blob: b222b063194a0255c30cc8db918233f1378df4f0 [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.">
Daniel Dunbar46d611a2012-04-19 20:20:34 +000010 <link rel="stylesheet" href="_static/llvm.css" type="text/css">
Misha Brukman76307852003-11-08 01:05:38 +000011</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
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +000015<h1>LLVM Language Reference Manual</h1>
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 Wendling578ee402010-08-20 22:05:50 +000028 <li><a href="#linkage_linker_private_weak_def_auto">'<tt>linker_private_weak_def_auto</tt>' Linkage</a></li>
Bill Wendling8693ef82009-07-20 02:41:50 +000029 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
30 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
31 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
32 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
33 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
35 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner80d73c72009-10-10 18:26:06 +000036 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendling8693ef82009-07-20 02:41:50 +000037 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
Bill Wendlingb4d076e2011-10-11 06:41:28 +000038 <li><a href="#linkage_external">'<tt>external</tt>' Linkage</a></li>
Bill Wendling8693ef82009-07-20 02:41:50 +000039 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
40 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendlinga3c6f6b2009-07-20 01:03:30 +000041 </ol>
42 </li>
Chris Lattner0132aff2005-05-06 22:57:40 +000043 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnerbc088212009-01-11 20:53:49 +000044 <li><a href="#namedtypes">Named Types</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000045 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +000046 <li><a href="#functionstructure">Functions</a></li>
Dan Gohmanef9462f2008-10-14 16:51:45 +000047 <li><a href="#aliasstructure">Aliases</a></li>
Devang Pateld1a89692010-01-11 19:35:55 +000048 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +000049 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel9eb525d2008-09-26 23:51:19 +000050 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen71183b62007-12-10 03:18:06 +000051 <li><a href="#gc">Garbage Collector Names</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +000052 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencer50c723a2007-02-19 23:54:10 +000053 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman6154a012009-07-27 18:07:55 +000054 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +000055 <li><a href="#volatile">Volatile Memory Accesses</a></li>
Eli Friedman35b54aa2011-07-20 21:35:53 +000056 <li><a href="#memmodel">Memory Model for Concurrent Operations</a></li>
Eli Friedmanc9a551e2011-07-28 21:48:00 +000057 <li><a href="#ordering">Atomic Memory Ordering Constraints</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +000058 </ol>
59 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000060 <li><a href="#typesystem">Type System</a>
61 <ol>
Chris Lattner7824d182008-01-04 04:32:38 +000062 <li><a href="#t_classifications">Type Classifications</a></li>
Eric Christopher455c5772009-12-05 02:46:03 +000063 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner48b383b02003-11-25 01:02:51 +000064 <ol>
Nick Lewycky84a1eeb2009-09-27 00:45:11 +000065 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner7824d182008-01-04 04:32:38 +000066 <li><a href="#t_floating">Floating Point Types</a></li>
Dale Johannesen33e5c352010-10-01 00:48:59 +000067 <li><a href="#t_x86mmx">X86mmx Type</a></li>
Chris Lattner7824d182008-01-04 04:32:38 +000068 <li><a href="#t_void">Void Type</a></li>
69 <li><a href="#t_label">Label Type</a></li>
Nick Lewyckyadbc2842009-05-30 05:06:04 +000070 <li><a href="#t_metadata">Metadata Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000071 </ol>
72 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +000073 <li><a href="#t_derived">Derived Types</a>
74 <ol>
Chris Lattner392be582010-02-12 20:49:41 +000075 <li><a href="#t_aggregate">Aggregate Types</a>
76 <ol>
77 <li><a href="#t_array">Array Type</a></li>
78 <li><a href="#t_struct">Structure Type</a></li>
Chris Lattner2a843822011-07-23 19:59:08 +000079 <li><a href="#t_opaque">Opaque Structure Types</a></li>
Chris Lattner392be582010-02-12 20:49:41 +000080 <li><a href="#t_vector">Vector Type</a></li>
81 </ol>
82 </li>
Misha Brukman76307852003-11-08 01:05:38 +000083 <li><a href="#t_function">Function Type</a></li>
84 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +000085 </ol>
86 </li>
87 </ol>
88 </li>
Chris Lattner6af02f32004-12-09 16:11:40 +000089 <li><a href="#constants">Constants</a>
Chris Lattner74d3f822004-12-09 17:30:23 +000090 <ol>
Dan Gohmanef9462f2008-10-14 16:51:45 +000091 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner361bfcd2009-02-28 18:32:25 +000092 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohmanef9462f2008-10-14 16:51:45 +000093 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
94 <li><a href="#undefvalues">Undefined Values</a></li>
Dan Gohman9a2a0932011-12-06 03:18:47 +000095 <li><a href="#poisonvalues">Poison Values</a></li>
Chris Lattner2bfd3202009-10-27 21:19:13 +000096 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
Dan Gohmanef9462f2008-10-14 16:51:45 +000097 <li><a href="#constantexprs">Constant Expressions</a></li>
Chris Lattner74d3f822004-12-09 17:30:23 +000098 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +000099 </li>
Chris Lattner98f013c2006-01-25 23:47:57 +0000100 <li><a href="#othervalues">Other Values</a>
101 <ol>
Dan Gohmanef9462f2008-10-14 16:51:45 +0000102 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Peter Collingbourneec9ff672011-10-27 19:19:07 +0000103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a>
104 <ol>
105 <li><a href="#tbaa">'<tt>tbaa</tt>' Metadata</a></li>
Duncan Sands34bd91a2012-04-14 12:36:06 +0000106 <li><a href="#fpmath">'<tt>fpmath</tt>' Metadata</a></li>
Rafael Espindolaef9f5502012-03-24 00:14:51 +0000107 <li><a href="#range">'<tt>range</tt>' Metadata</a></li>
Peter Collingbourneec9ff672011-10-27 19:19:07 +0000108 </ol>
109 </li>
Chris Lattner98f013c2006-01-25 23:47:57 +0000110 </ol>
111 </li>
Bill Wendling911fdf42012-02-11 11:59:36 +0000112 <li><a href="#module_flags">Module Flags Metadata</a>
113 <ol>
Bill Wendling73462772012-02-16 01:10:50 +0000114 <li><a href="#objc_gc_flags">Objective-C Garbage Collection Module Flags Metadata</a></li>
Bill Wendling911fdf42012-02-11 11:59:36 +0000115 </ol>
116 </li>
Chris Lattnerae76db52009-07-20 05:55:19 +0000117 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
118 <ol>
119 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner58f9bb22009-07-20 06:14:25 +0000120 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
121 Global Variable</a></li>
Chris Lattnerae76db52009-07-20 05:55:19 +0000122 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
123 Global Variable</a></li>
124 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
125 Global Variable</a></li>
126 </ol>
127 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000128 <li><a href="#instref">Instruction Reference</a>
129 <ol>
130 <li><a href="#terminators">Terminator Instructions</a>
131 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000132 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
133 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000134 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +0000135 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000136 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Bill Wendlingf891bf82011-07-31 06:30:59 +0000137 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000138 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000139 </ol>
