceptional" destination block
3209 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3210 instruction in the dynamic call chain, undefined behavior results.</p>
3211
Chris Lattner97257f82010-01-15 18:08:37 +00003212<p>Note that the code generator does not yet completely support unwind, and
3213that the invoke/unwind semantics are likely to change in future versions.</p>
3214
Misha Brukman76307852003-11-08 01:05:38 +00003215</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003216
3217<!-- _______________________________________________________________________ -->
3218
3219<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
3220Instruction</a> </div>
3221
3222<div class="doc_text">
3223
3224<h5>Syntax:</h5>
3225<pre>
3226 unreachable
3227</pre>
3228
3229<h5>Overview:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003230<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003231 instruction is used to inform the optimizer that a particular portion of the
3232 code is not reachable. This can be used to indicate that the code after a
3233 no-return function cannot be reached, and other facts.</p>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003234
3235<h5>Semantics:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003236<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003237
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003238</div>
3239
Chris Lattner2f7c9632001-06-06 20:29:01 +00003240<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +00003241<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003242
Misha Brukman76307852003-11-08 01:05:38 +00003243<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003244
3245<p>Binary operators are used to do most of the computation in a program. They
3246 require two operands of the same type, execute an operation on them, and
3247 produce a single value. The operands might represent multiple data, as is
3248 the case with the <a href="#t_vector">vector</a> data type. The result value
3249 has the same type as its operands.</p>
3250
Misha Brukman76307852003-11-08 01:05:38 +00003251<p>There are several different binary operators:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003252
Misha Brukman76307852003-11-08 01:05:38 +00003253</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003254
Chris Lattner2f7c9632001-06-06 20:29:01 +00003255<!-- _______________________________________________________________________ -->
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003256<div class="doc_subsubsection">
3257 <a name="i_add">'<tt>add</tt>' Instruction</a>
3258</div>
3259
Misha Brukman76307852003-11-08 01:05:38 +00003260<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003261
Chris Lattner2f7c9632001-06-06 20:29:01 +00003262<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003263<pre>
Dan Gohmanb07de442009-07-20 22:41:19 +00003264 &lt;result&gt; = add &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohman957b1312009-09-02 17:31:42 +00003265 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3266 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3267 &lt;result&gt; = add nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003268</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003269
Chris Lattner2f7c9632001-06-06 20:29:01 +00003270<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003271<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003272
Chris Lattner2f7c9632001-06-06 20:29:01 +00003273<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003274<p>The two arguments to the '<tt>add</tt>' instruction must
3275 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3276 integer values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003277
Chris Lattner2f7c9632001-06-06 20:29:01 +00003278<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003279<p>The value produced is the integer sum of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003280
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003281<p>If the sum has unsigned overflow, the result returned is the mathematical
3282 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003283
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003284<p>Because LLVM integers use a two's complement representation, this instruction
3285 is appropriate for both signed and unsigned integers.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003286
Dan Gohman902dfff2009-07-22 22:44:56 +00003287<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3288 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3289 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00003290 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3291 respectively, occurs.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00003292
Chris Lattner2f7c9632001-06-06 20:29:01 +00003293<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003294<pre>
3295 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003296</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003297
Misha Brukman76307852003-11-08 01:05:38 +00003298</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003299
Chris Lattner2f7c9632001-06-06 20:29:01 +00003300<!-- _______________________________________________________________________ -->
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003301<div class="doc_subsubsection">
Dan Gohmana5b96452009-06-04 22:49:04 +00003302 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3303</div>
3304
3305<div class="doc_text">
3306
3307<h5>Syntax:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003308<pre>
3309 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3310</pre>
3311
3312<h5>Overview:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003313<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3314
3315<h5>Arguments:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003316<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003317 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3318 floating point values. Both arguments must have identical types.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003319
3320<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003321<p>The value produced is the floating point sum of the two operands.</p>
3322
3323<h5>Example:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003324<pre>
3325 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3326</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003327
Dan Gohmana5b96452009-06-04 22:49:04 +00003328</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003329
Dan Gohmana5b96452009-06-04 22:49:04 +00003330<!-- _______________________________________________________________________ -->
3331<div class="doc_subsubsection">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003332 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3333</div>
3334
Misha Brukman76307852003-11-08 01:05:38 +00003335<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003336
Chris Lattner2f7c9632001-06-06 20:29:01 +00003337<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003338<pre>
Dan Gohman902dfff2009-07-22 22:44:56 +00003339 &lt;result&gt; = sub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohman957b1312009-09-02 17:31:42 +00003340 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3341 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3342 &lt;result&gt; = sub nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003343</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003344
Chris Lattner2f7c9632001-06-06 20:29:01 +00003345<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003346<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003347 operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003348
3349<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003350 '<tt>neg</tt>' instruction present in most other intermediate
3351 representations.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003352
Chris Lattner2f7c9632001-06-06 20:29:01 +00003353<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003354<p>The two arguments to the '<tt>sub</tt>' instruction must
3355 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3356 integer values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003357
Chris Lattner2f7c9632001-06-06 20:29:01 +00003358<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003359<p>The value produced is the integer difference of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003360
Dan Gohmana5b96452009-06-04 22:49:04 +00003361<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003362 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3363 result.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003364
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003365<p>Because LLVM integers use a two's complement representation, this instruction
3366 is appropriate for both signed and unsigned integers.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003367
Dan Gohman902dfff2009-07-22 22:44:56 +00003368<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3369 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3370 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00003371 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3372 respectively, occurs.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00003373
Chris Lattner2f7c9632001-06-06 20:29:01 +00003374<h5>Example:</h5>
Bill Wendling2d8b9a82007-05-29 09:42:13 +00003375<pre>
3376 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00003377 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003378</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003379
Misha Brukman76307852003-11-08 01:05:38 +00003380</div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003381
Chris Lattner2f7c9632001-06-06 20:29:01 +00003382<!-- _______________________________________________________________________ -->
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003383<div class="doc_subsubsection">
Dan Gohmana5b96452009-06-04 22:49:04 +00003384 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3385</div>
3386
3387<div class="doc_text">
3388
3389<h5>Syntax:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003390<pre>
3391 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3392</pre>
3393
3394<h5>Overview:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003395<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003396 operands.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003397
3398<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003399 '<tt>fneg</tt>' instruction present in most other intermediate
3400 representations.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003401
3402<h5>Arguments:</h5>
Bill Wendling972b7202009-07-20 02:32:41 +00003403<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003404 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3405 floating point values. Both arguments must have identical types.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003406
3407<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003408<p>The value produced is the floating point difference of the two operands.</p>
3409
3410<h5>Example:</h5>
3411<pre>
3412 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3413 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3414</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003415
Dan Gohmana5b96452009-06-04 22:49:04 +00003416</div>
3417
3418<!-- _______________________________________________________________________ -->
3419<div class="doc_subsubsection">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003420 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3421</div>
3422
Misha Brukman76307852003-11-08 01:05:38 +00003423<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003424
Chris Lattner2f7c9632001-06-06 20:29:01 +00003425<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003426<pre>
Dan Gohman902dfff2009-07-22 22:44:56 +00003427 &lt;result&gt; = mul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohman957b1312009-09-02 17:31:42 +00003428 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3429 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3430 &lt;result&gt; = mul nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003431</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003432
Chris Lattner2f7c9632001-06-06 20:29:01 +00003433<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003434<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003435
Chris Lattner2f7c9632001-06-06 20:29:01 +00003436<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003437<p>The two arguments to the '<tt>mul</tt>' instruction must
3438 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3439 integer values. Both arguments must have identical types.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00003440
Chris Lattner2f7c9632001-06-06 20:29:01 +00003441<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003442<p>The value produced is the integer product of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003443
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003444<p>If the result of the multiplication has unsigned overflow, the result
3445 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3446 width of the result.</p>
3447
3448<p>Because LLVM integers use a two's complement representation, and the result
3449 is the same width as the operands, this instruction returns the correct
3450 result for both signed and unsigned integers. If a full product
3451 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3452 be sign-extended or zero-extended as appropriate to the width of the full
3453 product.</p>
3454
Dan Gohman902dfff2009-07-22 22:44:56 +00003455<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3456 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3457 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00003458 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3459 respectively, occurs.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00003460
Chris Lattner2f7c9632001-06-06 20:29:01 +00003461<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003462<pre>
3463 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003464</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003465
Misha Brukman76307852003-11-08 01:05:38 +00003466</div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003467
Chris Lattner2f7c9632001-06-06 20:29:01 +00003468<!-- _______________________________________________________________________ -->
Dan Gohmana5b96452009-06-04 22:49:04 +00003469<div class="doc_subsubsection">
3470 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3471</div>
3472
3473<div class="doc_text">
3474
3475<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003476<pre>
3477 &lt;result&gt; = fmul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmana5b96452009-06-04 22:49:04 +00003478</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003479
Dan Gohmana5b96452009-06-04 22:49:04 +00003480<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003481<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003482
3483<h5>Arguments:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003484<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003485 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3486 floating point values. Both arguments must have identical types.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003487
3488<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003489<p>The value produced is the floating point product of the two operands.</p>
3490
3491<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003492<pre>
3493 &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohmana5b96452009-06-04 22:49:04 +00003494</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003495
Dan Gohmana5b96452009-06-04 22:49:04 +00003496</div>
3497
3498<!-- _______________________________________________________________________ -->
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003499<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3500</a></div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003501
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003502<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003503
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003504<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003505<pre>
3506 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003507</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003508
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003509<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003510<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003511
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003512<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003513<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003514 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3515 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003516
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003517<h5>Semantics:</h5>
Chris Lattner2f2427e2008-01-28 00:36:27 +00003518<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003519
Chris Lattner2f2427e2008-01-28 00:36:27 +00003520<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003521 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3522
Chris Lattner2f2427e2008-01-28 00:36:27 +00003523<p>Division by zero leads to undefined behavior.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003524
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003525<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003526<pre>
3527 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003528</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003529
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003530</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003531
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003532<!-- _______________________________________________________________________ -->
3533<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3534</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003535
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003536<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003537
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003538<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003539<pre>
Dan Gohmanb07de442009-07-20 22:41:19 +00003540 &lt;result&gt; = sdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohman957b1312009-09-02 17:31:42 +00003541 &lt;result&gt; = sdiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003542</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003543
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003544<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003545<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003546
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003547<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003548<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003549 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3550 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003551
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003552<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003553<p>The value produced is the signed integer quotient of the two operands rounded
3554 towards zero.</p>
3555
Chris Lattner2f2427e2008-01-28 00:36:27 +00003556<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003557 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3558
Chris Lattner2f2427e2008-01-28 00:36:27 +00003559<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003560 undefined behavior; this is a rare case, but can occur, for example, by doing
3561 a 32-bit division of -2147483648 by -1.</p>
3562
Dan Gohman71dfd782009-07-22 00:04:19 +00003563<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman57255802010-04-23 15:23:32 +00003564 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
Dan Gohmane501ff72010-07-11 00:08:34 +00003565 be rounded.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00003566
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003567<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003568<pre>
3569 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003570</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003571
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003572</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003573
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003574<!-- _______________________________________________________________________ -->
3575<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner48b383b02003-11-25 01:02:51 +00003576Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003577
Misha Brukman76307852003-11-08 01:05:38 +00003578<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003579
Chris Lattner2f7c9632001-06-06 20:29:01 +00003580<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003581<pre>
Gabor Greif0f75ad02008-08-07 21:46:00 +00003582 &lt;result&gt; = fdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner48b383b02003-11-25 01:02:51 +00003583</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003584
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003585<h5>Overview:</h5>
3586<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003587
Chris Lattner48b383b02003-11-25 01:02:51 +00003588<h5>Arguments:</h5>
Jeff Cohen5819f182007-04-22 01:17:39 +00003589<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003590 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3591 floating point values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003592
Chris Lattner48b383b02003-11-25 01:02:51 +00003593<h5>Semantics:</h5>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00003594<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003595
Chris Lattner48b383b02003-11-25 01:02:51 +00003596<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003597<pre>
3598 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner48b383b02003-11-25 01:02:51 +00003599</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003600
Chris Lattner48b383b02003-11-25 01:02:51 +00003601</div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003602
Chris Lattner48b383b02003-11-25 01:02:51 +00003603<!-- _______________________________________________________________________ -->
Reid Spencer7eb55b32006-11-02 01:53:59 +00003604<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3605</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003606
Reid Spencer7eb55b32006-11-02 01:53:59 +00003607<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003608
Reid Spencer7eb55b32006-11-02 01:53:59 +00003609<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003610<pre>
3611 &lt;result&gt; = urem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +00003612</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003613
Reid Spencer7eb55b32006-11-02 01:53:59 +00003614<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003615<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3616 division of its two arguments.</p>
3617
Reid Spencer7eb55b32006-11-02 01:53:59 +00003618<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003619<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003620 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3621 values. Both arguments must have identical types.</p>
3622
Reid Spencer7eb55b32006-11-02 01:53:59 +00003623<h5>Semantics:</h5>
3624<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003625 This instruction always performs an unsigned division to get the
3626 remainder.</p>
3627
Chris Lattner2f2427e2008-01-28 00:36:27 +00003628<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003629 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3630
Chris Lattner2f2427e2008-01-28 00:36:27 +00003631<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003632
Reid Spencer7eb55b32006-11-02 01:53:59 +00003633<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003634<pre>
3635 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +00003636</pre>
3637
3638</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003639
Reid Spencer7eb55b32006-11-02 01:53:59 +00003640<!-- _______________________________________________________________________ -->
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003641<div class="doc_subsubsection">
3642 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3643</div>
3644
Chris Lattner48b383b02003-11-25 01:02:51 +00003645<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003646
Chris Lattner48b383b02003-11-25 01:02:51 +00003647<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003648<pre>
Gabor Greif0f75ad02008-08-07 21:46:00 +00003649 &lt;result&gt; = srem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner48b383b02003-11-25 01:02:51 +00003650</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003651
Chris Lattner48b383b02003-11-25 01:02:51 +00003652<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003653<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3654 division of its two operands. This instruction can also take
3655 <a href="#t_vector">vector</a> versions of the values in which case the
3656 elements must be integers.</p>
Chris Lattnerb8f816e2008-01-04 04:33:49 +00003657
Chris Lattner48b383b02003-11-25 01:02:51 +00003658<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003659<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003660 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3661 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003662
Chris Lattner48b383b02003-11-25 01:02:51 +00003663<h5>Semantics:</h5>
Reid Spencer7eb55b32006-11-02 01:53:59 +00003664<p>This instruction returns the <i>remainder</i> of a division (where the result
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003665 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3666 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3667 a value. For more information about the difference,
3668 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3669 Math Forum</a>. For a table of how this is implemented in various languages,
3670 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3671 Wikipedia: modulo operation</a>.</p>
3672
Chris Lattner2f2427e2008-01-28 00:36:27 +00003673<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003674 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3675
Chris Lattner2f2427e2008-01-28 00:36:27 +00003676<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003677 Overflow also leads to undefined behavior; this is a rare case, but can
3678 occur, for example, by taking the remainder of a 32-bit division of
3679 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3680 lets srem be implemented using instructions that return both the result of
3681 the division and the remainder.)</p>
3682
Chris Lattner48b383b02003-11-25 01:02:51 +00003683<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003684<pre>
3685 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +00003686</pre>
3687
3688</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003689
Reid Spencer7eb55b32006-11-02 01:53:59 +00003690<!-- _______________________________________________________________________ -->
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003691<div class="doc_subsubsection">
3692 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3693
Reid Spencer7eb55b32006-11-02 01:53:59 +00003694<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003695
Reid Spencer7eb55b32006-11-02 01:53:59 +00003696<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003697<pre>
3698 &lt;result&gt; = frem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +00003699</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003700
Reid Spencer7eb55b32006-11-02 01:53:59 +00003701<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003702<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3703 its two operands.</p>
3704
Reid Spencer7eb55b32006-11-02 01:53:59 +00003705<h5>Arguments:</h5>
3706<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003707 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3708 floating point values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003709
Reid Spencer7eb55b32006-11-02 01:53:59 +00003710<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003711<p>This instruction returns the <i>remainder</i> of a division. The remainder
3712 has the same sign as the dividend.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003713
Reid Spencer7eb55b32006-11-02 01:53:59 +00003714<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003715<pre>
3716 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner48b383b02003-11-25 01:02:51 +00003717</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003718
Misha Brukman76307852003-11-08 01:05:38 +00003719</div>
Robert Bocchino820bc75b2006-02-17 21:18:08 +00003720
Reid Spencer2ab01932007-02-02 13:57:07 +00003721<!-- ======================================================================= -->
3722<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3723Operations</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003724
Reid Spencer2ab01932007-02-02 13:57:07 +00003725<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003726
3727<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3728 program. They are generally very efficient instructions and can commonly be
3729 strength reduced from other instructions. They require two operands of the
3730 same type, execute an operation on them, and produce a single value. The
3731 resulting value is the same type as its operands.</p>
3732
Reid Spencer2ab01932007-02-02 13:57:07 +00003733</div>
3734
Reid Spencer04e259b2007-01-31 21:39:12 +00003735<!-- _______________________________________________________________________ -->
3736<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3737Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003738
Reid Spencer04e259b2007-01-31 21:39:12 +00003739<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003740
Reid Spencer04e259b2007-01-31 21:39:12 +00003741<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003742<pre>
3743 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer04e259b2007-01-31 21:39:12 +00003744</pre>
Chris Lattnerf0e50112007-10-03 21:01:14 +00003745
Reid Spencer04e259b2007-01-31 21:39:12 +00003746<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003747<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3748 a specified number of bits.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +00003749
Reid Spencer04e259b2007-01-31 21:39:12 +00003750<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003751<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3752 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3753 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00003754
Reid Spencer04e259b2007-01-31 21:39:12 +00003755<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003756<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3757 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3758 is (statically or dynamically) negative or equal to or larger than the number
3759 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3760 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3761 shift amount in <tt>op2</tt>.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +00003762
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003763<h5>Example:</h5>
3764<pre>
Reid Spencer04e259b2007-01-31 21:39:12 +00003765 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3766 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3767 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerf0e50112007-10-03 21:01:14 +00003768 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang4dd832d2008-12-09 05:46:39 +00003769 &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 Spencer04e259b2007-01-31 21:39:12 +00003770</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003771
Reid Spencer04e259b2007-01-31 21:39:12 +00003772</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003773
Reid Spencer04e259b2007-01-31 21:39:12 +00003774<!-- _______________________________________________________________________ -->
3775<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3776Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003777
Reid Spencer04e259b2007-01-31 21:39:12 +00003778<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003779
Reid Spencer04e259b2007-01-31 21:39:12 +00003780<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003781<pre>
3782 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer04e259b2007-01-31 21:39:12 +00003783</pre>
3784
3785<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003786<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3787 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00003788
3789<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003790<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003791 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3792 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00003793
3794<h5>Semantics:</h5>
3795<p>This instruction always performs a logical shift right operation. The most
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003796 significant bits of the result will be filled with zero bits after the shift.