140 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000141 <li><a href="#binaryops">Binary Operations</a>
142 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000143 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmana5b96452009-06-04 22:49:04 +0000144 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000145 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmana5b96452009-06-04 22:49:04 +0000146 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000147 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmana5b96452009-06-04 22:49:04 +0000148 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000149 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
150 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
151 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000152 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
153 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
154 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000155 </ol>
156 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000157 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
158 <ol>
Reid Spencer2ab01932007-02-02 13:57:07 +0000159 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
160 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
161 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000162 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000163 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000164 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000165 </ol>
166 </li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000167 <li><a href="#vectorops">Vector Operations</a>
168 <ol>
169 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
170 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
171 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000172 </ol>
173 </li>
Dan Gohmanb9d66602008-05-12 23:51:09 +0000174 <li><a href="#aggregateops">Aggregate Operations</a>
175 <ol>
176 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
177 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
178 </ol>
179 </li>
Chris Lattner6ab66722006-08-15 00:45:58 +0000180 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000181 <ol>
Eli Friedmanc9a551e2011-07-28 21:48:00 +0000182 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
183 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
184 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
185 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
186 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
187 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
Robert Bocchino820bc75b2006-02-17 21:18:08 +0000188 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000189 </ol>
190 </li>
Reid Spencer97c5fa42006-11-08 01:18:52 +0000191 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000192 <ol>
193 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
194 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
195 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
196 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
197 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencer51b07252006-11-09 23:03:26 +0000198 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
199 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
200 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
201 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000202 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
203 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5b950642006-11-11 23:08:07 +0000204 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000205 </ol>
Dan Gohmanef9462f2008-10-14 16:51:45 +0000206 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000207 <li><a href="#otherops">Other Operations</a>
208 <ol>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000209 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
210 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000211 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000212 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000213 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattner33337472006-01-13 23:26:01 +0000214 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +0000215 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000216 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000217 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000218 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000219 </li>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000220 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000221 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000222 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
223 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000224 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
225 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
226 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000227 </ol>
228 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000229 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
230 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000231 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
232 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
233 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000234 </ol>
235 </li>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000236 <li><a href="#int_codegen">Code Generator Intrinsics</a>
237 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000238 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
239 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
240 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
241 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
242 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
243 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohmane58f7b32010-05-26 21:56:15 +0000244 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswellaa1c3c12004-04-09 16:43:20 +0000245 </ol>
246 </li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000247 <li><a href="#int_libc">Standard C Library Intrinsics</a>
248 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000249 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
250 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
251 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
252 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
253 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohmanb6324c12007-10-15 20:30:11 +0000254 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
255 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
256 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmane635c522011-05-27 00:36:31 +0000257 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
258 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
Cameron Zwarichf03fa182011-07-08 21:39:21 +0000259 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000260 </ol>
261 </li>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000262 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000263 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000264 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattnerb748c672006-01-16 22:34:14 +0000265 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
266 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
267 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000268 </ol>
269 </li>
Bill Wendlingf4d70622009-02-08 01:40:31 +0000270 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
271 <ol>
Bill Wendlingfd2bd722009-02-08 04:04:40 +0000272 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
273 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
274 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
275 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
276 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingb9a73272009-02-08 23:00:09 +0000277 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingf4d70622009-02-08 01:40:31 +0000278 </ol>
279 </li>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +0000280 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
281 <ol>
Chris Lattnerbbd8bd32010-03-14 23:03:31 +0000282 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
283 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +0000284 </ol>
285 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000286 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskey2211f492007-03-14 19:31:19 +0000287 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sandsa0984362011-09-06 13:37:06 +0000288 <li><a href="#int_trampoline">Trampoline Intrinsics</a>
Duncan Sands644f9172007-07-27 12:58:54 +0000289 <ol>
290 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sandsa0984362011-09-06 13:37:06 +0000291 <li><a href="#int_at">'<tt>llvm.adjust.trampoline</tt>' Intrinsic</a></li>
Duncan Sands644f9172007-07-27 12:58:54 +0000292 </ol>
293 </li>
Nick Lewycky6f7d8342009-10-13 07:03:23 +0000294 <li><a href="#int_memorymarkers">Memory Use Markers</a>
295 <ol>
Jakub Staszak5fd147f2011-12-04 20:44:25 +0000296 <li><a href="#int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a></li>
297 <li><a href="#int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a></li>
298 <li><a href="#int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a></li>
299 <li><a href="#int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a></li>
Nick Lewycky6f7d8342009-10-13 07:03:23 +0000300 </ol>
301 </li>
Reid Spencer5b2cb0f2007-07-20 19:59:11 +0000302 <li><a href="#int_general">General intrinsics</a>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000303 <ol>
Reid Spencer5b2cb0f2007-07-20 19:59:11 +0000304 <li><a href="#int_var_annotation">
Bill Wendling14313312008-11-19 05:56:17 +0000305 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattner293c0372007-09-21 22:59:12 +0000306 <li><a href="#int_annotation">
Bill Wendling14313312008-11-19 05:56:17 +0000307 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +0000308 <li><a href="#int_trap">
Bill Wendling14313312008-11-19 05:56:17 +0000309 '<tt>llvm.trap</tt>' Intrinsic</a></li>
310 <li><a href="#int_stackprotector">
311 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher73484322009-11-30 08:03:53 +0000312 <li><a href="#int_objectsize">
313 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Jakub Staszak5fef7922011-12-04 18:29:26 +0000314 <li><a href="#int_expect">
315 '<tt>llvm.expect</tt>' Intrinsic</a></li>
Tanya Lattner293c0372007-09-21 22:59:12 +0000316 </ol>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000317 </li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000318 </ol>
319 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000320</ol>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000321
322<div class="doc_author">
323 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
324 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman76307852003-11-08 01:05:38 +0000325</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000326
Chris Lattner2f7c9632001-06-06 20:29:01 +0000327<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000328<h2><a name="abstract">Abstract</a></h2>
Chris Lattner48b383b02003-11-25 01:02:51 +0000329<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000330
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000331<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000332
333<p>This document is a reference manual for the LLVM assembly language. LLVM is
334 a Static Single Assignment (SSA) based representation that provides type
335 safety, low-level operations, flexibility, and the capability of representing
336 'all' high-level languages cleanly. It is the common code representation
337 used throughout all phases of the LLVM compilation strategy.</p>
338
Misha Brukman76307852003-11-08 01:05:38 +0000339</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000340
Chris Lattner2f7c9632001-06-06 20:29:01 +0000341<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000342<h2><a name="introduction">Introduction</a></h2>
Chris Lattner48b383b02003-11-25 01:02:51 +0000343<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000344
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000345<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000346
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000347<p>The LLVM code representation is designed to be used in three different forms:
348 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
349 for fast loading by a Just-In-Time compiler), and as a human readable
350 assembly language representation. This allows LLVM to provide a powerful
351 intermediate representation for efficient compiler transformations and
352 analysis, while providing a natural means to debug and visualize the
353 transformations. The three different forms of LLVM are all equivalent. This
354 document describes the human readable representation and notation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000355
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000356<p>The LLVM representation aims to be light-weight and low-level while being
357 expressive, typed, and extensible at the same time. It aims to be a
358 "universal IR" of sorts, by being at a low enough level that high-level ideas
359 may be cleanly mapped to it (similar to how microprocessors are "universal
360 IR's", allowing many source languages to be mapped to them). By providing
361 type information, LLVM can be used as the target of optimizations: for
362 example, through pointer analysis, it can be proven that a C automatic
Bill Wendling7f4a3362009-11-02 00:24:16 +0000363 variable is never accessed outside of the current function, allowing it to
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000364 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000365
Chris Lattner2f7c9632001-06-06 20:29:01 +0000366<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000367<h4>
368 <a name="wellformed">Well-Formedness</a>
369</h4>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000370
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000371<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000372
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000373<p>It is important to note that this document describes 'well formed' LLVM
374 assembly language. There is a difference between what the parser accepts and
375 what is considered 'well formed'. For example, the following instruction is
376 syntactically okay, but not well formed:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000377
Benjamin Kramer79698be2010-07-13 12:26:09 +0000378<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +0000379%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattner757528b0b2004-05-23 21:06:01 +0000380</pre>
381
Bill Wendling7f4a3362009-11-02 00:24:16 +0000382<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
383 LLVM infrastructure provides a verification pass that may be used to verify
384 that an LLVM module is well formed. This pass is automatically run by the