3797 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3798 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3799 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3800 shift amount in <tt>op2</tt>.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00003801
3802<h5>Example:</h5>
3803<pre>
3804 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3805 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3806 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3807 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerf0e50112007-10-03 21:01:14 +00003808 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang4dd832d2008-12-09 05:46:39 +00003809 &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 Spencer04e259b2007-01-31 21:39:12 +00003810</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003811
Reid Spencer04e259b2007-01-31 21:39:12 +00003812</div>
3813
Reid Spencer2ab01932007-02-02 13:57:07 +00003814<!-- _______________________________________________________________________ -->
Reid Spencer04e259b2007-01-31 21:39:12 +00003815<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3816Instruction</a> </div>
3817<div class="doc_text">
3818
3819<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003820<pre>
3821 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer04e259b2007-01-31 21:39:12 +00003822</pre>
3823
3824<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003825<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3826 operand shifted to the right a specified number of bits with sign
3827 extension.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00003828
3829<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003830<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003831 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3832 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00003833
3834<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003835<p>This instruction always performs an arithmetic shift right operation, The
3836 most significant bits of the result will be filled with the sign bit
3837 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3838 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3839 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3840 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00003841
3842<h5>Example:</h5>
3843<pre>
3844 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3845 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3846 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3847 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerf0e50112007-10-03 21:01:14 +00003848 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang4dd832d2008-12-09 05:46:39 +00003849 &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 Spencer04e259b2007-01-31 21:39:12 +00003850</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003851
Reid Spencer04e259b2007-01-31 21:39:12 +00003852</div>
3853
Chris Lattner2f7c9632001-06-06 20:29:01 +00003854<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00003855<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3856Instruction</a> </div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003857
Misha Brukman76307852003-11-08 01:05:38 +00003858<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003859
Chris Lattner2f7c9632001-06-06 20:29:01 +00003860<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003861<pre>
Gabor Greif0f75ad02008-08-07 21:46:00 +00003862 &lt;result&gt; = and &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003863</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003864
Chris Lattner2f7c9632001-06-06 20:29:01 +00003865<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003866<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3867 operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003868
Chris Lattner2f7c9632001-06-06 20:29:01 +00003869<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003870<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003871 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3872 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003873
Chris Lattner2f7c9632001-06-06 20:29:01 +00003874<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003875<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003876
Misha Brukman76307852003-11-08 01:05:38 +00003877<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +00003878 <tbody>
3879 <tr>
3880 <td>In0</td>
3881 <td>In1</td>
3882 <td>Out</td>
3883 </tr>
3884 <tr>
3885 <td>0</td>
3886 <td>0</td>
3887 <td>0</td>
3888 </tr>
3889 <tr>
3890 <td>0</td>
3891 <td>1</td>
3892 <td>0</td>
3893 </tr>
3894 <tr>
3895 <td>1</td>
3896 <td>0</td>
3897 <td>0</td>
3898 </tr>
3899 <tr>
3900 <td>1</td>
3901 <td>1</td>
3902 <td>1</td>
3903 </tr>
3904 </tbody>
3905</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003906
Chris Lattner2f7c9632001-06-06 20:29:01 +00003907<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003908<pre>
3909 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00003910 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3911 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003912</pre>
Misha Brukman76307852003-11-08 01:05:38 +00003913</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003914<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00003915<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003916
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003917<div class="doc_text">
3918
3919<h5>Syntax:</h5>
3920<pre>
3921 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3922</pre>
3923
3924<h5>Overview:</h5>
3925<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3926 two operands.</p>
3927
3928<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003929<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003930 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3931 values. Both arguments must have identical types.</p>
3932
Chris Lattner2f7c9632001-06-06 20:29:01 +00003933<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003934<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003935
Chris Lattner48b383b02003-11-25 01:02:51 +00003936<table border="1" cellspacing="0" cellpadding="4">
3937 <tbody>
3938 <tr>
3939 <td>In0</td>
3940 <td>In1</td>
3941 <td>Out</td>
3942 </tr>
3943 <tr>
3944 <td>0</td>
3945 <td>0</td>
3946 <td>0</td>
3947 </tr>
3948 <tr>
3949 <td>0</td>
3950 <td>1</td>
3951 <td>1</td>
3952 </tr>
3953 <tr>
3954 <td>1</td>
3955 <td>0</td>
3956 <td>1</td>
3957 </tr>
3958 <tr>
3959 <td>1</td>
3960 <td>1</td>
3961 <td>1</td>
3962 </tr>
3963 </tbody>
3964</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003965
Chris Lattner2f7c9632001-06-06 20:29:01 +00003966<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003967<pre>
3968 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00003969 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3970 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003971</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003972
Misha Brukman76307852003-11-08 01:05:38 +00003973</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003974
Chris Lattner2f7c9632001-06-06 20:29:01 +00003975<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00003976<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3977Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003978
Misha Brukman76307852003-11-08 01:05:38 +00003979<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003980
Chris Lattner2f7c9632001-06-06 20:29:01 +00003981<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003982<pre>
3983 &lt;result&gt; = xor &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003984</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003985
Chris Lattner2f7c9632001-06-06 20:29:01 +00003986<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003987<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3988 its two operands. The <tt>xor</tt> is used to implement the "one's
3989 complement" operation, which is the "~" operator in C.</p>
3990
Chris Lattner2f7c9632001-06-06 20:29:01 +00003991<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00003992<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003993 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3994 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003995
Chris Lattner2f7c9632001-06-06 20:29:01 +00003996<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003997<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003998
Chris Lattner48b383b02003-11-25 01:02:51 +00003999<table border="1" cellspacing="0" cellpadding="4">
4000 <tbody>
4001 <tr>
4002 <td>In0</td>
4003 <td>In1</td>
4004 <td>Out</td>
4005 </tr>
4006 <tr>
4007 <td>0</td>
4008 <td>0</td>
4009 <td>0</td>
4010 </tr>
4011 <tr>
4012 <td>0</td>
4013 <td>1</td>
4014 <td>1</td>
4015 </tr>
4016 <tr>
4017 <td>1</td>
4018 <td>0</td>
4019 <td>1</td>
4020 </tr>
4021 <tr>
4022 <td>1</td>
4023 <td>1</td>
4024 <td>0</td>
4025 </tr>
4026 </tbody>
4027</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004028
Chris Lattner2f7c9632001-06-06 20:29:01 +00004029<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004030<pre>
4031 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004032 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4033 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4034 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004035</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004036
Misha Brukman76307852003-11-08 01:05:38 +00004037</div>
Chris Lattner54611b42005-11-06 08:02:57 +00004038
Chris Lattner2f7c9632001-06-06 20:29:01 +00004039<!-- ======================================================================= -->
Eric Christopher455c5772009-12-05 02:46:03 +00004040<div class="doc_subsection">
Chris Lattnerce83bff2006-04-08 23:07:04 +00004041 <a name="vectorops">Vector Operations</a>
4042</div>
4043
4044<div class="doc_text">
4045
4046<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004047 target-independent manner. These instructions cover the element-access and
4048 vector-specific operations needed to process vectors effectively. While LLVM
4049 does directly support these vector operations, many sophisticated algorithms
4050 will want to use target-specific intrinsics to take full advantage of a
4051 specific target.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004052
4053</div>
4054
4055<!-- _______________________________________________________________________ -->
4056<div class="doc_subsubsection">
4057 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
4058</div>
4059
4060<div class="doc_text">
4061
4062<h5>Syntax:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004063<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004064 &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 Lattnerce83bff2006-04-08 23:07:04 +00004065</pre>
4066
4067<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004068<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4069 from a vector at a specified index.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004070
4071
4072<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004073<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4074 of <a href="#t_vector">vector</a> type. The second operand is an index
4075 indicating the position from which to extract the element. The index may be
4076 a variable.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004077
4078<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004079<p>The result is a scalar of the same type as the element type of
4080 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4081 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4082 results are undefined.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004083
4084<h5>Example:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004085<pre>
Gabor Greif03ab4dc2009-10-28 13:14:50 +00004086 &lt;result&gt; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004087</pre>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004088
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004089</div>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004090
4091<!-- _______________________________________________________________________ -->
4092<div class="doc_subsubsection">
4093 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4094</div>
4095
4096<div class="doc_text">
4097
4098<h5>Syntax:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004099<pre>
Dan Gohman43ba0672008-05-12 23:38:42 +00004100 &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 Lattnerce83bff2006-04-08 23:07:04 +00004101</pre>
4102
4103<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004104<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4105 vector at a specified index.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004106
4107<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004108<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4109 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4110 whose type must equal the element type of the first operand. The third
4111 operand is an index indicating the position at which to insert the value.
4112 The index may be a variable.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004113
4114<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004115<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4116 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4117 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4118 results are undefined.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004119
4120<h5>Example:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004121<pre>
Gabor Greif03ab4dc2009-10-28 13:14:50 +00004122 &lt;result&gt; = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004123</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004124
Chris Lattnerce83bff2006-04-08 23:07:04 +00004125</div>
4126
4127<!-- _______________________________________________________________________ -->
4128<div class="doc_subsubsection">
4129 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4130</div>
4131
4132<div class="doc_text">
4133
4134<h5>Syntax:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004135<pre>
Mon P Wang25f01062008-11-10 04:46:22 +00004136 &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 Lattnerce83bff2006-04-08 23:07:04 +00004137</pre>
4138
4139<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004140<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4141 from two input vectors, returning a vector with the same element type as the
4142 input and length that is the same as the shuffle mask.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004143
4144<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004145<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4146 with types that match each other. The third argument is a shuffle mask whose
4147 element type is always 'i32'. The result of the instruction is a vector
4148 whose length is the same as the shuffle mask and whose element type is the
4149 same as the element type of the first two operands.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004150
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004151<p>The shuffle mask operand is required to be a constant vector with either
4152 constant integer or undef values.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004153
4154<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004155<p>The elements of the two input vectors are numbered from left to right across
4156 both of the vectors. The shuffle mask operand specifies, for each element of
4157 the result vector, which element of the two input vectors the result element
4158 gets. The element selector may be undef (meaning "don't care") and the
4159 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004160
4161<h5>Example:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004162<pre>
Eric Christopher455c5772009-12-05 02:46:03 +00004163 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen5819f182007-04-22 01:17:39 +00004164 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christopher455c5772009-12-05 02:46:03 +00004165 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004166 &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 Christopher455c5772009-12-05 02:46:03 +00004167 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wang25f01062008-11-10 04:46:22 +00004168 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christopher455c5772009-12-05 02:46:03 +00004169 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wang25f01062008-11-10 04:46:22 +00004170 &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 Lattnerce83bff2006-04-08 23:07:04 +00004171</pre>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004172
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004173</div>
Tanya Lattnerb138bbe2006-04-14 19:24:33 +00004174
Chris Lattnerce83bff2006-04-08 23:07:04 +00004175<!-- ======================================================================= -->
Eric Christopher455c5772009-12-05 02:46:03 +00004176<div class="doc_subsection">
Dan Gohmanb9d66602008-05-12 23:51:09 +00004177 <a name="aggregateops">Aggregate Operations</a>
4178</div>
4179
4180<div class="doc_text">
4181
Chris Lattner392be582010-02-12 20:49:41 +00004182<p>LLVM supports several instructions for working with
4183 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004184
4185</div>
4186
4187<!-- _______________________________________________________________________ -->
4188<div class="doc_subsubsection">
4189 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4190</div>
4191
4192<div class="doc_text">
4193
4194<h5>Syntax:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004195<pre>
4196 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4197</pre>
4198
4199<h5>Overview:</h5>
Chris Lattner392be582010-02-12 20:49:41 +00004200<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4201 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004202
4203<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004204<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattner392be582010-02-12 20:49:41 +00004205 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4206 <a href="#t_array">array</a> type. The operands are constant indices to
4207 specify which value to extract in a similar manner as indices in a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004208 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004209
4210<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004211<p>The result is the value at the position in the aggregate specified by the
4212 index operands.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004213
4214<h5>Example:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004215<pre>
Gabor Greif03ab4dc2009-10-28 13:14:50 +00004216 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004217</pre>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004218
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004219</div>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004220
4221<!-- _______________________________________________________________________ -->
4222<div class="doc_subsubsection">
4223 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4224</div>
4225
4226<div class="doc_text">
4227
4228<h5>Syntax:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004229<pre>
Jeffrey Yasskinf991bbb2010-01-11 19:19:26 +00004230 &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 Gohmanb9d66602008-05-12 23:51:09 +00004231</pre>
4232
4233<h5>Overview:</h5>
Chris Lattner392be582010-02-12 20:49:41 +00004234<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4235 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004236
4237<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004238<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattner392be582010-02-12 20:49:41 +00004239 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4240 <a href="#t_array">array</a> type. The second operand is a first-class
4241 value to insert. The following operands are constant indices indicating
4242 the position at which to insert the value in a similar manner as indices in a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004243 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
4244 value to insert must have the same type as the value identified by the
4245 indices.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004246
4247<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004248<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4249 that of <tt>val</tt> except that the value at the position specified by the
4250 indices is that of <tt>elt</tt>.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004251
4252<h5>Example:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004253<pre>
Jeffrey Yasskinf991bbb2010-01-11 19:19:26 +00004254 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4255 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004256</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004257
Dan Gohmanb9d66602008-05-12 23:51:09 +00004258</div>
4259
4260
4261<!-- ======================================================================= -->
Eric Christopher455c5772009-12-05 02:46:03 +00004262<div class="doc_subsection">
Chris Lattner6ab66722006-08-15 00:45:58 +00004263 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner54611b42005-11-06 08:02:57 +00004264</div>
4265
Misha Brukman76307852003-11-08 01:05:38 +00004266<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +00004267
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004268<p>A key design point of an SSA-based representation is how it represents
4269 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandeza70c6df2009-10-26 23:44:29 +00004270 very simple. This section describes how to read, write, and allocate
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004271 memory in LLVM.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004272
Misha Brukman76307852003-11-08 01:05:38 +00004273</div>
Chris Lattner54611b42005-11-06 08:02:57 +00004274
Chris Lattner2f7c9632001-06-06 20:29:01 +00004275<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +00004276<div class="doc_subsubsection">
Chris Lattner54611b42005-11-06 08:02:57 +00004277 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4278</div>
4279
Misha Brukman76307852003-11-08 01:05:38 +00004280<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +00004281
Chris Lattner2f7c9632001-06-06 20:29:01 +00004282<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00004283<pre>
Dan Gohman2140a742010-05-28 01:14:11 +00004284 &lt;result&gt; = alloca &lt;type&gt;[, &lt;ty&gt; &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004285</pre>
Chris Lattner54611b42005-11-06 08:02:57 +00004286
Chris Lattner2f7c9632001-06-06 20:29:01 +00004287<h5>Overview:</h5>
Jeff Cohen5819f182007-04-22 01:17:39 +00004288<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004289 currently executing function, to be automatically released when this function
4290 returns to its caller. The object is always allocated in the generic address
4291 space (address space zero).</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004292
Chris Lattner2f7c9632001-06-06 20:29:01 +00004293<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004294<p>The '<tt>alloca</tt>' instruction
4295 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4296 runtime stack, returning a pointer of the appropriate type to the program.
4297 If "NumElements" is specified, it is the number of elements allocated,
4298 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4299 specified, the value result of the allocation is guaranteed to be aligned to
4300 at least that boundary. If not specified, or if zero, the target can choose
4301 to align the allocation on any convenient boundary compatible with the
4302 type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004303
Misha Brukman76307852003-11-08 01:05:38 +00004304<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004305
Chris Lattner2f7c9632001-06-06 20:29:01 +00004306<h5>Semantics:</h5>
Bill Wendling9ee6a312009-05-08 20:49:29 +00004307<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004308 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4309 memory is automatically released when the function returns. The
4310 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4311 variables that must have an address available. When the function returns
4312 (either with the <tt><a href="#i_ret">ret</a></tt>
4313 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4314 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004315
Chris Lattner2f7c9632001-06-06 20:29:01 +00004316<h5>Example:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00004317<pre>
Dan Gohman7a5acb52009-01-04 23:49:44 +00004318 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4319 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4320 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4321 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004322</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004323
Misha Brukman76307852003-11-08 01:05:38 +00004324</div>
Chris Lattner54611b42005-11-06 08:02:57 +00004325
Chris Lattner2f7c9632001-06-06 20:29:01 +00004326<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00004327<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
4328Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004329
Misha Brukman76307852003-11-08 01:05:38 +00004330<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004331
Chris Lattner095735d2002-05-06 03:03:22 +00004332<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004333<pre>
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004334 &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4335 &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4336 !&lt;index&gt; = !{ i32 1 }
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004337</pre>
4338
Chris Lattner095735d2002-05-06 03:03:22 +00004339<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004340<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004341
Chris Lattner095735d2002-05-06 03:03:22 +00004342<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004343<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4344 from which to load. The pointer must point to
4345 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4346 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00004347 number or order of execution of this <tt>load</tt> with other <a
4348 href="#volatile">volatile operations</a>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004349
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004350<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004351 operation (that is, the alignment of the memory address). A value of 0 or an
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004352 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004353 alignment for the target. It is the responsibility of the code emitter to
4354 ensure that the alignment information is correct. Overestimating the
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004355 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004356 produce less efficient code. An alignment of 1 is always safe.</p>
4357
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004358<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4359 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmana269a0a2010-03-01 17:41:39 +00004360 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004361 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4362 and code generator that this load is not expected to be reused in the cache.