385 parser after parsing input assembly and by the optimizer before it outputs
386 bitcode. The violations pointed out by the verifier pass indicate bugs in
387 transformation passes or input to the parser.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000388
Bill Wendling3716c5d2007-05-29 09:04:49 +0000389</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000390
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000391</div>
392
Chris Lattner87a3dbe2007-10-03 17:34:29 +0000393<!-- Describe the typesetting conventions here. -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000394
Chris Lattner2f7c9632001-06-06 20:29:01 +0000395<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000396<h2><a name="identifiers">Identifiers</a></h2>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000397<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000398
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000399<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000400
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000401<p>LLVM identifiers come in two basic types: global and local. Global
402 identifiers (functions, global variables) begin with the <tt>'@'</tt>
403 character. Local identifiers (register names, types) begin with
404 the <tt>'%'</tt> character. Additionally, there are three different formats
405 for identifiers, for different purposes:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000406
Chris Lattner2f7c9632001-06-06 20:29:01 +0000407<ol>
Reid Spencerb23b65f2007-08-07 14:34:28 +0000408 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000409 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
410 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
411 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
412 other characters in their names can be surrounded with quotes. Special
413 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
414 ASCII code for the character in hexadecimal. In this way, any character
415 can be used in a name value, even quotes themselves.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000416
Reid Spencerb23b65f2007-08-07 14:34:28 +0000417 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000418 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000419
Reid Spencer8f08d802004-12-09 18:02:53 +0000420 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000421 constants</a>, below.</li>
Misha Brukman76307852003-11-08 01:05:38 +0000422</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000423
Reid Spencerb23b65f2007-08-07 14:34:28 +0000424<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000425 don't need to worry about name clashes with reserved words, and the set of
426 reserved words may be expanded in the future without penalty. Additionally,
427 unnamed identifiers allow a compiler to quickly come up with a temporary
428 variable without having to avoid symbol table conflicts.</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000429
Chris Lattner48b383b02003-11-25 01:02:51 +0000430<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000431 languages. There are keywords for different opcodes
432 ('<tt><a href="#i_add">add</a></tt>',
433 '<tt><a href="#i_bitcast">bitcast</a></tt>',
434 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
435 ('<tt><a href="#t_void">void</a></tt>',
436 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
437 reserved words cannot conflict with variable names, because none of them
438 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000439
440<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000441 '<tt>%X</tt>' by 8:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000442
Misha Brukman76307852003-11-08 01:05:38 +0000443<p>The easy way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000444
Benjamin Kramer79698be2010-07-13 12:26:09 +0000445<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +0000446%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnerd79749a2004-12-09 16:36:40 +0000447</pre>
448
Misha Brukman76307852003-11-08 01:05:38 +0000449<p>After strength reduction:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000450
Benjamin Kramer79698be2010-07-13 12:26:09 +0000451<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +0000452%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnerd79749a2004-12-09 16:36:40 +0000453</pre>
454
Misha Brukman76307852003-11-08 01:05:38 +0000455<p>And the hard way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000456
Benjamin Kramer79698be2010-07-13 12:26:09 +0000457<pre class="doc_code">
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>
462
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000463<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
464 lexical features of LLVM:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000465
Chris Lattner2f7c9632001-06-06 20:29:01 +0000466<ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000467 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000468 line.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000469
470 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000471 assigned to a named value.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000472
Misha Brukman76307852003-11-08 01:05:38 +0000473 <li>Unnamed temporaries are numbered sequentially</li>
474</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000475
Bill Wendling7f4a3362009-11-02 00:24:16 +0000476<p>It also shows a convention that we follow in this document. When
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000477 demonstrating instructions, we will follow an instruction with a comment that
478 defines the type and name of value produced. Comments are shown in italic
479 text.</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000480
Misha Brukman76307852003-11-08 01:05:38 +0000481</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000482
483<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000484<h2><a name="highlevel">High Level Structure</a></h2>
Chris Lattner6af02f32004-12-09 16:11:40 +0000485<!-- *********************************************************************** -->
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000486<div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000487<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000488<h3>
489 <a name="modulestructure">Module Structure</a>
490</h3>
Chris Lattner6af02f32004-12-09 16:11:40 +0000491
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000492<div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000493
Bill Wendling21ee0d22012-03-14 08:07:43 +0000494<p>LLVM programs are composed of <tt>Module</tt>s, each of which is a
495 translation unit of the input programs. Each module consists of functions,
496 global variables, and symbol table entries. Modules may be combined together
497 with the LLVM linker, which merges function (and global variable)
498 definitions, resolves forward declarations, and merges symbol table
499 entries. Here is an example of the "hello world" module:</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000500
Benjamin Kramer79698be2010-07-13 12:26:09 +0000501<pre class="doc_code">
Chris Lattner54a7be72010-08-17 17:13:42 +0000502<i>; Declare the string constant as a global constant.</i>&nbsp;
Bill Wendling21ee0d22012-03-14 08:07:43 +0000503<a href="#identifiers">@.str</a> = <a href="#linkage_private">private</a>&nbsp;<a href="#globalvars">unnamed_addr</a>&nbsp;<a href="#globalvars">constant</a>&nbsp;<a href="#t_array">[13 x i8]</a> c"hello world\0A\00"&nbsp;