4363 The code generator may select special instructions to save cache bandwidth,
Dan Gohmana269a0a2010-03-01 17:41:39 +00004364 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene9641d062010-02-16 20:50:18 +00004365
Chris Lattner095735d2002-05-06 03:03:22 +00004366<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004367<p>The location of memory pointed to is loaded. If the value being loaded is of
4368 scalar type then the number of bytes read does not exceed the minimum number
4369 of bytes needed to hold all bits of the type. For example, loading an
4370 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4371 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4372 is undefined if the value was not originally written using a store of the
4373 same type.</p>
4374
Chris Lattner095735d2002-05-06 03:03:22 +00004375<h5>Examples:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004376<pre>
4377 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4378 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004379 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner095735d2002-05-06 03:03:22 +00004380</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004381
Misha Brukman76307852003-11-08 01:05:38 +00004382</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004383
Chris Lattner095735d2002-05-06 03:03:22 +00004384<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +00004385<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4386Instruction</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004387
Reid Spencera89fb182006-11-09 21:18:01 +00004388<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004389
Chris Lattner095735d2002-05-06 03:03:22 +00004390<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004391<pre>
David Greene9641d062010-02-16 20:50:18 +00004392 store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !<index>] <i>; yields {void}</i>
4393 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !<index>] <i>; yields {void}</i>
Chris Lattner095735d2002-05-06 03:03:22 +00004394</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004395
Chris Lattner095735d2002-05-06 03:03:22 +00004396<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004397<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004398
Chris Lattner095735d2002-05-06 03:03:22 +00004399<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004400<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4401 and an address at which to store it. The type of the
4402 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4403 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00004404 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
4405 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4406 order of execution of this <tt>store</tt> with other <a
4407 href="#volatile">volatile operations</a>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004408
4409<p>The optional constant "align" argument specifies the alignment of the
4410 operation (that is, the alignment of the memory address). A value of 0 or an
4411 omitted "align" argument means that the operation has the preferential
4412 alignment for the target. It is the responsibility of the code emitter to
4413 ensure that the alignment information is correct. Overestimating the
4414 alignment results in an undefined behavior. Underestimating the alignment may
4415 produce less efficient code. An alignment of 1 is always safe.</p>
4416
David Greene9641d062010-02-16 20:50:18 +00004417<p>The optional !nontemporal metadata must reference a single metatadata
4418 name <index> corresponding to a metadata node with one i32 entry of
Dan Gohmana269a0a2010-03-01 17:41:39 +00004419 value 1. The existence of the !nontemporal metatadata on the
David Greene9641d062010-02-16 20:50:18 +00004420 instruction tells the optimizer and code generator that this load is
4421 not expected to be reused in the cache. The code generator may
4422 select special instructions to save cache bandwidth, such as the
Dan Gohmana269a0a2010-03-01 17:41:39 +00004423 MOVNT instruction on x86.</p>
David Greene9641d062010-02-16 20:50:18 +00004424
4425
Chris Lattner48b383b02003-11-25 01:02:51 +00004426<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004427<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4428 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4429 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4430 does not exceed the minimum number of bytes needed to hold all bits of the
4431 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4432 writing a value of a type like <tt>i20</tt> with a size that is not an
4433 integral number of bytes, it is unspecified what happens to the extra bits
4434 that do not belong to the type, but they will typically be overwritten.</p>
4435
Chris Lattner095735d2002-05-06 03:03:22 +00004436<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004437<pre>
4438 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8830ffe2007-10-22 05:10:05 +00004439 store i32 3, i32* %ptr <i>; yields {void}</i>
4440 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner095735d2002-05-06 03:03:22 +00004441</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004442
Reid Spencer443460a2006-11-09 21:15:49 +00004443</div>
4444
Chris Lattner095735d2002-05-06 03:03:22 +00004445<!-- _______________________________________________________________________ -->
Chris Lattner33fd7022004-04-05 01:30:49 +00004446<div class="doc_subsubsection">
4447 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4448</div>
4449
Misha Brukman76307852003-11-08 01:05:38 +00004450<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004451
Chris Lattner590645f2002-04-14 06:13:44 +00004452<h5>Syntax:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00004453<pre>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00004454 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohman1639c392009-07-27 21:53:46 +00004455 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Chris Lattner33fd7022004-04-05 01:30:49 +00004456</pre>
4457
Chris Lattner590645f2002-04-14 06:13:44 +00004458<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004459<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattner392be582010-02-12 20:49:41 +00004460 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4461 It performs address calculation only and does not access memory.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00004462
Chris Lattner590645f2002-04-14 06:13:44 +00004463<h5>Arguments:</h5>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00004464<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnera40b9122009-07-29 06:44:13 +00004465 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004466 elements of the aggregate object are indexed. The interpretation of each
4467 index is dependent on the type being indexed into. The first index always
4468 indexes the pointer value given as the first argument, the second index
4469 indexes a value of the type pointed to (not necessarily the value directly
4470 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattner392be582010-02-12 20:49:41 +00004471 indexed into must be a pointer value, subsequent types can be arrays,
4472 vectors, structs and unions. Note that subsequent types being indexed into
4473 can never be pointers, since that would require loading the pointer before
4474 continuing calculation.</p>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00004475
4476<p>The type of each index argument depends on the type it is indexing into.
Chris Lattner392be582010-02-12 20:49:41 +00004477 When indexing into a (optionally packed) structure or union, only <tt>i32</tt>
4478 integer <b>constants</b> are allowed. When indexing into an array, pointer
4479 or vector, integers of any width are allowed, and they are not required to be
Chris Lattnera40b9122009-07-29 06:44:13 +00004480 constant.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00004481
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004482<p>For example, let's consider a C code fragment and how it gets compiled to
4483 LLVM:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00004484
Bill Wendling3716c5d2007-05-29 09:04:49 +00004485<div class="doc_code">
Chris Lattner33fd7022004-04-05 01:30:49 +00004486<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00004487struct RT {
4488 char A;
Chris Lattnera446f1b2007-05-29 15:43:56 +00004489 int B[10][20];
Bill Wendling3716c5d2007-05-29 09:04:49 +00004490 char C;
4491};
4492struct ST {
Chris Lattnera446f1b2007-05-29 15:43:56 +00004493 int X;
Bill Wendling3716c5d2007-05-29 09:04:49 +00004494 double Y;
4495 struct RT Z;
4496};
Chris Lattner33fd7022004-04-05 01:30:49 +00004497
Chris Lattnera446f1b2007-05-29 15:43:56 +00004498int *foo(struct ST *s) {
Bill Wendling3716c5d2007-05-29 09:04:49 +00004499 return &amp;s[1].Z.B[5][13];
4500}
Chris Lattner33fd7022004-04-05 01:30:49 +00004501</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00004502</div>
Chris Lattner33fd7022004-04-05 01:30:49 +00004503
Misha Brukman76307852003-11-08 01:05:38 +00004504<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00004505
Bill Wendling3716c5d2007-05-29 09:04:49 +00004506<div class="doc_code">
Chris Lattner33fd7022004-04-05 01:30:49 +00004507<pre>
Chris Lattnerbc088212009-01-11 20:53:49 +00004508%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4509%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Chris Lattner33fd7022004-04-05 01:30:49 +00004510
Dan Gohman6b867702009-07-25 02:23:48 +00004511define i32* @foo(%ST* %s) {
Bill Wendling3716c5d2007-05-29 09:04:49 +00004512entry:
4513 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4514 ret i32* %reg
4515}
Chris Lattner33fd7022004-04-05 01:30:49 +00004516</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00004517</div>
Chris Lattner33fd7022004-04-05 01:30:49 +00004518
Chris Lattner590645f2002-04-14 06:13:44 +00004519<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004520<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004521 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4522 }</tt>' type, a structure. The second index indexes into the third element
4523 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4524 i8 }</tt>' type, another structure. The third index indexes into the second
4525 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4526 array. The two dimensions of the array are subscripted into, yielding an
4527 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4528 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00004529
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004530<p>Note that it is perfectly legal to index partially through a structure,
4531 returning a pointer to an inner element. Because of this, the LLVM code for
4532 the given testcase is equivalent to:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00004533
4534<pre>
Dan Gohman6b867702009-07-25 02:23:48 +00004535 define i32* @foo(%ST* %s) {
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004536 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
Jeff Cohen5819f182007-04-22 01:17:39 +00004537 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4538 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004539 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4540 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4541 ret i32* %t5
Chris Lattner33fd7022004-04-05 01:30:49 +00004542 }
Chris Lattnera8292f32002-05-06 22:08:29 +00004543</pre>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00004544
Dan Gohman1639c392009-07-27 21:53:46 +00004545<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman57255802010-04-23 15:23:32 +00004546 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
4547 base pointer is not an <i>in bounds</i> address of an allocated object,
4548 or if any of the addresses that would be formed by successive addition of
4549 the offsets implied by the indices to the base address with infinitely
4550 precise arithmetic are not an <i>in bounds</i> address of that allocated
4551 object. The <i>in bounds</i> addresses for an allocated object are all
4552 the addresses that point into the object, plus the address one byte past
4553 the end.</p>
Dan Gohman1639c392009-07-27 21:53:46 +00004554
4555<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4556 the base address with silently-wrapping two's complement arithmetic, and
4557 the result value of the <tt>getelementptr</tt> may be outside the object
4558 pointed to by the base pointer. The result value may not necessarily be
4559 used to access memory though, even if it happens to point into allocated
4560 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4561 section for more information.</p>
4562
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004563<p>The getelementptr instruction is often confusing. For some more insight into
4564 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner6ab66722006-08-15 00:45:58 +00004565
Chris Lattner590645f2002-04-14 06:13:44 +00004566<h5>Example:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00004567<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004568 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00004569 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4570 <i>; yields i8*:vptr</i>
Dan Gohmanef9462f2008-10-14 16:51:45 +00004571 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijman0e268272008-10-13 13:44:15 +00004572 <i>; yields i8*:eptr</i>
4573 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta0c155e62009-04-25 07:27:44 +00004574 <i>; yields i32*:iptr</i>
Sanjiv Gupta77abea02009-04-24 16:38:13 +00004575 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattner33fd7022004-04-05 01:30:49 +00004576</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004577
Chris Lattner33fd7022004-04-05 01:30:49 +00004578</div>
Reid Spencer443460a2006-11-09 21:15:49 +00004579
Chris Lattner2f7c9632001-06-06 20:29:01 +00004580<!-- ======================================================================= -->
Reid Spencer97c5fa42006-11-08 01:18:52 +00004581<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman76307852003-11-08 01:05:38 +00004582</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004583
Misha Brukman76307852003-11-08 01:05:38 +00004584<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004585
Reid Spencer97c5fa42006-11-08 01:18:52 +00004586<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004587 which all take a single operand and a type. They perform various bit
4588 conversions on the operand.</p>
4589
Misha Brukman76307852003-11-08 01:05:38 +00004590</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00004591
Chris Lattnera8292f32002-05-06 22:08:29 +00004592<!-- _______________________________________________________________________ -->
Chris Lattnerb53c28d2004-03-12 05:50:16 +00004593<div class="doc_subsubsection">
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004594 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4595</div>
4596<div class="doc_text">
4597
4598<h5>Syntax:</h5>
4599<pre>
4600 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4601</pre>
4602
4603<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004604<p>The '<tt>trunc</tt>' instruction truncates its operand to the
4605 type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004606
4607<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004608<p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4609 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4610 size and type of the result, which must be
4611 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4612 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4613 allowed.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004614
4615<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004616<p>The '<tt>trunc</tt>' instruction truncates the high order bits
4617 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4618 source size must be larger than the destination size, <tt>trunc</tt> cannot
4619 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004620
4621<h5>Example:</h5>
4622<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004623 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
Reid Spencer36a15422007-01-12 03:35:51 +00004624 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00004625 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004626</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004627
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004628</div>
4629
4630<!-- _______________________________________________________________________ -->
4631<div class="doc_subsubsection">
4632 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4633</div>
4634<div class="doc_text">
4635
4636<h5>Syntax:</h5>
4637<pre>
4638 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4639</pre>
4640
4641<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004642<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004643 <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004644
4645
4646<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004647<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004648 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4649 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christopher455c5772009-12-05 02:46:03 +00004650 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004651 <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004652
4653<h5>Semantics:</h5>
4654<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004655 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004656
Reid Spencer07c9c682007-01-12 15:46:11 +00004657<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004658
4659<h5>Example:</h5>
4660<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004661 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencer36a15422007-01-12 03:35:51 +00004662 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004663</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004664
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004665</div>
4666
4667<!-- _______________________________________________________________________ -->
4668<div class="doc_subsubsection">
4669 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4670</div>
4671<div class="doc_text">
4672
4673<h5>Syntax:</h5>
4674<pre>
4675 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4676</pre>
4677
4678<h5>Overview:</h5>
4679<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4680
4681<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004682<p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004683 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4684 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christopher455c5772009-12-05 02:46:03 +00004685 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004686 <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004687
4688<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004689<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4690 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4691 of the type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004692
Reid Spencer36a15422007-01-12 03:35:51 +00004693<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004694
4695<h5>Example:</h5>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004696<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004697 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencer36a15422007-01-12 03:35:51 +00004698 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004699</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004700
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004701</div>
4702
4703<!-- _______________________________________________________________________ -->
4704<div class="doc_subsubsection">
Reid Spencer2e2740d2006-11-09 21:48:10 +00004705 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4706</div>
4707
4708<div class="doc_text">
4709
4710<h5>Syntax:</h5>
Reid Spencer2e2740d2006-11-09 21:48:10 +00004711<pre>
4712 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4713</pre>
4714
4715<h5>Overview:</h5>
4716<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004717 <tt>ty2</tt>.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +00004718
4719<h5>Arguments:</h5>
4720<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004721 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4722 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christopher455c5772009-12-05 02:46:03 +00004723 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004724 <i>no-op cast</i>.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +00004725
4726<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004727<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher455c5772009-12-05 02:46:03 +00004728 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004729 <a href="#t_floating">floating point</a> type. If the value cannot fit
4730 within the destination type, <tt>ty2</tt>, then the results are
4731 undefined.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +00004732
4733<h5>Example:</h5>
4734<pre>
4735 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4736 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4737</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004738
Reid Spencer2e2740d2006-11-09 21:48:10 +00004739</div>
4740
4741<!-- _______________________________________________________________________ -->
4742<div class="doc_subsubsection">
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004743 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4744</div>
4745<div class="doc_text">
4746
4747<h5>Syntax:</h5>
4748<pre>
4749 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4750</pre>
4751
4752<h5>Overview:</h5>
4753<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004754 floating point value.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004755
4756<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004757<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004758 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4759 a <a href="#t_floating">floating point</a> type to cast it to. The source
4760 type must be smaller than the destination type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004761
4762<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +00004763<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004764 <a href="#t_floating">floating point</a> type to a larger
4765 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4766 used to make a <i>no-op cast</i> because it always changes bits. Use
4767 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004768
4769<h5>Example:</h5>
4770<pre>
4771 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4772 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4773</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004774
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004775</div>
4776
4777<!-- _______________________________________________________________________ -->
4778<div class="doc_subsubsection">
Reid Spencer2eadb532007-01-21 00:29:26 +00004779 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004780</div>
4781<div class="doc_text">
4782
4783<h5>Syntax:</h5>
4784<pre>
Reid Spencer753163d2007-07-31 14:40:14 +00004785 &lt;result&gt; = fptoui &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004786</pre>
4787
4788<h5>Overview:</h5>
Reid Spencer753163d2007-07-31 14:40:14 +00004789<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004790 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004791
4792<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004793<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4794 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4795 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4796 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4797 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004798
4799<h5>Semantics:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004800<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004801 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4802 towards zero) unsigned integer value. If the value cannot fit
4803 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004804
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004805<h5>Example:</h5>
4806<pre>
Reid Spencer753163d2007-07-31 14:40:14 +00004807 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner5b95a172007-09-22 03:17:52 +00004808 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00004809 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004810</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004811
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004812</div>
4813
4814<!-- _______________________________________________________________________ -->
4815<div class="doc_subsubsection">
Reid Spencer51b07252006-11-09 23:03:26 +00004816 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004817</div>
4818<div class="doc_text">
4819
4820<h5>Syntax:</h5>
4821<pre>
Reid Spencer51b07252006-11-09 23:03:26 +00004822 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004823</pre>
4824
4825<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004826<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004827 <a href="#t_floating">floating point</a> <tt>value</tt> to
4828 type <tt>ty2</tt>.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00004829
Chris Lattnera8292f32002-05-06 22:08:29 +00004830<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004831<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4832 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4833 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4834 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4835 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00004836
Chris Lattnera8292f32002-05-06 22:08:29 +00004837<h5>Semantics:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004838<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004839 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4840 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4841 the results are undefined.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00004842
Chris Lattner70de6632001-07-09 00:26:23 +00004843<h5>Example:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00004844<pre>
Reid Spencer36a15422007-01-12 03:35:51 +00004845 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner5b95a172007-09-22 03:17:52 +00004846 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00004847 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004848</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004849
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004850</div>
4851
4852<!-- _______________________________________________________________________ -->
4853<div class="doc_subsubsection">
Reid Spencer51b07252006-11-09 23:03:26 +00004854 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004855</div>
4856<div class="doc_text">
4857
4858<h5>Syntax:</h5>
4859<pre>
Reid Spencer51b07252006-11-09 23:03:26 +00004860 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004861</pre>
4862
4863<h5>Overview:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +00004864<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004865 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004866
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004867<h5>Arguments:</h5>
Nate Begemand4d45c22007-11-17 03:58:34 +00004868<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004869 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4870 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4871 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4872 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004873
4874<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +00004875<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004876 integer quantity and converts it to the corresponding floating point
4877 value. If the value cannot fit in the floating point value, the results are
4878 undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004879
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004880<h5>Example:</h5>
4881<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004882 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohmanef9462f2008-10-14 16:51:45 +00004883 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004884</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004885
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004886</div>
4887
4888<!-- _______________________________________________________________________ -->
4889<div class="doc_subsubsection">
Reid Spencer51b07252006-11-09 23:03:26 +00004890 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004891</div>
4892<div class="doc_text">
4893
4894<h5>Syntax:</h5>
4895<pre>
Reid Spencer51b07252006-11-09 23:03:26 +00004896 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004897</pre>
4898
4899<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004900<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4901 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004902
4903<h5>Arguments:</h5>
Nate Begemand4d45c22007-11-17 03:58:34 +00004904<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004905 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4906 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4907 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4908 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004909
4910<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004911<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4912 quantity and converts it to the corresponding floating point value. If the
4913 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004914
4915<h5>Example:</h5>
4916<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004917 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohmanef9462f2008-10-14 16:51:45 +00004918 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004919</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004920
Reid Spencer59b6b7d2006-11-08 01:11:31 +00004921</div>
4922
4923<!-- _______________________________________________________________________ -->
4924<div class="doc_subsubsection">
Reid Spencerb7344ff2006-11-11 21:00:47 +00004925 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4926</div>
4927<div class="doc_text">
4928
4929<h5>Syntax:</h5>
4930<pre>
4931 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4932</pre>
4933
4934<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004935<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4936 the integer type <tt>ty2</tt>.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004937
4938<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004939<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4940 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4941 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004942
4943<h5>Semantics:</h5>
4944<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004945 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4946 truncating or zero extending that value to the size of the integer type. If
4947 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4948 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4949 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4950 change.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004951
4952<h5>Example:</h5>
4953<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +00004954 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4955 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004956</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004957
Reid Spencerb7344ff2006-11-11 21:00:47 +00004958</div>
4959
4960<!-- _______________________________________________________________________ -->
4961<div class="doc_subsubsection">
4962 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4963</div>
4964<div class="doc_text">
4965
4966<h5>Syntax:</h5>
4967<pre>
4968 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4969</pre>
4970
4971<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004972<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4973 pointer type, <tt>ty2</tt>.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004974
4975<h5>Arguments:</h5>
Duncan Sands16f122e2007-03-30 12:22:09 +00004976<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004977 value to cast, and a type to cast it to, which must be a
4978 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004979
4980<h5>Semantics:</h5>
4981<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004982 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4983 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4984 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4985 than the size of a pointer then a zero extension is done. If they are the
4986 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004987
4988<h5>Example:</h5>
4989<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +00004990 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00004991 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4992 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Reid Spencerb7344ff2006-11-11 21:00:47 +00004993</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004994
Reid Spencerb7344ff2006-11-11 21:00:47 +00004995</div>
4996
4997<!-- _______________________________________________________________________ -->
4998<div class="doc_subsubsection">
Reid Spencer5b950642006-11-11 23:08:07 +00004999 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005000</div>
5001<div class="doc_text">
5002
5003<h5>Syntax:</h5>
5004<pre>
Reid Spencer5b950642006-11-11 23:08:07 +00005005 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005006</pre>
5007
5008<h5>Overview:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +00005009<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005010 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005011
5012<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005013<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5014 non-aggregate first class value, and a type to cast it to, which must also be
5015 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5016 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5017 identical. If the source type is a pointer, the destination type must also be