Chris Lattner6af02f32004-12-09 16:11:40 +0000504
Chris Lattner54a7be72010-08-17 17:13:42 +0000505<i>; External declaration of the puts function</i>&nbsp;
Bill Wendling21ee0d22012-03-14 08:07:43 +0000506<a href="#functionstructure">declare</a> i32 @puts(i8* <a href="#nocapture">nocapture</a>) <a href="#fnattrs">nounwind</a>&nbsp;
Chris Lattner6af02f32004-12-09 16:11:40 +0000507
508<i>; Definition of main function</i>
Chris Lattner54a7be72010-08-17 17:13:42 +0000509define i32 @main() { <i>; i32()* </i>&nbsp;
510 <i>; Convert [13 x i8]* to i8 *...</i>&nbsp;
Bill Wendling21ee0d22012-03-14 08:07:43 +0000511 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.str, i64 0, i64 0
Chris Lattner6af02f32004-12-09 16:11:40 +0000512
Chris Lattner54a7be72010-08-17 17:13:42 +0000513 <i>; Call puts function to write out the string to stdout.</i>&nbsp;
Bill Wendling21ee0d22012-03-14 08:07:43 +0000514 <a href="#i_call">call</a> i32 @puts(i8* %cast210)
Chris Lattner54a7be72010-08-17 17:13:42 +0000515 <a href="#i_ret">ret</a> i32 0&nbsp;
516}
Devang Pateld1a89692010-01-11 19:35:55 +0000517
518<i>; Named metadata</i>
Bill Wendling21ee0d22012-03-14 08:07:43 +0000519!1 = metadata !{i32 42}
Devang Pateld1a89692010-01-11 19:35:55 +0000520!foo = !{!1, null}
Bill Wendling3716c5d2007-05-29 09:04:49 +0000521</pre>
Chris Lattner6af02f32004-12-09 16:11:40 +0000522
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000523<p>This example is made up of a <a href="#globalvars">global variable</a> named
Bill Wendling21ee0d22012-03-14 08:07:43 +0000524 "<tt>.str</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000525 a <a href="#functionstructure">function definition</a> for
Devang Pateld1a89692010-01-11 19:35:55 +0000526 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
Bill Wendling21ee0d22012-03-14 08:07:43 +0000527 "<tt>foo</tt>".</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000528
Bill Wendling21ee0d22012-03-14 08:07:43 +0000529<p>In general, a module is made up of a list of global values (where both
530 functions and global variables are global values). Global values are
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000531 represented by a pointer to a memory location (in this case, a pointer to an
532 array of char, and a pointer to a function), and have one of the
533 following <a href="#linkage">linkage types</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000534
Chris Lattnerd79749a2004-12-09 16:36:40 +0000535</div>
536
537<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000538<h3>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000539 <a name="linkage">Linkage Types</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000540</h3>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000541
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000542<div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000543
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000544<p>All Global Variables and Functions have one of the following types of
545 linkage:</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000546
547<dl>
Bill Wendling7f4a3362009-11-02 00:24:16 +0000548 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling03bcd6e2010-07-01 21:55:59 +0000549 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
550 by objects in the current module. In particular, linking code into a
551 module with an private global value may cause the private to be renamed as
552 necessary to avoid collisions. Because the symbol is private to the
553 module, all references can be updated. This doesn't show up in any symbol
554 table in the object file.</dd>
Rafael Espindola6de96a12009-01-15 20:18:42 +0000555
Bill Wendling7f4a3362009-11-02 00:24:16 +0000556 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling03bcd6e2010-07-01 21:55:59 +0000557 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
558 assembler and evaluated by the linker. Unlike normal strong symbols, they
559 are removed by the linker from the final linked image (executable or
560 dynamic library).</dd>
561
562 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
563 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
564 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
565 linker. The symbols are removed by the linker from the final linked image
566 (executable or dynamic library).</dd>
Bill Wendlinga3c6f6b2009-07-20 01:03:30 +0000567
Bill Wendling578ee402010-08-20 22:05:50 +0000568 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
569 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
570 of the object is not taken. For instance, functions that had an inline
571 definition, but the compiler decided not to inline it. Note,
572 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
573 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
574 visibility. The symbols are removed by the linker from the final linked
575 image (executable or dynamic library).</dd>
576
Bill Wendling7f4a3362009-11-02 00:24:16 +0000577 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling36321712010-06-29 22:34:52 +0000578 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000579 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
580 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000581
Bill Wendling7f4a3362009-11-02 00:24:16 +0000582 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner184f1be2009-04-13 05:44:34 +0000583 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000584 into the object file corresponding to the LLVM module. They exist to
585 allow inlining and other optimizations to take place given knowledge of
586 the definition of the global, which is known to be somewhere outside the
587 module. Globals with <tt>available_externally</tt> linkage are allowed to
588 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
589 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner184f1be2009-04-13 05:44:34 +0000590
Bill Wendling7f4a3362009-11-02 00:24:16 +0000591 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattnere20b4702007-01-14 06:51:48 +0000592 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner0de4caa2010-01-09 19:15:14 +0000593 the same name when linkage occurs. This can be used to implement
594 some forms of inline functions, templates, or other code which must be
595 generated in each translation unit that uses it, but where the body may
596 be overridden with a more definitive definition later. Unreferenced
597 <tt>linkonce</tt> globals are allowed to be discarded. Note that
598 <tt>linkonce</tt> linkage does not actually allow the optimizer to
599 inline the body of this function into callers because it doesn't know if
600 this definition of the function is the definitive definition within the
601 program or whether it will be overridden by a stronger definition.
602 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
603 linkage.</dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000604
Bill Wendling7f4a3362009-11-02 00:24:16 +0000605 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattnerd0554882009-08-05 05:21:07 +0000606 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
607 <tt>linkonce</tt> linkage, except that unreferenced globals with
608 <tt>weak</tt> linkage may not be discarded. This is used for globals that
609 are declared "weak" in C source code.</dd>
610
Bill Wendling7f4a3362009-11-02 00:24:16 +0000611 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattnerd0554882009-08-05 05:21:07 +0000612 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
613 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
614 global scope.