5018 a pointer. This instruction supports bitwise conversion of vectors to
5019 integers and to vectors of other types (as long as they have the same
5020 size).</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005021
5022<h5>Semantics:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +00005023<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005024 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5025 this conversion. The conversion is done as if the <tt>value</tt> had been
5026 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
5027 be converted to other pointer types with this instruction. To convert
5028 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5029 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005030
5031<h5>Example:</h5>
5032<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005033 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005034 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Eric Christopher455c5772009-12-05 02:46:03 +00005035 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Chris Lattner70de6632001-07-09 00:26:23 +00005036</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005037
Misha Brukman76307852003-11-08 01:05:38 +00005038</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005039
Reid Spencer97c5fa42006-11-08 01:18:52 +00005040<!-- ======================================================================= -->
5041<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005042
Reid Spencer97c5fa42006-11-08 01:18:52 +00005043<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005044
5045<p>The instructions in this category are the "miscellaneous" instructions, which
5046 defy better classification.</p>
5047
Reid Spencer97c5fa42006-11-08 01:18:52 +00005048</div>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005049
5050<!-- _______________________________________________________________________ -->
5051<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5052</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005053
Reid Spencerc828a0e2006-11-18 21:50:54 +00005054<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005055
Reid Spencerc828a0e2006-11-18 21:50:54 +00005056<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005057<pre>
5058 &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 Spencerc828a0e2006-11-18 21:50:54 +00005059</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005060
Reid Spencerc828a0e2006-11-18 21:50:54 +00005061<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005062<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5063 boolean values based on comparison of its two integer, integer vector, or
5064 pointer operands.</p>
5065
Reid Spencerc828a0e2006-11-18 21:50:54 +00005066<h5>Arguments:</h5>
5067<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005068 the condition code indicating the kind of comparison to perform. It is not a
5069 value, just a keyword. The possible condition code are:</p>
5070
Reid Spencerc828a0e2006-11-18 21:50:54 +00005071<ol>
5072 <li><tt>eq</tt>: equal</li>
5073 <li><tt>ne</tt>: not equal </li>
5074 <li><tt>ugt</tt>: unsigned greater than</li>
5075 <li><tt>uge</tt>: unsigned greater or equal</li>
5076 <li><tt>ult</tt>: unsigned less than</li>
5077 <li><tt>ule</tt>: unsigned less or equal</li>
5078 <li><tt>sgt</tt>: signed greater than</li>
5079 <li><tt>sge</tt>: signed greater or equal</li>
5080 <li><tt>slt</tt>: signed less than</li>
5081 <li><tt>sle</tt>: signed less or equal</li>
5082</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005083
Chris Lattnerc0f423a2007-01-15 01:54:13 +00005084<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005085 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5086 typed. They must also be identical types.</p>
5087
Reid Spencerc828a0e2006-11-18 21:50:54 +00005088<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005089<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5090 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00005091 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005092 result, as follows:</p>
5093
Reid Spencerc828a0e2006-11-18 21:50:54 +00005094<ol>
Eric Christopher455c5772009-12-05 02:46:03 +00005095 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005096 <tt>false</tt> otherwise. No sign interpretation is necessary or
5097 performed.</li>
5098
Eric Christopher455c5772009-12-05 02:46:03 +00005099 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005100 <tt>false</tt> otherwise. No sign interpretation is necessary or
5101 performed.</li>
5102
Reid Spencerc828a0e2006-11-18 21:50:54 +00005103 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005104 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5105
Reid Spencerc828a0e2006-11-18 21:50:54 +00005106 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005107 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5108 to <tt>op2</tt>.</li>
5109
Reid Spencerc828a0e2006-11-18 21:50:54 +00005110 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005111 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5112
Reid Spencerc828a0e2006-11-18 21:50:54 +00005113 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005114 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5115
Reid Spencerc828a0e2006-11-18 21:50:54 +00005116 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005117 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5118
Reid Spencerc828a0e2006-11-18 21:50:54 +00005119 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005120 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5121 to <tt>op2</tt>.</li>
5122
Reid Spencerc828a0e2006-11-18 21:50:54 +00005123 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005124 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5125
Reid Spencerc828a0e2006-11-18 21:50:54 +00005126 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005127 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005128</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005129
Reid Spencerc828a0e2006-11-18 21:50:54 +00005130<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005131 values are compared as if they were integers.</p>
5132
5133<p>If the operands are integer vectors, then they are compared element by
5134 element. The result is an <tt>i1</tt> vector with the same number of elements
5135 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005136
5137<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005138<pre>
5139 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005140 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
5141 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
5142 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5143 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5144 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005145</pre>
Dan Gohmana5127ab2009-01-22 01:39:38 +00005146
5147<p>Note that the code generator does not yet support vector types with
5148 the <tt>icmp</tt> instruction.</p>
5149
Reid Spencerc828a0e2006-11-18 21:50:54 +00005150</div>
5151
5152<!-- _______________________________________________________________________ -->
5153<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5154</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005155
Reid Spencerc828a0e2006-11-18 21:50:54 +00005156<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005157
Reid Spencerc828a0e2006-11-18 21:50:54 +00005158<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005159<pre>
5160 &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 Spencerc828a0e2006-11-18 21:50:54 +00005161</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005162
Reid Spencerc828a0e2006-11-18 21:50:54 +00005163<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005164<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5165 values based on comparison of its operands.</p>
5166
5167<p>If the operands are floating point scalars, then the result type is a boolean
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00005168(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005169
5170<p>If the operands are floating point vectors, then the result type is a vector
5171 of boolean with the same number of elements as the operands being
5172 compared.</p>
5173
Reid Spencerc828a0e2006-11-18 21:50:54 +00005174<h5>Arguments:</h5>
5175<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005176 the condition code indicating the kind of comparison to perform. It is not a
5177 value, just a keyword. The possible condition code are:</p>
5178
Reid Spencerc828a0e2006-11-18 21:50:54 +00005179<ol>
Reid Spencerf69acf32006-11-19 03:00:14 +00005180 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005181 <li><tt>oeq</tt>: ordered and equal</li>
5182 <li><tt>ogt</tt>: ordered and greater than </li>
5183 <li><tt>oge</tt>: ordered and greater than or equal</li>
5184 <li><tt>olt</tt>: ordered and less than </li>
5185 <li><tt>ole</tt>: ordered and less than or equal</li>
5186 <li><tt>one</tt>: ordered and not equal</li>
5187 <li><tt>ord</tt>: ordered (no nans)</li>
5188 <li><tt>ueq</tt>: unordered or equal</li>
5189 <li><tt>ugt</tt>: unordered or greater than </li>
5190 <li><tt>uge</tt>: unordered or greater than or equal</li>
5191 <li><tt>ult</tt>: unordered or less than </li>
5192 <li><tt>ule</tt>: unordered or less than or equal</li>
5193 <li><tt>une</tt>: unordered or not equal</li>
5194 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerf69acf32006-11-19 03:00:14 +00005195 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005196</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005197
Jeff Cohen222a8a42007-04-29 01:07:00 +00005198<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005199 <i>unordered</i> means that either operand may be a QNAN.</p>
5200
5201<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5202 a <a href="#t_floating">floating point</a> type or
5203 a <a href="#t_vector">vector</a> of floating point type. They must have
5204 identical types.</p>
5205
Reid Spencerc828a0e2006-11-18 21:50:54 +00005206<h5>Semantics:</h5>
Gabor Greif0f75ad02008-08-07 21:46:00 +00005207<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005208 according to the condition code given as <tt>cond</tt>. If the operands are
5209 vectors, then the vectors are compared element by element. Each comparison
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00005210 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005211 follows:</p>
5212
Reid Spencerc828a0e2006-11-18 21:50:54 +00005213<ol>
5214 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005215
Eric Christopher455c5772009-12-05 02:46:03 +00005216 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005217 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5218
Reid Spencerf69acf32006-11-19 03:00:14 +00005219 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Dan Gohmana269a0a2010-03-01 17:41:39 +00005220 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005221
Eric Christopher455c5772009-12-05 02:46:03 +00005222 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005223 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5224
Eric Christopher455c5772009-12-05 02:46:03 +00005225 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005226 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5227
Eric Christopher455c5772009-12-05 02:46:03 +00005228 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005229 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5230
Eric Christopher455c5772009-12-05 02:46:03 +00005231 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005232 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5233
Reid Spencerf69acf32006-11-19 03:00:14 +00005234 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005235
Eric Christopher455c5772009-12-05 02:46:03 +00005236 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005237 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5238
Eric Christopher455c5772009-12-05 02:46:03 +00005239 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005240 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5241
Eric Christopher455c5772009-12-05 02:46:03 +00005242 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005243 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5244
Eric Christopher455c5772009-12-05 02:46:03 +00005245 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005246 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5247
Eric Christopher455c5772009-12-05 02:46:03 +00005248 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005249 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5250
Eric Christopher455c5772009-12-05 02:46:03 +00005251 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005252 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5253
Reid Spencerf69acf32006-11-19 03:00:14 +00005254 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005255
Reid Spencerc828a0e2006-11-18 21:50:54 +00005256 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5257</ol>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005258
5259<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005260<pre>
5261 &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanc579d972008-09-09 01:02:47 +00005262 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5263 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5264 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +00005265</pre>
Dan Gohmana5127ab2009-01-22 01:39:38 +00005266
5267<p>Note that the code generator does not yet support vector types with
5268 the <tt>fcmp</tt> instruction.</p>
5269
Reid Spencerc828a0e2006-11-18 21:50:54 +00005270</div>
5271
Reid Spencer97c5fa42006-11-08 01:18:52 +00005272<!-- _______________________________________________________________________ -->
Nate Begemand2195702008-05-12 19:01:56 +00005273<div class="doc_subsubsection">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005274 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5275</div>
5276
Reid Spencer97c5fa42006-11-08 01:18:52 +00005277<div class="doc_text">
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005278
Reid Spencer97c5fa42006-11-08 01:18:52 +00005279<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005280<pre>
5281 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
5282</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005283
Reid Spencer97c5fa42006-11-08 01:18:52 +00005284<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005285<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
5286 SSA graph representing the function.</p>
5287
Reid Spencer97c5fa42006-11-08 01:18:52 +00005288<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005289<p>The type of the incoming values is specified with the first type field. After
5290 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5291 one pair for each predecessor basic block of the current block. Only values
5292 of <a href="#t_firstclass">first class</a> type may be used as the value
5293 arguments to the PHI node. Only labels may be used as the label
5294 arguments.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005295
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005296<p>There must be no non-phi instructions between the start of a basic block and
5297 the PHI instructions: i.e. PHI instructions must be first in a basic
5298 block.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005299
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005300<p>For the purposes of the SSA form, the use of each incoming value is deemed to
5301 occur on the edge from the corresponding predecessor block to the current
5302 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5303 value on the same edge).</p>
Jay Foad1a4eea52009-06-03 10:20:10 +00005304
Reid Spencer97c5fa42006-11-08 01:18:52 +00005305<h5>Semantics:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005306<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005307 specified by the pair corresponding to the predecessor basic block that
5308 executed just prior to the current block.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005309
Reid Spencer97c5fa42006-11-08 01:18:52 +00005310<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00005311<pre>
5312Loop: ; Infinite loop that counts from 0 on up...
5313 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5314 %nextindvar = add i32 %indvar, 1
5315 br label %Loop
5316</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005317
Reid Spencer97c5fa42006-11-08 01:18:52 +00005318</div>
5319
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005320<!-- _______________________________________________________________________ -->
5321<div class="doc_subsubsection">
5322 <a name="i_select">'<tt>select</tt>' Instruction</a>
5323</div>
5324
5325<div class="doc_text">
5326
5327<h5>Syntax:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005328<pre>
Dan Gohmanc579d972008-09-09 01:02:47 +00005329 &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>
5330
Dan Gohmanef9462f2008-10-14 16:51:45 +00005331 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005332</pre>
5333
5334<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005335<p>The '<tt>select</tt>' instruction is used to choose one value based on a
5336 condition, without branching.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005337
5338
5339<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005340<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5341 values indicating the condition, and two values of the
5342 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5343 vectors and the condition is a scalar, then entire vectors are selected, not
5344 individual elements.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005345
5346<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005347<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5348 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005349
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005350<p>If the condition is a vector of i1, then the value arguments must be vectors
5351 of the same size, and the selection is done element by element.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005352
5353<h5>Example:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005354<pre>
Reid Spencer36a15422007-01-12 03:35:51 +00005355 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005356</pre>
Dan Gohmana5127ab2009-01-22 01:39:38 +00005357
5358<p>Note that the code generator does not yet support conditions
5359 with vector type.</p>
5360
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005361</div>
5362
Robert Bocchinof72fdfe2006-01-15 20:48:27 +00005363<!-- _______________________________________________________________________ -->
5364<div class="doc_subsubsection">
Chris Lattnere23c1392005-05-06 05:47:36 +00005365 <a name="i_call">'<tt>call</tt>' Instruction</a>
5366</div>
5367
Misha Brukman76307852003-11-08 01:05:38 +00005368<div class="doc_text">
Chris Lattnere23c1392005-05-06 05:47:36 +00005369
Chris Lattner2f7c9632001-06-06 20:29:01 +00005370<h5>Syntax:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00005371<pre>
Devang Patel02256232008-10-07 17:48:33 +00005372 &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 Lattnere23c1392005-05-06 05:47:36 +00005373</pre>
5374
Chris Lattner2f7c9632001-06-06 20:29:01 +00005375<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00005376<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00005377
Chris Lattner2f7c9632001-06-06 20:29:01 +00005378<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00005379<p>This instruction requires several arguments:</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00005380
Chris Lattnera8292f32002-05-06 22:08:29 +00005381<ol>
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00005382 <li>The optional "tail" marker indicates that the callee function does not
5383 access any allocas or varargs in the caller. Note that calls may be
5384 marked "tail" even if they do not occur before
5385 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5386 present, the function call is eligible for tail call optimization,
5387 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Cheng59676492010-03-08 21:05:02 +00005388 optimized into a jump</a>. The code generator may optimize calls marked
5389 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5390 sibling call optimization</a> when the caller and callee have
5391 matching signatures, or 2) forced tail call optimization when the
5392 following extra requirements are met:
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00005393 <ul>
5394 <li>Caller and callee both have the calling
5395 convention <tt>fastcc</tt>.</li>
5396 <li>The call is in tail position (ret immediately follows call and ret
5397 uses value of call or is void).</li>
5398 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohman6232f732010-03-02 01:08:11 +00005399 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00005400 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5401 constraints are met.</a></li>
5402 </ul>
5403 </li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00005404
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005405 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5406 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00005407 defaults to using C calling conventions. The calling convention of the
5408 call must match the calling convention of the target function, or else the
5409 behavior is undefined.</li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00005410
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005411 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5412 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5413 '<tt>inreg</tt>' attributes are valid here.</li>
5414
5415 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5416 type of the return value. Functions that return no value are marked
5417 <tt><a href="#t_void">void</a></tt>.</li>
5418
5419 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5420 being invoked. The argument types must match the types implied by this
5421 signature. This type can be omitted if the function is not varargs and if
5422 the function type does not return a pointer to a function.</li>
5423
5424 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5425 be invoked. In most cases, this is a direct function invocation, but
5426 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5427 to function value.</li>
5428
5429 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner47f2a832010-03-02 06:36:51 +00005430 signature argument types and parameter attributes. All arguments must be
5431 of <a href="#t_firstclass">first class</a> type. If the function
5432 signature indicates the function accepts a variable number of arguments,
5433 the extra arguments can be specified.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005434
5435 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5436 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5437 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattnera8292f32002-05-06 22:08:29 +00005438</ol>
Chris Lattnere23c1392005-05-06 05:47:36 +00005439
Chris Lattner2f7c9632001-06-06 20:29:01 +00005440<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005441<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5442 a specified function, with its incoming arguments bound to the specified
5443 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5444 function, control flow continues with the instruction after the function
5445 call, and the return value of the function is bound to the result
5446 argument.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00005447
Chris Lattner2f7c9632001-06-06 20:29:01 +00005448<h5>Example:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00005449<pre>
Nick Lewyckya9b13d52007-09-08 13:57:50 +00005450 %retval = call i32 @test(i32 %argc)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00005451 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattnerfb7c88d2008-03-21 17:24:17 +00005452 %X = tail call i32 @foo() <i>; yields i32</i>
5453 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5454 call void %foo(i8 97 signext)
Devang Pateld6cff512008-03-10 20:49:15 +00005455
5456 %struct.A = type { i32, i8 }
Devang Patel7e9b05e2008-10-06 18:50:38 +00005457 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmancc3132e2008-10-04 19:00:07 +00005458 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5459 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner6cbe8e92008-10-08 06:26:11 +00005460 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmaneefa7df2008-10-07 10:03:45 +00005461 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattnere23c1392005-05-06 05:47:36 +00005462</pre>
5463
Dale Johannesen68f971b2009-09-24 18:38:21 +00005464<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen722212d2009-09-25 17:04:42 +00005465standard C99 library as being the C99 library functions, and may perform
5466optimizations or generate code for them under that assumption. This is
5467something we'd like to change in the future to provide better support for
Dan Gohmana269a0a2010-03-01 17:41:39 +00005468freestanding environments and non-C-based languages.</p>
Dale Johannesen68f971b2009-09-24 18:38:21 +00005469
Misha Brukman76307852003-11-08 01:05:38 +00005470</div>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005471
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005472<!-- _______________________________________________________________________ -->
Chris Lattner6a4a0492004-09-27 21:51:25 +00005473<div class="doc_subsubsection">
Chris Lattner33337472006-01-13 23:26:01 +00005474 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005475</div>
5476
Misha Brukman76307852003-11-08 01:05:38 +00005477<div class="doc_text">
Chris Lattner6a4a0492004-09-27 21:51:25 +00005478
Chris Lattner26ca62e2003-10-18 05:51:36 +00005479<h5>Syntax:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005480<pre>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00005481 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattner6a4a0492004-09-27 21:51:25 +00005482</pre>
5483
Chris Lattner26ca62e2003-10-18 05:51:36 +00005484<h5>Overview:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00005485<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005486 the "variable argument" area of a function call. It is used to implement the
5487 <tt>va_arg</tt> macro in C.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005488
Chris Lattner26ca62e2003-10-18 05:51:36 +00005489<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005490<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5491 argument. It returns a value of the specified argument type and increments
5492 the <tt>va_list</tt> to point to the next argument. The actual type
5493 of <tt>va_list</tt> is target specific.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005494
Chris Lattner26ca62e2003-10-18 05:51:36 +00005495<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005496<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5497 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5498 to the next argument. For more information, see the variable argument
5499 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005500
5501<p>It is legal for this instruction to be called in a function which does not
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005502 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5503 function.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005504
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005505<p><tt>va_arg</tt> is an LLVM instruction instead of
5506 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5507 argument.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005508
Chris Lattner26ca62e2003-10-18 05:51:36 +00005509<h5>Example:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00005510<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5511
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005512<p>Note that the code generator does not yet fully support va_arg on many
5513 targets. Also, it does not currently support va_arg with aggregate types on
5514 any target.</p>
Dan Gohman3065b612009-01-12 23:12:39 +00005515
Misha Brukman76307852003-11-08 01:05:38 +00005516</div>
Chris Lattner941515c2004-01-06 05:31:32 +00005517
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005518<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +00005519<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5520<!-- *********************************************************************** -->
Chris Lattner941515c2004-01-06 05:31:32 +00005521
Misha Brukman76307852003-11-08 01:05:38 +00005522<div class="doc_text">
Chris Lattnerfee11462004-02-12 17:01:32 +00005523
5524<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005525 well known names and semantics and are required to follow certain
5526 restrictions. Overall, these intrinsics represent an extension mechanism for
5527 the LLVM language that does not require changing all of the transformations
5528 in LLVM when adding to the language (or the bitcode reader/writer, the
5529 parser, etc...).</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00005530
John Criswell88190562005-05-16 16:17:45 +00005531<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005532 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5533 begin with this prefix. Intrinsic functions must always be external
5534 functions: you cannot define the body of intrinsic functions. Intrinsic
5535 functions may only be used in call or invoke instructions: it is illegal to
5536 take the address of an intrinsic function. Additionally, because intrinsic
5537 functions are part of the LLVM language, it is required if any are added that
5538 they be documented here.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00005539
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005540<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5541 family of functions that perform the same operation but on different data
5542 types. Because LLVM can represent over 8 million different integer types,
5543 overloading is used commonly to allow an intrinsic function to operate on any
5544 integer type. One or more of the argument types or the result type can be
5545 overloaded to accept any integer type. Argument types may also be defined as
5546 exactly matching a previous argument's type or the result type. This allows
5547 an intrinsic function which accepts multiple arguments, but needs all of them
5548 to be of the same type, to only be overloaded with respect to a single
5549 argument or the result.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00005550
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005551<p>Overloaded intrinsics will have the names of its overloaded argument types
5552 encoded into its function name, each preceded by a period. Only those types
5553 which are overloaded result in a name suffix. Arguments whose type is matched
5554 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5555 can take an integer of any width and returns an integer of exactly the same
5556 integer width. This leads to a family of functions such as
5557 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5558 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5559 suffix is required. Because the argument's type is matched against the return
5560 type, it does not require its own name suffix.</p>
Reid Spencer4eefaab2007-04-01 08:04:23 +00005561
Eric Christopher455c5772009-12-05 02:46:03 +00005562<p>To learn how to add an intrinsic function, please see the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005563 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00005564
Misha Brukman76307852003-11-08 01:05:38 +00005565</div>
Chris Lattner941515c2004-01-06 05:31:32 +00005566
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005567<!-- ======================================================================= -->
Chris Lattner941515c2004-01-06 05:31:32 +00005568<div class="doc_subsection">
5569 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5570</div>
5571
Misha Brukman76307852003-11-08 01:05:38 +00005572<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +00005573
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005574<p>Variable argument support is defined in LLVM with
5575 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5576 intrinsic functions. These functions are related to the similarly named
5577 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005578
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005579<p>All of these functions operate on arguments that use a target-specific value
5580 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5581 not define what this type is, so all transformations should be prepared to
5582 handle these functions regardless of the type used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005583
Chris Lattner30b868d2006-05-15 17:26:46 +00005584<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005585 instruction and the variable argument handling intrinsic functions are
5586 used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005587
Bill Wendling3716c5d2007-05-29 09:04:49 +00005588<div class="doc_code">
Chris Lattnerfee11462004-02-12 17:01:32 +00005589<pre>
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00005590define i32 @test(i32 %X, ...) {
Chris Lattnerfee11462004-02-12 17:01:32 +00005591 ; Initialize variable argument processing
Jeff Cohen222a8a42007-04-29 01:07:00 +00005592 %ap = alloca i8*
Chris Lattnerdb0790c2007-01-08 07:55:15 +00005593 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00005594 call void @llvm.va_start(i8* %ap2)
Chris Lattnerfee11462004-02-12 17:01:32 +00005595
5596 ; Read a single integer argument
Jeff Cohen222a8a42007-04-29 01:07:00 +00005597 %tmp = va_arg i8** %ap, i32
Chris Lattnerfee11462004-02-12 17:01:32 +00005598
5599 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohen222a8a42007-04-29 01:07:00 +00005600 %aq = alloca i8*
Chris Lattnerdb0790c2007-01-08 07:55:15 +00005601 %aq2 = bitcast i8** %aq to i8*
Jeff Cohen222a8a42007-04-29 01:07:00 +00005602 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00005603 call void @llvm.va_end(i8* %aq2)
Chris Lattnerfee11462004-02-12 17:01:32 +00005604
5605 ; Stop processing of arguments.
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00005606 call void @llvm.va_end(i8* %ap2)
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005607 ret i32 %tmp
Chris Lattnerfee11462004-02-12 17:01:32 +00005608}
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00005609
5610declare void @llvm.va_start(i8*)
5611declare void @llvm.va_copy(i8*, i8*)
5612declare void @llvm.va_end(i8*)
Chris Lattnerfee11462004-02-12 17:01:32 +00005613</pre>
Misha Brukman76307852003-11-08 01:05:38 +00005614</div>
Chris Lattner941515c2004-01-06 05:31:32 +00005615
Bill Wendling3716c5d2007-05-29 09:04:49 +00005616</div>
5617
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005618<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +00005619<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005620 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
Chris Lattner941515c2004-01-06 05:31:32 +00005621</div>
5622
5623
Misha Brukman76307852003-11-08 01:05:38 +00005624<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005625
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005626<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005627<pre>
5628 declare void %llvm.va_start(i8* &lt;arglist&gt;)
5629</pre>
5630
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005631<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005632<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
5633 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00005634
5635<h5>Arguments:</h5>
Dan Gohmanef9462f2008-10-14 16:51:45 +00005636<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00005637
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005638<h5>Semantics:</h5>
Dan Gohmanef9462f2008-10-14 16:51:45 +00005639<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005640 macro available in C. In a target-dependent way, it initializes
5641 the <tt>va_list</tt> element to which the argument points, so that the next
5642 call to <tt>va_arg</tt> will produce the first variable argument passed to
5643 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5644 need to know the last argument of the function as the compiler can figure
5645 that out.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00005646
Misha Brukman76307852003-11-08 01:05:38 +00005647</div>
Chris Lattner941515c2004-01-06 05:31:32 +00005648
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005649<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +00005650<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005651 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
Chris Lattner941515c2004-01-06 05:31:32 +00005652</div>
5653
Misha Brukman76307852003-11-08 01:05:38 +00005654<div class="doc_text">
Chris Lattnerdb0790c2007-01-08 07:55:15 +00005655
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005656<h5>Syntax:</h5>
5657<pre>
5658 declare void @llvm.va_end(i8* &lt;arglist&gt;)
5659</pre>
5660
5661<h5>Overview:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005662<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005663 which has been initialized previously
5664 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5665 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00005666
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005667<h5>Arguments:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005668<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00005669
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005670<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00005671<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005672 macro available in C. In a target-dependent way, it destroys
5673 the <tt>va_list</tt> element to which the argument points. Calls
5674 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5675 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5676 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00005677
Misha Brukman76307852003-11-08 01:05:38 +00005678</div>
Chris Lattner941515c2004-01-06 05:31:32 +00005679
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005680<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +00005681<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005682 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
Chris Lattner941515c2004-01-06 05:31:32 +00005683</div>
5684
Misha Brukman76307852003-11-08 01:05:38 +00005685<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +00005686
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005687<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005688<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00005689 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattner757528b0b2004-05-23 21:06:01 +00005690</pre>
5691
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005692<h5>Overview:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005693<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005694 from the source argument list to the destination argument list.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005695
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005696<h5>Arguments:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00005697<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005698 The second argument is a pointer to a <tt>va_list</tt> element to copy
5699 from.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005700
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00005701<h5>Semantics:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005702<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005703 macro available in C. In a target-dependent way, it copies the
5704 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5705 element. This intrinsic is necessary because
5706 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5707 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005708
Misha Brukman76307852003-11-08 01:05:38 +00005709</div>
Chris Lattner941515c2004-01-06 05:31:32 +00005710
Chris Lattnerfee11462004-02-12 17:01:32 +00005711<!-- ======================================================================= -->
5712<div class="doc_subsection">
Chris Lattner757528b0b2004-05-23 21:06:01 +00005713 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5714</div>
5715
5716<div class="doc_text">
5717
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005718<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner67c37d12008-08-05 18:29:16 +00005719Collection</a> (GC) requires the implementation and generation of these
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005720intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5721roots on the stack</a>, as well as garbage collector implementations that
5722require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5723barriers. Front-ends for type-safe garbage collected languages should generate
5724these intrinsics to make use of the LLVM garbage collectors. For more details,
5725see <a href="GarbageCollection.html">Accurate Garbage Collection with
5726LLVM</a>.</p>
Christopher Lamb55c6d4f2007-12-17 01:00:21 +00005727
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005728<p>The garbage collection intrinsics only operate on objects in the generic
5729 address space (address space zero).</p>
Christopher Lamb55c6d4f2007-12-17 01:00:21 +00005730
Chris Lattner757528b0b2004-05-23 21:06:01 +00005731</div>
5732
5733<!-- _______________________________________________________________________ -->
5734<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005735 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005736</div>
5737
5738<div class="doc_text">
5739
5740<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005741<pre>
Chris Lattner12477732007-09-21 17:30:40 +00005742 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattner757528b0b2004-05-23 21:06:01 +00005743</pre>
5744
5745<h5>Overview:</h5>
John Criswelldfe6a862004-12-10 15:51:16 +00005746<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005747 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005748
5749<h5>Arguments:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005750<p>The first argument specifies the address of a stack object that contains the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005751 root pointer. The second pointer (which must be either a constant or a
5752 global value address) contains the meta-data to be associated with the
5753 root.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005754
5755<h5>Semantics:</h5>
Chris Lattner851b7712008-04-24 05:59:56 +00005756<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005757 location. At compile-time, the code generator generates information to allow
5758 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5759 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5760 algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005761
5762</div>
5763
Chris Lattner757528b0b2004-05-23 21:06:01 +00005764<!-- _______________________________________________________________________ -->
5765<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005766 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005767</div>
5768
5769<div class="doc_text">
5770
5771<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005772<pre>
Chris Lattner12477732007-09-21 17:30:40 +00005773 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattner757528b0b2004-05-23 21:06:01 +00005774</pre>
5775
5776<h5>Overview:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005777<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005778 locations, allowing garbage collector implementations that require read
5779 barriers.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005780
5781<h5>Arguments:</h5>
Chris Lattnerf9228072006-03-14 20:02:51 +00005782<p>The second argument is the address to read from, which should be an address
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005783 allocated from the garbage collector. The first object is a pointer to the
5784 start of the referenced object, if needed by the language runtime (otherwise
5785 null).</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005786
5787<h5>Semantics:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005788<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005789 instruction, but may be replaced with substantially more complex code by the
5790 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5791 may only be used in a function which <a href="#gc">specifies a GC
5792 algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005793
5794</div>
5795
Chris Lattner757528b0b2004-05-23 21:06:01 +00005796<!-- _______________________________________________________________________ -->
5797<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005798 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005799</div>
5800
5801<div class="doc_text">
5802
5803<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005804<pre>
Chris Lattner12477732007-09-21 17:30:40 +00005805 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattner757528b0b2004-05-23 21:06:01 +00005806</pre>
5807
5808<h5>Overview:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005809<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005810 locations, allowing garbage collector implementations that require write
5811 barriers (such as generational or reference counting collectors).</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005812
5813<h5>Arguments:</h5>
Chris Lattnerf9228072006-03-14 20:02:51 +00005814<p>The first argument is the reference to store, the second is the start of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005815 object to store it to, and the third is the address of the field of Obj to
5816 store to. If the runtime does not require a pointer to the object, Obj may
5817 be null.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005818
5819<h5>Semantics:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005820<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005821 instruction, but may be replaced with substantially more complex code by the
5822 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5823 may only be used in a function which <a href="#gc">specifies a GC
5824 algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00005825
5826</div>
5827
Chris Lattner757528b0b2004-05-23 21:06:01 +00005828<!-- ======================================================================= -->
5829<div class="doc_subsection">
Chris Lattner3649c3a2004-02-14 04:08:35 +00005830 <a name="int_codegen">Code Generator Intrinsics</a>
5831</div>
5832
5833<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005834
5835<p>These intrinsics are provided by LLVM to expose special features that may
5836 only be implemented with code generator support.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005837
5838</div>
5839
5840<!-- _______________________________________________________________________ -->
5841<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005842 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005843</div>
5844
5845<div class="doc_text">
5846
5847<h5>Syntax:</h5>
5848<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00005849 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00005850</pre>
5851
5852<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005853<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5854 target-specific value indicating the return address of the current function
5855 or one of its callers.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005856
5857<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005858<p>The argument to this intrinsic indicates which function to return the address
5859 for. Zero indicates the calling function, one indicates its caller, etc.
5860 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005861
5862<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005863<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5864 indicating the return address of the specified call frame, or zero if it
5865 cannot be identified. The value returned by this intrinsic is likely to be
5866 incorrect or 0 for arguments other than zero, so it should only be used for
5867 debugging purposes.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005868
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005869<p>Note that calling this intrinsic does not prevent function inlining or other
5870 aggressive transformations, so the value returned may not be that of the
5871 obvious source-language caller.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005872
Chris Lattner3649c3a2004-02-14 04:08:35 +00005873</div>
5874
Chris Lattner3649c3a2004-02-14 04:08:35 +00005875<!-- _______________________________________________________________________ -->
5876<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005877 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005878</div>
5879
5880<div class="doc_text">
5881
5882<h5>Syntax:</h5>
5883<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00005884 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00005885</pre>
5886
5887<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005888<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5889 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005890
5891<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005892<p>The argument to this intrinsic indicates which function to return the frame
5893 pointer for. Zero indicates the calling function, one indicates its caller,
5894 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005895
5896<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005897<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5898 indicating the frame address of the specified call frame, or zero if it
5899 cannot be identified. The value returned by this intrinsic is likely to be
5900 incorrect or 0 for arguments other than zero, so it should only be used for
5901 debugging purposes.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005902
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005903<p>Note that calling this intrinsic does not prevent function inlining or other
5904 aggressive transformations, so the value returned may not be that of the
5905 obvious source-language caller.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00005906
Chris Lattner3649c3a2004-02-14 04:08:35 +00005907</div>
5908
Chris Lattnerc8a2c222005-02-28 19:24:19 +00005909<!-- _______________________________________________________________________ -->
5910<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005911 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
Chris Lattner2f0f0012006-01-13 02:03:13 +00005912</div>
5913
5914<div class="doc_text">
5915
5916<h5>Syntax:</h5>
5917<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00005918 declare i8* @llvm.stacksave()
Chris Lattner2f0f0012006-01-13 02:03:13 +00005919</pre>
5920
5921<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005922<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5923 of the function stack, for use
5924 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5925 useful for implementing language features like scoped automatic variable
5926 sized arrays in C99.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00005927
5928<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005929<p>This intrinsic returns a opaque pointer value that can be passed
5930 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5931 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5932 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5933 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5934 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5935 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00005936
5937</div>
5938
5939<!-- _______________________________________________________________________ -->
5940<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005941 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
Chris Lattner2f0f0012006-01-13 02:03:13 +00005942</div>
5943
5944<div class="doc_text">
5945
5946<h5>Syntax:</h5>
5947<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00005948 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner2f0f0012006-01-13 02:03:13 +00005949</pre>
5950
5951<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005952<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5953 the function stack to the state it was in when the
5954 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5955 executed. This is useful for implementing language features like scoped
5956 automatic variable sized arrays in C99.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00005957
5958<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005959<p>See the description
5960 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00005961
5962</div>
5963
Chris Lattner2f0f0012006-01-13 02:03:13 +00005964<!-- _______________________________________________________________________ -->
5965<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005966 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00005967</div>
5968
5969<div class="doc_text">
5970
5971<h5>Syntax:</h5>
5972<pre>
Chris Lattner12477732007-09-21 17:30:40 +00005973 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Chris Lattnerc8a2c222005-02-28 19:24:19 +00005974</pre>
5975
5976<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005977<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5978 insert a prefetch instruction if supported; otherwise, it is a noop.
5979 Prefetches have no effect on the behavior of the program but can change its
5980 performance characteristics.</p>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00005981
5982<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005983<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5984 specifier determining if the fetch should be for a read (0) or write (1),
5985 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5986 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5987 and <tt>locality</tt> arguments must be constant integers.</p>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00005988
5989<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005990<p>This intrinsic does not modify the behavior of the program. In particular,
5991 prefetches cannot trap and do not produce a value. On targets that support
5992 this intrinsic, the prefetch can provide hints to the processor cache for
5993 better performance.</p>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00005994
5995</div>
5996
Andrew Lenharthb4427912005-03-28 20:05:49 +00005997<!-- _______________________________________________________________________ -->
5998<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00005999 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006000</div>
6001
6002<div class="doc_text">
6003
6004<h5>Syntax:</h5>
6005<pre>
Chris Lattner12477732007-09-21 17:30:40 +00006006 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharthb4427912005-03-28 20:05:49 +00006007</pre>
6008
6009<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006010<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6011 Counter (PC) in a region of code to simulators and other tools. The method
6012 is target specific, but it is expected that the marker will use exported
6013 symbols to transmit the PC of the marker. The marker makes no guarantees
6014 that it will remain with any specific instruction after optimizations. It is
6015 possible that the presence of a marker will inhibit optimizations. The
6016 intended use is to be inserted after optimizations to allow correlations of
6017 simulation runs.</p>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006018
6019<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006020<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006021
6022<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006023<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmana269a0a2010-03-01 17:41:39 +00006024 not support this intrinsic may ignore it.</p>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006025
6026</div>
6027
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006028<!-- _______________________________________________________________________ -->
6029<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006030 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006031</div>
6032
6033<div class="doc_text">
6034
6035<h5>Syntax:</h5>
6036<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00006037 declare i64 @llvm.readcyclecounter()
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006038</pre>
6039
6040<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006041<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6042 counter register (or similar low latency, high accuracy clocks) on those
6043 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6044 should map to RPCC. As the backing counters overflow quickly (on the order
6045 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006046
6047<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006048<p>When directly supported, reading the cycle counter should not modify any
6049 memory. Implementations are allowed to either return a application specific
6050 value or a system wide value. On backends without support, this is lowered
6051 to a constant 0.</p>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006052
6053</div>
6054
Chris Lattner3649c3a2004-02-14 04:08:35 +00006055<!-- ======================================================================= -->
6056<div class="doc_subsection">
Chris Lattnerfee11462004-02-12 17:01:32 +00006057 <a name="int_libc">Standard C Library Intrinsics</a>
6058</div>
6059
6060<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006061
6062<p>LLVM provides intrinsics for a few important standard C library functions.