615 Symbols with "<tt>common</tt>" linkage are merged in the same way as
616 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattner0aff0b22009-08-05 05:41:44 +0000617 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher455c5772009-12-05 02:46:03 +0000618 must have a zero initializer, and may not be marked '<a
Chris Lattner0aff0b22009-08-05 05:41:44 +0000619 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
620 have common linkage.</dd>
Chris Lattnerd0554882009-08-05 05:21:07 +0000621
Chris Lattnerd79749a2004-12-09 16:36:40 +0000622
Bill Wendling7f4a3362009-11-02 00:24:16 +0000623 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000624 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000625 pointer to array type. When two global variables with appending linkage
626 are linked together, the two global arrays are appended together. This is
627 the LLVM, typesafe, equivalent of having the system linker append together
628 "sections" with identical names when .o files are linked.</dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000629
Bill Wendling7f4a3362009-11-02 00:24:16 +0000630 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000631 <dd>The semantics of this linkage follow the ELF object file model: the symbol
632 is weak until linked, if not linked, the symbol becomes null instead of
633 being an undefined reference.</dd>
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000634
Bill Wendling7f4a3362009-11-02 00:24:16 +0000635 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
636 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000637 <dd>Some languages allow differing globals to be merged, such as two functions
638 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling03bcd6e2010-07-01 21:55:59 +0000639 that only equivalent globals are ever merged (the "one definition rule"
640 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000641 and <tt>weak_odr</tt> linkage types to indicate that the global will only
642 be merged with equivalent globals. These linkage types are otherwise the
643 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands12da8ce2009-03-07 15:45:40 +0000644
Bill Wendlingef3cdea2011-11-04 20:40:41 +0000645 <dt><tt><b><a name="linkage_external">external</a></b></tt></dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000646 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000647 visible, meaning that it participates in linkage and can be used to
648 resolve external symbol references.</dd>
Reid Spencer7972c472007-04-11 23:49:50 +0000649</dl>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000650
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000651<p>The next two types of linkage are targeted for Microsoft Windows platform
652 only. They are designed to support importing (exporting) symbols from (to)
653 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000654
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000655<dl>
Bill Wendling7f4a3362009-11-02 00:24:16 +0000656 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000657 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000658 or variable via a global pointer to a pointer that is set up by the DLL
659 exporting the symbol. On Microsoft Windows targets, the pointer name is
660 formed by combining <code>__imp_</code> and the function or variable
661 name.</dd>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000662
Bill Wendling7f4a3362009-11-02 00:24:16 +0000663 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000664 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000665 pointer to a pointer in a DLL, so that it can be referenced with the
666 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
667 name is formed by combining <code>__imp_</code> and the function or
668 variable name.</dd>
Chris Lattner6af02f32004-12-09 16:11:40 +0000669</dl>
670
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000671<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
672 another module defined a "<tt>.LC0</tt>" variable and was linked with this
673 one, one of the two would be renamed, preventing a collision. Since
674 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
675 declarations), they are accessible outside of the current module.</p>
676
677<p>It is illegal for a function <i>declaration</i> to have any linkage type
Bill Wendlingb4d076e2011-10-11 06:41:28 +0000678 other than <tt>external</tt>, <tt>dllimport</tt>
679 or <tt>extern_weak</tt>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000680
Duncan Sands12da8ce2009-03-07 15:45:40 +0000681<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000682 or <tt>weak_odr</tt> linkages.</p>
683
Chris Lattner6af02f32004-12-09 16:11:40 +0000684</div>
685
686<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000687<h3>
Chris Lattner0132aff2005-05-06 22:57:40 +0000688 <a name="callingconv">Calling Conventions</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000689</h3>
Chris Lattner0132aff2005-05-06 22:57:40 +0000690
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000691<div>
Chris Lattner0132aff2005-05-06 22:57:40 +0000692
693<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000694 and <a href="#i_invoke">invokes</a> can all have an optional calling
695 convention specified for the call. The calling convention of any pair of
696 dynamic caller/callee must match, or the behavior of the program is
697 undefined. The following calling conventions are supported by LLVM, and more
698 may be added in the future:</p>
Chris Lattner0132aff2005-05-06 22:57:40 +0000699
700<dl>
701 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000702 <dd>This calling convention (the default if no other calling convention is
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000703 specified) matches the target C calling conventions. This calling
704 convention supports varargs function calls and tolerates some mismatch in
705 the declared prototype and implemented declaration of the function (as
706 does normal C).</dd>
Chris Lattner0132aff2005-05-06 22:57:40 +0000707
708 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000709 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000710 (e.g. by passing things in registers). This calling convention allows the
711 target to use whatever tricks it wants to produce fast code for the
712 target, without having to conform to an externally specified ABI
Jeffrey Yasskinb8677462010-01-09 19:44:16 +0000713 (Application Binary Interface).
714 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattnera179e4d2010-03-11 00:22:57 +0000715 when this or the GHC convention is used.</a> 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
719 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000720 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000721 as possible under the assumption that the call is not commonly executed.
722 As such, these calls often preserve all registers so that the call does
723 not break any live ranges in the caller side. This calling convention
724 does not support varargs and requires the prototype of all callees to
725 exactly match the prototype of the function definition.</dd>
Chris Lattner0132aff2005-05-06 22:57:40 +0000726
Chris Lattnera179e4d2010-03-11 00:22:57 +0000727 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
728 <dd>This calling convention has been implemented specifically for use by the
729 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
730 It passes everything in registers, going to extremes to achieve this by
731 disabling callee save registers. This calling convention should not be
732 used lightly but only for specific situations such as an alternative to
733 the <em>register pinning</em> performance technique often used when
734 implementing functional programming languages.At the moment only X86
735 supports this convention and it has the following limitations:
736 <ul>
737 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
738 floating point types are supported.</li>
739 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
740 6 floating point parameters.</li>
741 </ul>
742 This calling convention supports
743 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
744 requires both the caller and callee are using it.
745 </dd>
746
Chris Lattner573f64e2005-05-07 01:46:40 +0000747 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000748 <dd>Any calling convention may be specified by number, allowing
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000749 target-specific calling conventions to be used. Target specific calling
750 conventions start at 64.</dd>
Chris Lattner573f64e2005-05-07 01:46:40 +0000751</dl>
Chris Lattner0132aff2005-05-06 22:57:40 +0000752
753<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000754 support Pascal conventions or any other well-known target-independent
755 convention.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +0000756
757</div>
758
759<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000760<h3>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000761 <a name="visibility">Visibility Styles</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000762</h3>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000763
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000764<div>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000765
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000766<p>All Global Variables and Functions have one of the following visibility
767 styles:</p>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000768
769<dl>
770 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattner67c37d12008-08-05 18:29:16 +0000771 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000772 that the declaration is visible to other modules and, in shared libraries,
773 means that the declared entity may be overridden. On Darwin, default
774 visibility means that the declaration is visible to other modules. Default
775 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000776
777 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000778 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000779 object if they are in the same shared object. Usually, hidden visibility
780 indicates that the symbol will not be placed into the dynamic symbol
781 table, so no other module (executable or shared library) can reference it
782 directly.</dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000783
Anton Korobeynikov39f3cff2007-04-29 18:35:00 +0000784 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov39f3cff2007-04-29 18:35:00 +0000785 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000786 the dynamic symbol table, but that references within the defining module
787 will bind to the local symbol. That is, the symbol cannot be overridden by
788 another module.</dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000789</dl>
790
791</div>
792
793<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000794<h3>
Chris Lattnerbc088212009-01-11 20:53:49 +0000795 <a name="namedtypes">Named Types</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000796</h3>
Chris Lattnerbc088212009-01-11 20:53:49 +0000797
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000798<div>
Chris Lattnerbc088212009-01-11 20:53:49 +0000799
800<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000801 it easier to read the IR and make the IR more condensed (particularly when
802 recursive types are involved). An example of a name specification is:</p>
Chris Lattnerbc088212009-01-11 20:53:49 +0000803
Benjamin Kramer79698be2010-07-13 12:26:09 +0000804<pre class="doc_code">
Chris Lattnerbc088212009-01-11 20:53:49 +0000805%mytype = type { %mytype*, i32 }
806</pre>
Chris Lattnerbc088212009-01-11 20:53:49 +0000807
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000808<p>You may give a name to any <a href="#typesystem">type</a> except
Chris Lattner249b9762010-08-17 23:26:04 +0000809 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000810 is expected with the syntax "%mytype".</p>
Chris Lattnerbc088212009-01-11 20:53:49 +0000811
812<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000813 and that you can therefore specify multiple names for the same type. This
814 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
815 uses structural typing, the name is not part of the type. When printing out
816 LLVM IR, the printer will pick <em>one name</em> to render all types of a
817 particular shape. This means that if you have code where two different
818 source types end up having the same LLVM type, that the dumper will sometimes
819 print the "wrong" or unexpected type. This is an important design point and
820 isn't going to change.</p>
Chris Lattnerbc088212009-01-11 20:53:49 +0000821
822</div>
823
Chris Lattnerbc088212009-01-11 20:53:49 +0000824<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000825<h3>
Chris Lattner6af02f32004-12-09 16:11:40 +0000826 <a name="globalvars">Global Variables</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000827</h3>
Chris Lattner6af02f32004-12-09 16:11:40 +0000828
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000829<div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000830
Chris Lattner5d5aede2005-02-12 19:30:21 +0000831<p>Global variables define regions of memory allocated at compilation time
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000832 instead of run-time. Global variables may optionally be initialized, may
833 have an explicit section to be placed in, and may have an optional explicit
834 alignment specified. A variable may be defined as "thread_local", which
835 means that it will not be shared by threads (each thread will have a
836 separated copy of the variable). A variable may be defined as a global
837 "constant," which indicates that the contents of the variable
838 will <b>never</b> be modified (enabling better optimization, allowing the
839 global data to be placed in the read-only section of an executable, etc).
840 Note that variables that need runtime initialization cannot be marked
841 "constant" as there is a store to the variable.</p>
Chris Lattner5d5aede2005-02-12 19:30:21 +0000842
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000843<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
844 constant, even if the final definition of the global is not. This capability
845 can be used to enable slightly better optimization of the program, but
846 requires the language definition to guarantee that optimizations based on the
847 'constantness' are valid for the translation units that do not include the
848 definition.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000849
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000850<p>As SSA values, global variables define pointer values that are in scope
851 (i.e. they dominate) all basic blocks in the program. Global variables
852 always define a pointer to their "content" type because they describe a
853 region of memory, and all memory objects in LLVM are accessed through
854 pointers.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000855
Rafael Espindola45e6c192011-01-08 16:42:36 +0000856<p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
857 that the address is not significant, only the content. Constants marked
Rafael Espindolaf1ed7812011-01-15 08:20:57 +0000858 like this can be merged with other constants if they have the same
859 initializer. Note that a constant with significant address <em>can</em>
860 be merged with a <tt>unnamed_addr</tt> constant, the result being a
861 constant whose address is significant.</p>
Rafael Espindola45e6c192011-01-08 16:42:36 +0000862
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000863<p>A global variable may be declared to reside in a target-specific numbered
864 address space. For targets that support them, address spaces may affect how
865 optimizations are performed and/or what target instructions are used to
866 access the variable. The default address space is zero. The address space
867 qualifier must precede any other attributes.</p>
Christopher Lamb308121c2007-12-11 09:31:00 +0000868
Chris Lattner662c8722005-11-12 00:45:07 +0000869<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000870 supports it, it will emit globals to the section specified.</p>