6063 These intrinsics allow source-language front-ends to pass information about
6064 the alignment of the pointer arguments to the code generator, providing
6065 opportunity for more efficient code generation.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006066
6067</div>
6068
6069<!-- _______________________________________________________________________ -->
6070<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006071 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
Chris Lattnerfee11462004-02-12 17:01:32 +00006072</div>
6073
6074<div class="doc_text">
6075
6076<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006077<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
Mon P Wang508127b2010-04-07 06:35:53 +00006078 integer bit width and for different address spaces. Not all targets support
6079 all bit widths however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006080
Chris Lattnerfee11462004-02-12 17:01:32 +00006081<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006082 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006083 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006084 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006085 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattnerfee11462004-02-12 17:01:32 +00006086</pre>
6087
6088<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006089<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6090 source location to the destination location.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006091
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006092<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006093 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6094 and the pointers can be in specified address spaces.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006095
6096<h5>Arguments:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006097
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006098<p>The first argument is a pointer to the destination, the second is a pointer
6099 to the source. The third argument is an integer argument specifying the
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006100 number of bytes to copy, the fourth argument is the alignment of the
6101 source and destination locations, and the fifth is a boolean indicating a
6102 volatile access.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006103
Dan Gohmana269a0a2010-03-01 17:41:39 +00006104<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006105 then the caller guarantees that both the source and destination pointers are
6106 aligned to that boundary.</p>
Chris Lattner4c67c482004-02-12 21:18:15 +00006107
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00006108<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6109 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6110 The detailed access behavior is not very cleanly specified and it is unwise
6111 to depend on it.</p>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006112
Chris Lattnerfee11462004-02-12 17:01:32 +00006113<h5>Semantics:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006114
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006115<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6116 source location to the destination location, which are not allowed to
6117 overlap. It copies "len" bytes of memory over. If the argument is known to
6118 be aligned to some boundary, this can be specified as the fourth argument,
6119 otherwise it should be set to 0 or 1.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006120
Chris Lattnerfee11462004-02-12 17:01:32 +00006121</div>
6122
Chris Lattnerf30152e2004-02-12 18:10:10 +00006123<!-- _______________________________________________________________________ -->
6124<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006125 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
Chris Lattnerf30152e2004-02-12 18:10:10 +00006126</div>
6127
6128<div class="doc_text">
6129
6130<h5>Syntax:</h5>
Chris Lattnerdd708342008-11-21 16:42:48 +00006131<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Mon P Wang508127b2010-04-07 06:35:53 +00006132 width and for different address space. Not all targets support all bit
6133 widths however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006134
Chris Lattnerf30152e2004-02-12 18:10:10 +00006135<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006136 declare void @llvm.memmove.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006137 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006138 declare void @llvm.memmove.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006139 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattnerf30152e2004-02-12 18:10:10 +00006140</pre>
6141
6142<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006143<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6144 source location to the destination location. It is similar to the
6145 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6146 overlap.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00006147
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006148<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006149 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6150 and the pointers can be in specified address spaces.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00006151
6152<h5>Arguments:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006153
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006154<p>The first argument is a pointer to the destination, the second is a pointer
6155 to the source. The third argument is an integer argument specifying the
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006156 number of bytes to copy, the fourth argument is the alignment of the
6157 source and destination locations, and the fifth is a boolean indicating a
6158 volatile access.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00006159
Dan Gohmana269a0a2010-03-01 17:41:39 +00006160<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006161 then the caller guarantees that the source and destination pointers are
6162 aligned to that boundary.</p>
Chris Lattner4c67c482004-02-12 21:18:15 +00006163
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00006164<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6165 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6166 The detailed access behavior is not very cleanly specified and it is unwise
6167 to depend on it.</p>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006168
Chris Lattnerf30152e2004-02-12 18:10:10 +00006169<h5>Semantics:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006170
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006171<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6172 source location to the destination location, which may overlap. It copies
6173 "len" bytes of memory over. If the argument is known to be aligned to some
6174 boundary, this can be specified as the fourth argument, otherwise it should
6175 be set to 0 or 1.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00006176
Chris Lattnerf30152e2004-02-12 18:10:10 +00006177</div>
6178
Chris Lattner3649c3a2004-02-14 04:08:35 +00006179<!-- _______________________________________________________________________ -->
6180<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006181 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006182</div>
6183
6184<div class="doc_text">
6185
6186<h5>Syntax:</h5>
Chris Lattnerdd708342008-11-21 16:42:48 +00006187<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
Mon P Wang508127b2010-04-07 06:35:53 +00006188 width and for different address spaces. Not all targets support all bit
6189 widths however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006190
Chris Lattner3649c3a2004-02-14 04:08:35 +00006191<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006192 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattner685db9d2010-04-08 00:54:34 +00006193 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006194 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattner685db9d2010-04-08 00:54:34 +00006195 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00006196</pre>
6197
6198<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006199<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6200 particular byte value.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006201
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006202<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006203 intrinsic does not return a value, takes extra alignment/volatile arguments,
6204 and the destination can be in an arbitrary address space.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006205
6206<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006207<p>The first argument is a pointer to the destination to fill, the second is the
6208 byte value to fill it with, the third argument is an integer argument
6209 specifying the number of bytes to fill, and the fourth argument is the known
6210 alignment of destination location.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006211
Dan Gohmana269a0a2010-03-01 17:41:39 +00006212<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006213 then the caller guarantees that the destination pointer is aligned to that
6214 boundary.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006215
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00006216<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6217 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6218 The detailed access behavior is not very cleanly specified and it is unwise
6219 to depend on it.</p>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00006220
Chris Lattner3649c3a2004-02-14 04:08:35 +00006221<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006222<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6223 at the destination location. If the argument is known to be aligned to some
6224 boundary, this can be specified as the fourth argument, otherwise it should
6225 be set to 0 or 1.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006226
Chris Lattner3649c3a2004-02-14 04:08:35 +00006227</div>
6228
Chris Lattner3b4f4372004-06-11 02:28:03 +00006229<!-- _______________________________________________________________________ -->
6230<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006231 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006232</div>
6233
6234<div class="doc_text">
6235
6236<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006237<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6238 floating point or vector of floating point type. Not all targets support all
6239 types however.</p>
6240
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006241<pre>
Dale Johannesendd89d272007-10-02 17:47:38 +00006242 declare float @llvm.sqrt.f32(float %Val)
6243 declare double @llvm.sqrt.f64(double %Val)
6244 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6245 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6246 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006247</pre>
6248
6249<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006250<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6251 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6252 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6253 behavior for negative numbers other than -0.0 (which allows for better
6254 optimization, because there is no need to worry about errno being
6255 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006256
6257<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006258<p>The argument and return value are floating point numbers of the same
6259 type.</p>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006260
6261<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006262<p>This function returns the sqrt of the specified operand if it is a
6263 nonnegative floating point number.</p>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006264
Chris Lattner8a8f2e52005-07-21 01:29:16 +00006265</div>
6266
Chris Lattner33b73f92006-09-08 06:34:02 +00006267<!-- _______________________________________________________________________ -->
6268<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006269 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
Chris Lattner33b73f92006-09-08 06:34:02 +00006270</div>
6271
6272<div class="doc_text">
6273
6274<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006275<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6276 floating point or vector of floating point type. Not all targets support all
6277 types however.</p>
6278
Chris Lattner33b73f92006-09-08 06:34:02 +00006279<pre>
Dale Johannesendd89d272007-10-02 17:47:38 +00006280 declare float @llvm.powi.f32(float %Val, i32 %power)
6281 declare double @llvm.powi.f64(double %Val, i32 %power)
6282 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6283 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6284 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattner33b73f92006-09-08 06:34:02 +00006285</pre>
6286
6287<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006288<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6289 specified (positive or negative) power. The order of evaluation of
6290 multiplications is not defined. When a vector of floating point type is
6291 used, the second argument remains a scalar integer value.</p>
Chris Lattner33b73f92006-09-08 06:34:02 +00006292
6293<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006294<p>The second argument is an integer power, and the first is a value to raise to
6295 that power.</p>
Chris Lattner33b73f92006-09-08 06:34:02 +00006296
6297<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006298<p>This function returns the first value raised to the second power with an
6299 unspecified sequence of rounding operations.</p>
Chris Lattner33b73f92006-09-08 06:34:02 +00006300
Chris Lattner33b73f92006-09-08 06:34:02 +00006301</div>
6302
Dan Gohmanb6324c12007-10-15 20:30:11 +00006303<!-- _______________________________________________________________________ -->
6304<div class="doc_subsubsection">
6305 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6306</div>
6307
6308<div class="doc_text">
6309
6310<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006311<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6312 floating point or vector of floating point type. Not all targets support all
6313 types however.</p>
6314
Dan Gohmanb6324c12007-10-15 20:30:11 +00006315<pre>
6316 declare float @llvm.sin.f32(float %Val)
6317 declare double @llvm.sin.f64(double %Val)
6318 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6319 declare fp128 @llvm.sin.f128(fp128 %Val)
6320 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6321</pre>
6322
6323<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006324<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006325
6326<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006327<p>The argument and return value are floating point numbers of the same
6328 type.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006329
6330<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006331<p>This function returns the sine of the specified operand, returning the same
6332 values as the libm <tt>sin</tt> functions would, and handles error conditions
6333 in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006334
Dan Gohmanb6324c12007-10-15 20:30:11 +00006335</div>
6336
6337<!-- _______________________________________________________________________ -->
6338<div class="doc_subsubsection">
6339 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6340</div>
6341
6342<div class="doc_text">
6343
6344<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006345<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6346 floating point or vector of floating point type. Not all targets support all
6347 types however.</p>
6348
Dan Gohmanb6324c12007-10-15 20:30:11 +00006349<pre>
6350 declare float @llvm.cos.f32(float %Val)
6351 declare double @llvm.cos.f64(double %Val)
6352 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6353 declare fp128 @llvm.cos.f128(fp128 %Val)
6354 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6355</pre>
6356
6357<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006358<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006359
6360<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006361<p>The argument and return value are floating point numbers of the same
6362 type.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006363
6364<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006365<p>This function returns the cosine of the specified operand, returning the same
6366 values as the libm <tt>cos</tt> functions would, and handles error conditions
6367 in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006368
Dan Gohmanb6324c12007-10-15 20:30:11 +00006369</div>
6370
6371<!-- _______________________________________________________________________ -->
6372<div class="doc_subsubsection">
6373 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6374</div>
6375
6376<div class="doc_text">
6377
6378<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006379<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6380 floating point or vector of floating point type. Not all targets support all
6381 types however.</p>
6382
Dan Gohmanb6324c12007-10-15 20:30:11 +00006383<pre>
6384 declare float @llvm.pow.f32(float %Val, float %Power)
6385 declare double @llvm.pow.f64(double %Val, double %Power)
6386 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6387 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6388 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6389</pre>
6390
6391<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006392<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6393 specified (positive or negative) power.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006394
6395<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006396<p>The second argument is a floating point power, and the first is a value to
6397 raise to that power.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006398
6399<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006400<p>This function returns the first value raised to the second power, returning
6401 the same values as the libm <tt>pow</tt> functions would, and handles error
6402 conditions in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00006403
Dan Gohmanb6324c12007-10-15 20:30:11 +00006404</div>
6405
Andrew Lenharth1d463522005-05-03 18:01:48 +00006406<!-- ======================================================================= -->
6407<div class="doc_subsection">
Nate Begeman0f223bb2006-01-13 23:26:38 +00006408 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006409</div>
6410
6411<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006412
6413<p>LLVM provides intrinsics for a few important bit manipulation operations.
6414 These allow efficient code generation for some algorithms.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006415
6416</div>
6417
6418<!-- _______________________________________________________________________ -->
6419<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +00006420 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
Nate Begeman0f223bb2006-01-13 23:26:38 +00006421</div>
6422
6423<div class="doc_text">
6424
6425<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +00006426<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006427 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6428
Nate Begeman0f223bb2006-01-13 23:26:38 +00006429<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +00006430 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
6431 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
6432 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman0f223bb2006-01-13 23:26:38 +00006433</pre>
6434
6435<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006436<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6437 values with an even number of bytes (positive multiple of 16 bits). These
6438 are useful for performing operations on data that is not in the target's
6439 native byte order.</p>
Nate Begeman0f223bb2006-01-13 23:26:38 +00006440
6441<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006442<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6443 and low byte of the input i16 swapped. Similarly,
6444 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6445 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6446 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6447 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6448 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6449 more, respectively).</p>
Nate Begeman0f223bb2006-01-13 23:26:38 +00006450
6451</div>
6452
6453<!-- _______________________________________________________________________ -->
6454<div class="doc_subsubsection">
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00006455 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006456</div>
6457
6458<div class="doc_text">
6459
6460<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +00006461<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006462 width. Not all targets support all bit widths however.</p>
6463
Andrew Lenharth1d463522005-05-03 18:01:48 +00006464<pre>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006465 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +00006466 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00006467 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +00006468 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
6469 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +00006470</pre>
6471
6472<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006473<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6474 in a value.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006475
6476<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006477<p>The only argument is the value to be counted. The argument may be of any
6478 integer type. The return type must match the argument type.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006479
6480<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006481<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006482
Andrew Lenharth1d463522005-05-03 18:01:48 +00006483</div>
6484
6485<!-- _______________________________________________________________________ -->
6486<div class="doc_subsubsection">
Chris Lattnerb748c672006-01-16 22:34:14 +00006487 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006488</div>
6489
6490<div class="doc_text">
6491
6492<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006493<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6494 integer bit width. Not all targets support all bit widths however.</p>
6495
Andrew Lenharth1d463522005-05-03 18:01:48 +00006496<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +00006497 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
6498 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00006499 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +00006500 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
6501 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +00006502</pre>
6503
6504<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006505<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6506 leading zeros in a variable.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006507
6508<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006509<p>The only argument is the value to be counted. The argument may be of any
6510 integer type. The return type must match the argument type.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006511
6512<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006513<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6514 zeros in a variable. If the src == 0 then the result is the size in bits of
6515 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00006516
Andrew Lenharth1d463522005-05-03 18:01:48 +00006517</div>
Chris Lattner3b4f4372004-06-11 02:28:03 +00006518
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006519<!-- _______________________________________________________________________ -->
6520<div class="doc_subsubsection">
Chris Lattnerb748c672006-01-16 22:34:14 +00006521 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006522</div>
6523
6524<div class="doc_text">
6525
6526<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006527<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6528 integer bit width. Not all targets support all bit widths however.</p>
6529
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006530<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +00006531 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
6532 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00006533 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +00006534 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
6535 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006536</pre>
6537
6538<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006539<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6540 trailing zeros.</p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006541
6542<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006543<p>The only argument is the value to be counted. The argument may be of any
6544 integer type. The return type must match the argument type.</p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006545
6546<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006547<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6548 zeros in a variable. If the src == 0 then the result is the size in bits of
6549 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006550
Chris Lattnerefa20fa2005-05-15 19:39:26 +00006551</div>
6552
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006553<!-- ======================================================================= -->
6554<div class="doc_subsection">
6555 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6556</div>
6557
6558<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006559
6560<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006561
6562</div>
6563
Bill Wendlingf4d70622009-02-08 01:40:31 +00006564<!-- _______________________________________________________________________ -->
6565<div class="doc_subsubsection">
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006566 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006567</div>
6568
6569<div class="doc_text">
6570
6571<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006572<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006573 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006574
6575<pre>
6576 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6577 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6578 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6579</pre>
6580
6581<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006582<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006583 a signed addition of the two arguments, and indicate whether an overflow
6584 occurred during the signed summation.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006585
6586<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006587<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006588 be of integer types of any bit width, but they must have the same bit
6589 width. The second element of the result structure must be of
6590 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6591 undergo signed addition.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006592
6593<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006594<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006595 a signed addition of the two variables. They return a structure &mdash; the
6596 first element of which is the signed summation, and the second element of
6597 which is a bit specifying if the signed summation resulted in an
6598 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006599
6600<h5>Examples:</h5>
6601<pre>
6602 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6603 %sum = extractvalue {i32, i1} %res, 0
6604 %obit = extractvalue {i32, i1} %res, 1
6605 br i1 %obit, label %overflow, label %normal
6606</pre>
6607
6608</div>
6609
6610<!-- _______________________________________________________________________ -->
6611<div class="doc_subsubsection">
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006612 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006613</div>
6614
6615<div class="doc_text">
6616
6617<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006618<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006619 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006620
6621<pre>
6622 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6623 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6624 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6625</pre>
6626
6627<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006628<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006629 an unsigned addition of the two arguments, and indicate whether a carry
6630 occurred during the unsigned summation.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006631
6632<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006633<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006634 be of integer types of any bit width, but they must have the same bit
6635 width. The second element of the result structure must be of
6636 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6637 undergo unsigned addition.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006638
6639<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006640<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006641 an unsigned addition of the two arguments. They return a structure &mdash;
6642 the first element of which is the sum, and the second element of which is a
6643 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006644
6645<h5>Examples:</h5>
6646<pre>
6647 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6648 %sum = extractvalue {i32, i1} %res, 0
6649 %obit = extractvalue {i32, i1} %res, 1
6650 br i1 %obit, label %carry, label %normal
6651</pre>
6652
6653</div>
6654
6655<!-- _______________________________________________________________________ -->
6656<div class="doc_subsubsection">
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006657 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006658</div>
6659
6660<div class="doc_text">
6661
6662<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006663<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006664 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006665
6666<pre>
6667 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6668 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6669 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6670</pre>
6671
6672<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006673<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006674 a signed subtraction of the two arguments, and indicate whether an overflow
6675 occurred during the signed subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006676
6677<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006678<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006679 be of integer types of any bit width, but they must have the same bit
6680 width. The second element of the result structure must be of
6681 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6682 undergo signed subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006683
6684<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006685<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006686 a signed subtraction of the two arguments. They return a structure &mdash;
6687 the first element of which is the subtraction, and the second element of
6688 which is a bit specifying if the signed subtraction resulted in an
6689 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006690
6691<h5>Examples:</h5>
6692<pre>
6693 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6694 %sum = extractvalue {i32, i1} %res, 0
6695 %obit = extractvalue {i32, i1} %res, 1
6696 br i1 %obit, label %overflow, label %normal
6697</pre>
6698
6699</div>
6700
6701<!-- _______________________________________________________________________ -->
6702<div class="doc_subsubsection">
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006703 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006704</div>
6705
6706<div class="doc_text">
6707
6708<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006709<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006710 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006711
6712<pre>
6713 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6714 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6715 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6716</pre>
6717
6718<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006719<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006720 an unsigned subtraction of the two arguments, and indicate whether an
6721 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006722
6723<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006724<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006725 be of integer types of any bit width, but they must have the same bit
6726 width. The second element of the result structure must be of
6727 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6728 undergo unsigned subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006729
6730<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006731<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006732 an unsigned subtraction of the two arguments. They return a structure &mdash;
6733 the first element of which is the subtraction, and the second element of
6734 which is a bit specifying if the unsigned subtraction resulted in an
6735 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006736
6737<h5>Examples:</h5>
6738<pre>
6739 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6740 %sum = extractvalue {i32, i1} %res, 0
6741 %obit = extractvalue {i32, i1} %res, 1
6742 br i1 %obit, label %overflow, label %normal
6743</pre>
6744
6745</div>
6746
6747<!-- _______________________________________________________________________ -->
6748<div class="doc_subsubsection">
Bill Wendlingfd2bd722009-02-08 04:04:40 +00006749 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006750</div>
6751
6752<div class="doc_text">
6753
6754<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006755<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006756 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006757
6758<pre>
6759 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6760 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6761 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6762</pre>
6763
6764<h5>Overview:</h5>
6765
6766<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006767 a signed multiplication of the two arguments, and indicate whether an
6768 overflow occurred during the signed multiplication.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006769
6770<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006771<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006772 be of integer types of any bit width, but they must have the same bit
6773 width. The second element of the result structure must be of
6774 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6775 undergo signed multiplication.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006776
6777<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006778<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006779 a signed multiplication of the two arguments. They return a structure &mdash;
6780 the first element of which is the multiplication, and the second element of
6781 which is a bit specifying if the signed multiplication resulted in an
6782 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00006783
6784<h5>Examples:</h5>
6785<pre>
6786 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6787 %sum = extractvalue {i32, i1} %res, 0
6788 %obit = extractvalue {i32, i1} %res, 1
6789 br i1 %obit, label %overflow, label %normal
6790</pre>
6791
Reid Spencer5bf54c82007-04-11 23:23:49 +00006792</div>
6793
Bill Wendlingb9a73272009-02-08 23:00:09 +00006794<!-- _______________________________________________________________________ -->
6795<div class="doc_subsubsection">
6796 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6797</div>
6798
6799<div class="doc_text">
6800
6801<h5>Syntax:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006802<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006803 on any integer bit width.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006804
6805<pre>
6806 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6807 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6808 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6809</pre>
6810
6811<h5>Overview:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006812<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006813 a unsigned multiplication of the two arguments, and indicate whether an
6814 overflow occurred during the unsigned multiplication.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006815
6816<h5>Arguments:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006817<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006818 be of integer types of any bit width, but they must have the same bit
6819 width. The second element of the result structure must be of
6820 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6821 undergo unsigned multiplication.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006822
6823<h5>Semantics:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006824<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006825 an unsigned multiplication of the two arguments. They return a structure
6826 &mdash; the first element of which is the multiplication, and the second
6827 element of which is a bit specifying if the unsigned multiplication resulted
6828 in an overflow.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00006829
6830<h5>Examples:</h5>
6831<pre>
6832 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6833 %sum = extractvalue {i32, i1} %res, 0
6834 %obit = extractvalue {i32, i1} %res, 1
6835 br i1 %obit, label %overflow, label %normal
6836</pre>
6837
6838</div>
6839
Chris Lattner941515c2004-01-06 05:31:32 +00006840<!-- ======================================================================= -->
6841<div class="doc_subsection">
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00006842 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
6843</div>
6844
6845<div class="doc_text">
6846
Chris Lattner022a9fb2010-03-15 04:12:21 +00006847<p>Half precision floating point is a storage-only format. This means that it is
6848 a dense encoding (in memory) but does not support computation in the
6849 format.</p>
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006850
Chris Lattner022a9fb2010-03-15 04:12:21 +00006851<p>This means that code must first load the half-precision floating point
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006852 value as an i16, then convert it to float with <a
6853 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
6854 Computation can then be performed on the float value (including extending to
Chris Lattner022a9fb2010-03-15 04:12:21 +00006855 double etc). To store the value back to memory, it is first converted to
6856 float if needed, then converted to i16 with
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006857 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
6858 storing as an i16 value.</p>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00006859</div>
6860
6861<!-- _______________________________________________________________________ -->
6862<div class="doc_subsubsection">
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006863 <a name="int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00006864</div>
6865
6866<div class="doc_text">
6867
6868<h5>Syntax:</h5>
6869<pre>
6870 declare i16 @llvm.convert.to.fp16(f32 %a)
6871</pre>
6872
6873<h5>Overview:</h5>
6874<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6875 a conversion from single precision floating point format to half precision
6876 floating point format.</p>
6877
6878<h5>Arguments:</h5>
6879<p>The intrinsic function contains single argument - the value to be
6880 converted.</p>
6881
6882<h5>Semantics:</h5>
6883<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6884 a conversion from single precision floating point format to half precision
Chris Lattner022a9fb2010-03-15 04:12:21 +00006885 floating point format. The return value is an <tt>i16</tt> which
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006886 contains the converted number.</p>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00006887
6888<h5>Examples:</h5>
6889<pre>
6890 %res = call i16 @llvm.convert.to.fp16(f32 %a)
6891 store i16 %res, i16* @x, align 2
6892</pre>
6893
6894</div>
6895
6896<!-- _______________________________________________________________________ -->
6897<div class="doc_subsubsection">
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006898 <a name="int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00006899</div>
6900
6901<div class="doc_text">
6902
6903<h5>Syntax:</h5>
6904<pre>
6905 declare f32 @llvm.convert.from.fp16(i16 %a)
6906</pre>
6907
6908<h5>Overview:</h5>
6909<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
6910 a conversion from half precision floating point format to single precision
6911 floating point format.</p>
6912
6913<h5>Arguments:</h5>
6914<p>The intrinsic function contains single argument - the value to be
6915 converted.</p>
6916
6917<h5>Semantics:</h5>
6918<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner022a9fb2010-03-15 04:12:21 +00006919 conversion from half single precision floating point format to single
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00006920 precision floating point format. The input half-float value is represented by
6921 an <tt>i16</tt> value.</p>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00006922
6923<h5>Examples:</h5>
6924<pre>
6925 %a = load i16* @x, align 2
6926 %res = call f32 @llvm.convert.from.fp16(i16 %a)
6927</pre>
6928
6929</div>
6930
6931<!-- ======================================================================= -->
6932<div class="doc_subsection">
Chris Lattner941515c2004-01-06 05:31:32 +00006933 <a name="int_debugger">Debugger Intrinsics</a>
6934</div>
6935
6936<div class="doc_text">
Chris Lattner941515c2004-01-06 05:31:32 +00006937
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006938<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6939 prefix), are described in
6940 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6941 Level Debugging</a> document.</p>
6942
6943</div>
Chris Lattner941515c2004-01-06 05:31:32 +00006944
Jim Laskey2211f492007-03-14 19:31:19 +00006945<!-- ======================================================================= -->
6946<div class="doc_subsection">
6947 <a name="int_eh">Exception Handling Intrinsics</a>
6948</div>
6949
6950<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006951
6952<p>The LLVM exception handling intrinsics (which all start with
6953 <tt>llvm.eh.</tt> prefix), are described in
6954 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6955 Handling</a> document.</p>
6956
Jim Laskey2211f492007-03-14 19:31:19 +00006957</div>
6958
Tanya Lattnercb1b9602007-06-15 20:50:54 +00006959<!-- ======================================================================= -->
6960<div class="doc_subsection">
Duncan Sands86e01192007-09-11 14:10:23 +00006961 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands644f9172007-07-27 12:58:54 +00006962</div>
6963
6964<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006965
6966<p>This intrinsic makes it possible to excise one parameter, marked with
Dan Gohman3770af52010-07-02 23:18:08 +00006967 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
6968 The result is a callable
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006969 function pointer lacking the nest parameter - the caller does not need to
6970 provide a value for it. Instead, the value to use is stored in advance in a
6971 "trampoline", a block of memory usually allocated on the stack, which also
6972 contains code to splice the nest value into the argument list. This is used
6973 to implement the GCC nested function address extension.</p>
6974
6975<p>For example, if the function is
6976 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6977 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6978 follows:</p>
6979
6980<div class="doc_code">
Duncan Sands644f9172007-07-27 12:58:54 +00006981<pre>
Duncan Sands86e01192007-09-11 14:10:23 +00006982 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6983 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Dan Gohmand6a6f612010-05-28 17:07:41 +00006984 %p = call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval)
Duncan Sands86e01192007-09-11 14:10:23 +00006985 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands644f9172007-07-27 12:58:54 +00006986</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006987</div>
6988
Dan Gohmand6a6f612010-05-28 17:07:41 +00006989<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
6990 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006991
Duncan Sands644f9172007-07-27 12:58:54 +00006992</div>
6993
6994<!-- _______________________________________________________________________ -->
6995<div class="doc_subsubsection">
6996 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6997</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006998
Duncan Sands644f9172007-07-27 12:58:54 +00006999<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007000
Duncan Sands644f9172007-07-27 12:58:54 +00007001<h5>Syntax:</h5>
7002<pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007003 declare i8* @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands644f9172007-07-27 12:58:54 +00007004</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007005
Duncan Sands644f9172007-07-27 12:58:54 +00007006<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007007<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
7008 function pointer suitable for executing it.</p>
7009
Duncan Sands644f9172007-07-27 12:58:54 +00007010<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007011<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7012 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7013 sufficiently aligned block of memory; this memory is written to by the
7014 intrinsic. Note that the size and the alignment are target-specific - LLVM
7015 currently provides no portable way of determining them, so a front-end that
7016 generates this intrinsic needs to have some target-specific knowledge.
7017 The <tt>func</tt> argument must hold a function bitcast to
7018 an <tt>i8*</tt>.</p>
7019
Duncan Sands644f9172007-07-27 12:58:54 +00007020<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007021<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
7022 dependent code, turning it into a function. A pointer to this function is
7023 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
7024 function pointer type</a> before being called. The new function's signature
7025 is the same as that of <tt>func</tt> with any arguments marked with
7026 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
7027 is allowed, and it must be of pointer type. Calling the new function is
7028 equivalent to calling <tt>func</tt> with the same argument list, but
7029 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
7030 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
7031 by <tt>tramp</tt> is modified, then the effect of any later call to the
7032 returned function pointer is undefined.</p>
7033
Duncan Sands644f9172007-07-27 12:58:54 +00007034</div>
7035
7036<!-- ======================================================================= -->
7037<div class="doc_subsection">
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007038 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
7039</div>
7040
7041<div class="doc_text">
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007042
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007043<p>These intrinsic functions expand the "universal IR" of LLVM to represent
7044 hardware constructs for atomic operations and memory synchronization. This
7045 provides an interface to the hardware, not an interface to the programmer. It
7046 is aimed at a low enough level to allow any programming models or APIs
7047 (Application Programming Interfaces) which need atomic behaviors to map
7048 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
7049 hardware provides a "universal IR" for source languages, it also provides a
7050 starting point for developing a "universal" atomic operation and
7051 synchronization IR.</p>
7052
7053<p>These do <em>not</em> form an API such as high-level threading libraries,
7054 software transaction memory systems, atomic primitives, and intrinsic
7055 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
7056 application libraries. The hardware interface provided by LLVM should allow
7057 a clean implementation of all of these APIs and parallel programming models.
7058 No one model or paradigm should be selected above others unless the hardware
7059 itself ubiquitously does so.</p>
7060
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007061</div>
7062
7063<!-- _______________________________________________________________________ -->
7064<div class="doc_subsubsection">
7065 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
7066</div>
7067<div class="doc_text">
7068<h5>Syntax:</h5>
7069<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007070 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 Lenharth9b254ee2008-02-16 01:24:58 +00007071</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007072
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007073<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007074<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7075 specific pairs of memory access types.</p>
7076
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007077<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007078<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7079 The first four arguments enables a specific barrier as listed below. The
Dan Gohmana269a0a2010-03-01 17:41:39 +00007080 fifth argument specifies that the barrier applies to io or device or uncached
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007081 memory.</p>
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007082
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007083<ul>
7084 <li><tt>ll</tt>: load-load barrier</li>
7085 <li><tt>ls</tt>: load-store barrier</li>
7086 <li><tt>sl</tt>: store-load barrier</li>
7087 <li><tt>ss</tt>: store-store barrier</li>
7088 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7089</ul>
7090
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007091<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007092<p>This intrinsic causes the system to enforce some ordering constraints upon
7093 the loads and stores of the program. This barrier does not
7094 indicate <em>when</em> any events will occur, it only enforces
7095 an <em>order</em> in which they occur. For any of the specified pairs of load
7096 and store operations (f.ex. load-load, or store-load), all of the first
7097 operations preceding the barrier will complete before any of the second
7098 operations succeeding the barrier begin. Specifically the semantics for each
7099 pairing is as follows:</p>
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007100
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007101<ul>
7102 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7103 after the barrier begins.</li>
Eric Christopher455c5772009-12-05 02:46:03 +00007104 <li><tt>ls</tt>: All loads before the barrier must complete before any
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007105 store after the barrier begins.</li>
Eric Christopher455c5772009-12-05 02:46:03 +00007106 <li><tt>ss</tt>: All stores before the barrier must complete before any
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007107 store after the barrier begins.</li>
Eric Christopher455c5772009-12-05 02:46:03 +00007108 <li><tt>sl</tt>: All stores before the barrier must complete before any
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007109 load after the barrier begins.</li>
7110</ul>
7111
7112<p>These semantics are applied with a logical "and" behavior when more than one
7113 is enabled in a single memory barrier intrinsic.</p>
7114
7115<p>Backends may implement stronger barriers than those requested when they do
7116 not support as fine grained a barrier as requested. Some architectures do
7117 not need all types of barriers and on such architectures, these become
7118 noops.</p>
7119
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007120<h5>Example:</h5>
7121<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007122%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7123%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007124 store i32 4, %ptr
7125
7126%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007127 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false)
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007128 <i>; guarantee the above finishes</i>
7129 store i32 8, %ptr <i>; before this begins</i>
7130</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007131
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007132</div>
7133
Andrew Lenharth95528942008-02-21 06:45:13 +00007134<!-- _______________________________________________________________________ -->
7135<div class="doc_subsubsection">
Mon P Wang6a490372008-06-25 08:15:39 +00007136 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharth95528942008-02-21 06:45:13 +00007137</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007138
Andrew Lenharth95528942008-02-21 06:45:13 +00007139<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007140
Andrew Lenharth95528942008-02-21 06:45:13 +00007141<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007142<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7143 any integer bit width and for different address spaces. Not all targets
7144 support all bit widths however.</p>
Andrew Lenharth95528942008-02-21 06:45:13 +00007145
7146<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007147 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt;)
7148 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt;)
7149 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt;)
7150 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt;)
Andrew Lenharth95528942008-02-21 06:45:13 +00007151</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007152
Andrew Lenharth95528942008-02-21 06:45:13 +00007153<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007154<p>This loads a value in memory and compares it to a given value. If they are
7155 equal, it stores a new value into the memory.</p>
7156
Andrew Lenharth95528942008-02-21 06:45:13 +00007157<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007158<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7159 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7160 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7161 this integer type. While any bit width integer may be used, targets may only
7162 lower representations they support in hardware.</p>
Andrew Lenharth95528942008-02-21 06:45:13 +00007163
Andrew Lenharth95528942008-02-21 06:45:13 +00007164<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007165<p>This entire intrinsic must be executed atomically. It first loads the value
7166 in memory pointed to by <tt>ptr</tt> and compares it with the
7167 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7168 memory. The loaded value is yielded in all cases. This provides the
7169 equivalent of an atomic compare-and-swap operation within the SSA
7170 framework.</p>
Andrew Lenharth95528942008-02-21 06:45:13 +00007171
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007172<h5>Examples:</h5>
Andrew Lenharth95528942008-02-21 06:45:13 +00007173<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007174%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7175%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth95528942008-02-21 06:45:13 +00007176 store i32 4, %ptr
7177
7178%val1 = add i32 4, 4
Dan Gohmand6a6f612010-05-28 17:07:41 +00007179%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
Andrew Lenharth95528942008-02-21 06:45:13 +00007180 <i>; yields {i32}:result1 = 4</i>
7181%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7182%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7183
7184%val2 = add i32 1, 1
Dan Gohmand6a6f612010-05-28 17:07:41 +00007185%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
Andrew Lenharth95528942008-02-21 06:45:13 +00007186 <i>; yields {i32}:result2 = 8</i>
7187%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7188
7189%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7190</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007191
Andrew Lenharth95528942008-02-21 06:45:13 +00007192</div>
7193
7194<!-- _______________________________________________________________________ -->
7195<div class="doc_subsubsection">
7196 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7197</div>
7198<div class="doc_text">
7199<h5>Syntax:</h5>
7200
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007201<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7202 integer bit width. Not all targets support all bit widths however.</p>
7203
Andrew Lenharth95528942008-02-21 06:45:13 +00007204<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007205 declare i8 @llvm.atomic.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;val&gt;)
7206 declare i16 @llvm.atomic.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;val&gt;)
7207 declare i32 @llvm.atomic.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;val&gt;)
7208 declare i64 @llvm.atomic.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;val&gt;)
Andrew Lenharth95528942008-02-21 06:45:13 +00007209</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007210
Andrew Lenharth95528942008-02-21 06:45:13 +00007211<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007212<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7213 the value from memory. It then stores the value in <tt>val</tt> in the memory
7214 at <tt>ptr</tt>.</p>
7215
Andrew Lenharth95528942008-02-21 06:45:13 +00007216<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007217<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7218 the <tt>val</tt> argument and the result must be integers of the same bit
7219 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7220 integer type. The targets may only lower integer representations they
7221 support.</p>
Andrew Lenharth95528942008-02-21 06:45:13 +00007222
Andrew Lenharth95528942008-02-21 06:45:13 +00007223<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007224<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7225 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7226 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharth95528942008-02-21 06:45:13 +00007227
Andrew Lenharth95528942008-02-21 06:45:13 +00007228<h5>Examples:</h5>
7229<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007230%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7231%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth95528942008-02-21 06:45:13 +00007232 store i32 4, %ptr
7233
7234%val1 = add i32 4, 4
Dan Gohmand6a6f612010-05-28 17:07:41 +00007235%result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
Andrew Lenharth95528942008-02-21 06:45:13 +00007236 <i>; yields {i32}:result1 = 4</i>
7237%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7238%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7239
7240%val2 = add i32 1, 1
Dan Gohmand6a6f612010-05-28 17:07:41 +00007241%result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
Andrew Lenharth95528942008-02-21 06:45:13 +00007242 <i>; yields {i32}:result2 = 8</i>
7243
7244%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7245%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7246</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007247
Andrew Lenharth95528942008-02-21 06:45:13 +00007248</div>
7249
7250<!-- _______________________________________________________________________ -->
7251<div class="doc_subsubsection">
Mon P Wang6a490372008-06-25 08:15:39 +00007252 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharth95528942008-02-21 06:45:13 +00007253
7254</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007255
Andrew Lenharth95528942008-02-21 06:45:13 +00007256<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007257
Andrew Lenharth95528942008-02-21 06:45:13 +00007258<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007259<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7260 any integer bit width. Not all targets support all bit widths however.</p>
7261
Andrew Lenharth95528942008-02-21 06:45:13 +00007262<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007263 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7264 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7265 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7266 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Andrew Lenharth95528942008-02-21 06:45:13 +00007267</pre>
Andrew Lenharth95528942008-02-21 06:45:13 +00007268
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007269<h5>Overview:</h5>
7270<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7271 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7272
7273<h5>Arguments:</h5>
7274<p>The intrinsic takes two arguments, the first a pointer to an integer value
7275 and the second an integer value. The result is also an integer value. These
7276 integer types can have any bit width, but they must all have the same bit
7277 width. The targets may only lower integer representations they support.</p>
7278
Andrew Lenharth95528942008-02-21 06:45:13 +00007279<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007280<p>This intrinsic does a series of operations atomically. It first loads the
7281 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7282 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
Andrew Lenharth95528942008-02-21 06:45:13 +00007283
7284<h5>Examples:</h5>
7285<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007286%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7287%ptr = bitcast i8* %mallocP to i32*
7288 store i32 4, %ptr
Dan Gohmand6a6f612010-05-28 17:07:41 +00007289%result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
Andrew Lenharth95528942008-02-21 06:45:13 +00007290 <i>; yields {i32}:result1 = 4</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007291%result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
Andrew Lenharth95528942008-02-21 06:45:13 +00007292 <i>; yields {i32}:result2 = 8</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007293%result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
Andrew Lenharth95528942008-02-21 06:45:13 +00007294 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6a490372008-06-25 08:15:39 +00007295%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharth95528942008-02-21 06:45:13 +00007296</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007297
Andrew Lenharth95528942008-02-21 06:45:13 +00007298</div>
7299
Mon P Wang6a490372008-06-25 08:15:39 +00007300<!-- _______________________________________________________________________ -->
7301<div class="doc_subsubsection">
7302 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
7303
7304</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007305
Mon P Wang6a490372008-06-25 08:15:39 +00007306<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007307
Mon P Wang6a490372008-06-25 08:15:39 +00007308<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007309<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
7310 any integer bit width and for different address spaces. Not all targets
7311 support all bit widths however.</p>
7312
Mon P Wang6a490372008-06-25 08:15:39 +00007313<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007314 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7315 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7316 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7317 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007318</pre>
Mon P Wang6a490372008-06-25 08:15:39 +00007319
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007320<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00007321<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007322 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7323
7324<h5>Arguments:</h5>
7325<p>The intrinsic takes two arguments, the first a pointer to an integer value
7326 and the second an integer value. The result is also an integer value. These
7327 integer types can have any bit width, but they must all have the same bit
7328 width. The targets may only lower integer representations they support.</p>
7329
Mon P Wang6a490372008-06-25 08:15:39 +00007330<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007331<p>This intrinsic does a series of operations atomically. It first loads the
7332 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
7333 result to <tt>ptr</tt>. It yields the original value stored
7334 at <tt>ptr</tt>.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007335
7336<h5>Examples:</h5>
7337<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007338%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7339%ptr = bitcast i8* %mallocP to i32*
7340 store i32 8, %ptr
Dan Gohmand6a6f612010-05-28 17:07:41 +00007341%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
Mon P Wang6a490372008-06-25 08:15:39 +00007342 <i>; yields {i32}:result1 = 8</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007343%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
Mon P Wang6a490372008-06-25 08:15:39 +00007344 <i>; yields {i32}:result2 = 4</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007345%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
Mon P Wang6a490372008-06-25 08:15:39 +00007346 <i>; yields {i32}:result3 = 2</i>
7347%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
7348</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007349
Mon P Wang6a490372008-06-25 08:15:39 +00007350</div>
7351
7352<!-- _______________________________________________________________________ -->
7353<div class="doc_subsubsection">
7354 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
7355 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
7356 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
7357 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
Mon P Wang6a490372008-06-25 08:15:39 +00007358</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007359
Mon P Wang6a490372008-06-25 08:15:39 +00007360<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007361
Mon P Wang6a490372008-06-25 08:15:39 +00007362<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007363<p>These are overloaded intrinsics. You can
7364 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
7365 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
7366 bit width and for different address spaces. Not all targets support all bit
7367 widths however.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007368
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007369<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007370 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7371 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7372 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7373 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007374</pre>
7375
7376<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007377 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7378 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7379 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7380 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007381</pre>
7382
7383<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007384 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7385 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7386 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7387 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007388</pre>
7389
7390<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007391 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7392 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7393 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7394 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007395</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007396
Mon P Wang6a490372008-06-25 08:15:39 +00007397<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007398<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
7399 the value stored in memory at <tt>ptr</tt>. It yields the original value
7400 at <tt>ptr</tt>.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007401
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007402<h5>Arguments:</h5>
7403<p>These intrinsics take two arguments, the first a pointer to an integer value
7404 and the second an integer value. The result is also an integer value. These
7405 integer types can have any bit width, but they must all have the same bit
7406 width. The targets may only lower integer representations they support.</p>
7407
Mon P Wang6a490372008-06-25 08:15:39 +00007408<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007409<p>These intrinsics does a series of operations atomically. They first load the
7410 value stored at <tt>ptr</tt>. They then do the bitwise
7411 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
7412 original value stored at <tt>ptr</tt>.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007413
7414<h5>Examples:</h5>
7415<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007416%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7417%ptr = bitcast i8* %mallocP to i32*
7418 store i32 0x0F0F, %ptr
Dan Gohmand6a6f612010-05-28 17:07:41 +00007419%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang6a490372008-06-25 08:15:39 +00007420 <i>; yields {i32}:result0 = 0x0F0F</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007421%result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang6a490372008-06-25 08:15:39 +00007422 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007423%result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang6a490372008-06-25 08:15:39 +00007424 <i>; yields {i32}:result2 = 0xF0</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007425%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang6a490372008-06-25 08:15:39 +00007426 <i>; yields {i32}:result3 = FF</i>
7427%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
7428</pre>
Mon P Wang6a490372008-06-25 08:15:39 +00007429
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007430</div>
Mon P Wang6a490372008-06-25 08:15:39 +00007431
7432<!-- _______________________________________________________________________ -->
7433<div class="doc_subsubsection">
7434 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
7435 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
7436 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
7437 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
Mon P Wang6a490372008-06-25 08:15:39 +00007438</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007439
Mon P Wang6a490372008-06-25 08:15:39 +00007440<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007441
Mon P Wang6a490372008-06-25 08:15:39 +00007442<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007443<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
7444 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
7445 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
7446 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007447
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007448<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007449 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7450 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7451 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7452 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007453</pre>
7454
7455<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007456 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7457 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7458 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7459 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007460</pre>
7461
7462<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007463 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7464 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7465 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7466 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007467</pre>
7468
7469<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007470 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7471 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7472 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7473 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6a490372008-06-25 08:15:39 +00007474</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007475
Mon P Wang6a490372008-06-25 08:15:39 +00007476<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00007477<p>These intrinsics takes the signed or unsigned minimum or maximum of
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007478 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7479 original value at <tt>ptr</tt>.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007480
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007481<h5>Arguments:</h5>
7482<p>These intrinsics take two arguments, the first a pointer to an integer value
7483 and the second an integer value. The result is also an integer value. These
7484 integer types can have any bit width, but they must all have the same bit
7485 width. The targets may only lower integer representations they support.</p>
7486
Mon P Wang6a490372008-06-25 08:15:39 +00007487<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007488<p>These intrinsics does a series of operations atomically. They first load the
7489 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7490 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7491 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang6a490372008-06-25 08:15:39 +00007492
7493<h5>Examples:</h5>
7494<pre>
Victor Hernandeza70c6df2009-10-26 23:44:29 +00007495%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7496%ptr = bitcast i8* %mallocP to i32*
7497 store i32 7, %ptr
Dan Gohmand6a6f612010-05-28 17:07:41 +00007498%result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
Mon P Wang6a490372008-06-25 08:15:39 +00007499 <i>; yields {i32}:result0 = 7</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007500%result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
Mon P Wang6a490372008-06-25 08:15:39 +00007501 <i>; yields {i32}:result1 = -2</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007502%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
Mon P Wang6a490372008-06-25 08:15:39 +00007503 <i>; yields {i32}:result2 = 8</i>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007504%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
Mon P Wang6a490372008-06-25 08:15:39 +00007505 <i>; yields {i32}:result3 = 8</i>
7506%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7507</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007508
Mon P Wang6a490372008-06-25 08:15:39 +00007509</div>
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007510
Nick Lewycky6f7d8342009-10-13 07:03:23 +00007511
7512<!-- ======================================================================= -->
7513<div class="doc_subsection">
7514 <a name="int_memorymarkers">Memory Use Markers</a>
7515</div>
7516
7517<div class="doc_text">
7518
7519<p>This class of intrinsics exists to information about the lifetime of memory
7520 objects and ranges where variables are immutable.</p>
7521
7522</div>
7523
7524<!-- _______________________________________________________________________ -->
7525<div class="doc_subsubsection">
7526 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7527</div>
7528
7529<div class="doc_text">
7530
7531<h5>Syntax:</h5>
7532<pre>
7533 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7534</pre>
7535
7536<h5>Overview:</h5>
7537<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7538 object's lifetime.</p>
7539
7540<h5>Arguments:</h5>
Nick Lewycky9bc89042009-10-13 07:57:33 +00007541<p>The first argument is a constant integer representing the size of the
7542 object, or -1 if it is variable sized. The second argument is a pointer to
7543 the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00007544
7545<h5>Semantics:</h5>
7546<p>This intrinsic indicates that before this point in the code, the value of the
7547 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
Nick Lewyckyd20fd592009-10-27 16:56:58 +00007548 never be used and has an undefined value. A load from the pointer that
7549 precedes this intrinsic can be replaced with
Nick Lewycky6f7d8342009-10-13 07:03:23 +00007550 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7551
7552</div>
7553
7554<!-- _______________________________________________________________________ -->
7555<div class="doc_subsubsection">
7556 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7557</div>
7558
7559<div class="doc_text">
7560
7561<h5>Syntax:</h5>
7562<pre>
7563 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7564</pre>
7565
7566<h5>Overview:</h5>
7567<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7568 object's lifetime.</p>
7569
7570<h5>Arguments:</h5>
Nick Lewycky9bc89042009-10-13 07:57:33 +00007571<p>The first argument is a constant integer representing the size of the
7572 object, or -1 if it is variable sized. The second argument is a pointer to
7573 the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00007574
7575<h5>Semantics:</h5>
7576<p>This intrinsic indicates that after this point in the code, the value of the
7577 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7578 never be used and has an undefined value. Any stores into the memory object
7579 following this intrinsic may be removed as dead.
7580
7581</div>
7582
7583<!-- _______________________________________________________________________ -->
7584<div class="doc_subsubsection">
7585 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7586</div>
7587
7588<div class="doc_text">
7589
7590<h5>Syntax:</h5>
7591<pre>
7592 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;) readonly
7593</pre>
7594
7595<h5>Overview:</h5>
7596<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7597 a memory object will not change.</p>
7598
7599<h5>Arguments:</h5>
Nick Lewycky9bc89042009-10-13 07:57:33 +00007600<p>The first argument is a constant integer representing the size of the
7601 object, or -1 if it is variable sized. The second argument is a pointer to
7602 the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00007603
7604<h5>Semantics:</h5>
7605<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7606 the return value, the referenced memory location is constant and
7607 unchanging.</p>
7608
7609</div>
7610
7611<!-- _______________________________________________________________________ -->
7612<div class="doc_subsubsection">
7613 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7614</div>
7615
7616<div class="doc_text">
7617
7618<h5>Syntax:</h5>
7619<pre>
7620 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7621</pre>
7622
7623<h5>Overview:</h5>
7624<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7625 a memory object are mutable.</p>
7626
7627<h5>Arguments:</h5>
7628<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky9bc89042009-10-13 07:57:33 +00007629 The second argument is a constant integer representing the size of the
7630 object, or -1 if it is variable sized and the third argument is a pointer
7631 to the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00007632
7633<h5>Semantics:</h5>
7634<p>This intrinsic indicates that the memory is mutable again.</p>
7635
7636</div>
7637
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00007638<!-- ======================================================================= -->
7639<div class="doc_subsection">
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007640 <a name="int_general">General Intrinsics</a>
7641</div>
7642
7643<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007644
7645<p>This class of intrinsics is designed to be generic and has no specific
7646 purpose.</p>
7647
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007648</div>
7649
7650<!-- _______________________________________________________________________ -->
7651<div class="doc_subsubsection">
7652 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7653</div>
7654
7655<div class="doc_text">
7656
7657<h5>Syntax:</h5>
7658<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007659 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007660</pre>
7661
7662<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007663<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007664
7665<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007666<p>The first argument is a pointer to a value, the second is a pointer to a
7667 global string, the third is a pointer to a global string which is the source
7668 file name, and the last argument is the line number.</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007669
7670<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007671<p>This intrinsic allows annotation of local variables with arbitrary strings.
7672 This can be useful for special purpose optimizations that want to look for
7673 these annotations. These have no other defined use, they are ignored by code
7674 generation and optimization.</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007675
Tanya Lattnercb1b9602007-06-15 20:50:54 +00007676</div>
7677
Tanya Lattner293c0372007-09-21 22:59:12 +00007678<!-- _______________________________________________________________________ -->
7679<div class="doc_subsubsection">
Tanya Lattner0186a652007-09-21 23:57:59 +00007680 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattner293c0372007-09-21 22:59:12 +00007681</div>
7682
7683<div class="doc_text">
7684
7685<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007686<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7687 any integer bit width.</p>
7688
Tanya Lattner293c0372007-09-21 22:59:12 +00007689<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007690 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7691 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7692 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7693 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7694 declare i256 @llvm.annotation.i256(i256 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Tanya Lattner293c0372007-09-21 22:59:12 +00007695</pre>
7696
7697<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007698<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattner293c0372007-09-21 22:59:12 +00007699
7700<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007701<p>The first argument is an integer value (result of some expression), the
7702 second is a pointer to a global string, the third is a pointer to a global
7703 string which is the source file name, and the last argument is the line
7704 number. It returns the value of the first argument.</p>
Tanya Lattner293c0372007-09-21 22:59:12 +00007705
7706<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007707<p>This intrinsic allows annotations to be put on arbitrary expressions with
7708 arbitrary strings. This can be useful for special purpose optimizations that
7709 want to look for these annotations. These have no other defined use, they
7710 are ignored by code generation and optimization.</p>
Tanya Lattner293c0372007-09-21 22:59:12 +00007711
Tanya Lattner293c0372007-09-21 22:59:12 +00007712</div>
Jim Laskey2211f492007-03-14 19:31:19 +00007713
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00007714<!-- _______________________________________________________________________ -->
7715<div class="doc_subsubsection">
7716 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7717</div>
7718
7719<div class="doc_text">
7720
7721<h5>Syntax:</h5>
7722<pre>
7723 declare void @llvm.trap()
7724</pre>
7725
7726<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007727<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00007728
7729<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007730<p>None.</p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00007731
7732<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007733<p>This intrinsics is lowered to the target dependent trap instruction. If the
7734 target does not have a trap instruction, this intrinsic will be lowered to
7735 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00007736
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00007737</div>
7738
Bill Wendling14313312008-11-19 05:56:17 +00007739<!-- _______________________________________________________________________ -->
7740<div class="doc_subsubsection">
Misha Brukman50de2b22008-11-22 23:55:29 +00007741 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendling14313312008-11-19 05:56:17 +00007742</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007743
Bill Wendling14313312008-11-19 05:56:17 +00007744<div class="doc_text">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007745
Bill Wendling14313312008-11-19 05:56:17 +00007746<h5>Syntax:</h5>
7747<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007748 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling14313312008-11-19 05:56:17 +00007749</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007750
Bill Wendling14313312008-11-19 05:56:17 +00007751<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007752<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7753 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7754 ensure that it is placed on the stack before local variables.</p>
7755
Bill Wendling14313312008-11-19 05:56:17 +00007756<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007757<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7758 arguments. The first argument is the value loaded from the stack
7759 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7760 that has enough space to hold the value of the guard.</p>
7761
Bill Wendling14313312008-11-19 05:56:17 +00007762<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007763<p>This intrinsic causes the prologue/epilogue inserter to force the position of
7764 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7765 stack. This is to ensure that if a local variable on the stack is
7766 overwritten, it will destroy the value of the guard. When the function exits,
7767 the guard on the stack is checked against the original guard. If they're
7768 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7769 function.</p>
7770
Bill Wendling14313312008-11-19 05:56:17 +00007771</div>
7772
Eric Christopher73484322009-11-30 08:03:53 +00007773<!-- _______________________________________________________________________ -->
7774<div class="doc_subsubsection">
7775 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7776</div>
7777
7778<div class="doc_text">
7779
7780<h5>Syntax:</h5>
7781<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00007782 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;type&gt;)
7783 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;type&gt;)
Eric Christopher73484322009-11-30 08:03:53 +00007784</pre>
7785
7786<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00007787<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
Eric Christopher3070e162010-01-08 21:42:39 +00007788 to the optimizers to discover at compile time either a) when an
Eric Christopher455c5772009-12-05 02:46:03 +00007789 operation like memcpy will either overflow a buffer that corresponds to
7790 an object, or b) to determine that a runtime check for overflow isn't
7791 necessary. An object in this context means an allocation of a
Eric Christopher31e39bd2009-12-23 00:29:49 +00007792 specific class, structure, array, or other object.</p>
Eric Christopher73484322009-11-30 08:03:53 +00007793
7794<h5>Arguments:</h5>
7795<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher31e39bd2009-12-23 00:29:49 +00007796 argument is a pointer to or into the <tt>object</tt>. The second argument
7797 is a boolean 0 or 1. This argument determines whether you want the
7798 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7799 1, variables are not allowed.</p>
7800
Eric Christopher73484322009-11-30 08:03:53 +00007801<h5>Semantics:</h5>
7802<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Eric Christopher455c5772009-12-05 02:46:03 +00007803 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7804 (depending on the <tt>type</tt> argument if the size cannot be determined
7805 at compile time.</p>
Eric Christopher73484322009-11-30 08:03:53 +00007806
7807</div>
7808
Chris Lattner2f7c9632001-06-06 20:29:01 +00007809<!-- *********************************************************************** -->
Chris Lattner2f7c9632001-06-06 20:29:01 +00007810<hr>
Misha Brukmanc501f552004-03-01 17:47:27 +00007811<address>
7812 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +00007813 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmanc501f552004-03-01 17:47:27 +00007814 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +00007815 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukmanc501f552004-03-01 17:47:27 +00007816
7817 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencerca058542006-03-14 05:39:39 +00007818 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmanc501f552004-03-01 17:47:27 +00007819 Last modified: $Date$
7820</address>
Chris Lattnerb8f816e2008-01-04 04:33:49 +00007821
Misha Brukman76307852003-11-08 01:05:38 +00007822</body>
7823</html>