Chris Lattner662c8722005-11-12 00:45:07 +0000871
Chris Lattner78e00bc2010-04-28 00:13:42 +0000872<p>An explicit alignment may be specified for a global, which must be a power
873 of 2. If not present, or if the alignment is set to zero, the alignment of
874 the global is set by the target to whatever it feels convenient. If an
875 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner4bd85e42010-04-28 00:31:12 +0000876 alignment. Targets and optimizers are not allowed to over-align the global
877 if the global has an assigned section. In this case, the extra alignment
878 could be observable: for example, code could assume that the globals are
879 densely packed in their section and try to iterate over them as an array,
880 alignment padding would break this iteration.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000881
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000882<p>For example, the following defines a global in a numbered address space with
883 an initializer, section, and alignment:</p>
Chris Lattner5760c502007-01-14 00:27:09 +0000884
Benjamin Kramer79698be2010-07-13 12:26:09 +0000885<pre class="doc_code">
Dan Gohmanaaa679b2009-01-11 00:40:00 +0000886@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner5760c502007-01-14 00:27:09 +0000887</pre>
888
Chris Lattner6af02f32004-12-09 16:11:40 +0000889</div>
890
891
892<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000893<h3>
Chris Lattner6af02f32004-12-09 16:11:40 +0000894 <a name="functionstructure">Functions</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000895</h3>
Chris Lattner6af02f32004-12-09 16:11:40 +0000896
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000897<div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000898
Dan Gohmana269a0a2010-03-01 17:41:39 +0000899<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000900 optional <a href="#linkage">linkage type</a>, an optional
901 <a href="#visibility">visibility style</a>, an optional
Rafael Espindola45e6c192011-01-08 16:42:36 +0000902 <a href="#callingconv">calling convention</a>,
903 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000904 <a href="#paramattrs">parameter attribute</a> for the return type, a function
905 name, a (possibly empty) argument list (each with optional
906 <a href="#paramattrs">parameter attributes</a>), optional
907 <a href="#fnattrs">function attributes</a>, an optional section, an optional
908 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
909 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000910
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000911<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
912 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher455c5772009-12-05 02:46:03 +0000913 <a href="#visibility">visibility style</a>, an optional
Rafael Espindola45e6c192011-01-08 16:42:36 +0000914 <a href="#callingconv">calling convention</a>,
915 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000916 <a href="#paramattrs">parameter attribute</a> for the return type, a function
917 name, a possibly empty list of arguments, an optional alignment, and an
918 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000919
Chris Lattner67c37d12008-08-05 18:29:16 +0000920<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000921 (Control Flow Graph) for the function. Each basic block may optionally start
922 with a label (giving the basic block a symbol table entry), contains a list
923 of instructions, and ends with a <a href="#terminators">terminator</a>
924 instruction (such as a branch or function return).</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000925
Chris Lattnera59fb102007-06-08 16:52:14 +0000926<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000927 executed on entrance to the function, and it is not allowed to have
928 predecessor basic blocks (i.e. there can not be any branches to the entry
929 block of a function). Because the block can have no predecessors, it also
930 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000931
Chris Lattner662c8722005-11-12 00:45:07 +0000932<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000933 supports it, it will emit functions to the section specified.</p>
Chris Lattner662c8722005-11-12 00:45:07 +0000934
Chris Lattner54611b42005-11-06 08:02:57 +0000935<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000936 the alignment is set to zero, the alignment of the function is set by the
937 target to whatever it feels convenient. If an explicit alignment is
938 specified, the function is forced to have at least that much alignment. All
939 alignments must be a power of 2.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000940
Rafael Espindola45e6c192011-01-08 16:42:36 +0000941<p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
Bill Wendlingef3cdea2011-11-04 20:40:41 +0000942 be significant and two identical functions can be merged.</p>
Rafael Espindola45e6c192011-01-08 16:42:36 +0000943
Bill Wendling30235112009-07-20 02:39:26 +0000944<h5>Syntax:</h5>
Benjamin Kramer79698be2010-07-13 12:26:09 +0000945<pre class="doc_code">
Chris Lattner0ae02092008-10-13 16:55:18 +0000946define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000947 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
948 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
949 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
950 [<a href="#gc">gc</a>] { ... }
951</pre>
Devang Patel02256232008-10-07 17:48:33 +0000952
Chris Lattner6af02f32004-12-09 16:11:40 +0000953</div>
954
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000955<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000956<h3>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000957 <a name="aliasstructure">Aliases</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000958</h3>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000959
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000960<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +0000961
962<p>Aliases act as "second name" for the aliasee value (which can be either
963 function, global variable, another alias or bitcast of global value). Aliases
964 may have an optional <a href="#linkage">linkage type</a>, and an
965 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000966
Bill Wendling30235112009-07-20 02:39:26 +0000967<h5>Syntax:</h5>
Benjamin Kramer79698be2010-07-13 12:26:09 +0000968<pre class="doc_code">
Duncan Sands7e99a942008-09-12 20:48:21 +0000969@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000970</pre>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000971
972</div>
973
Chris Lattner91c15c42006-01-23 23:23:47 +0000974<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000975<h3>
Devang Pateld1a89692010-01-11 19:35:55 +0000976 <a name="namedmetadatastructure">Named Metadata</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000977</h3>
Devang Pateld1a89692010-01-11 19:35:55 +0000978
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +0000979<div>
Devang Pateld1a89692010-01-11 19:35:55 +0000980
Chris Lattnerc2f8f162010-01-15 21:50:19 +0000981<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
Dan Gohman093cb792010-07-21 18:54:18 +0000982 nodes</a> (but not metadata strings) are the only valid operands for
Chris Lattnerc2f8f162010-01-15 21:50:19 +0000983 a named metadata.</p>
Devang Pateld1a89692010-01-11 19:35:55 +0000984
985<h5>Syntax:</h5>
Benjamin Kramer79698be2010-07-13 12:26:09 +0000986<pre class="doc_code">
Dan Gohman093cb792010-07-21 18:54:18 +0000987; Some unnamed metadata nodes, which are referenced by the named metadata.
988!0 = metadata !{metadata !"zero"}
Devang Pateld1a89692010-01-11 19:35:55 +0000989!1 = metadata !{metadata !"one"}
Dan Gohman093cb792010-07-21 18:54:18 +0000990!2 = metadata !{metadata !"two"}
Dan Gohman58cd65f2010-07-13 19:48:13 +0000991; A named metadata.
Dan Gohman093cb792010-07-21 18:54:18 +0000992!name = !{!0, !1, !2}
Devang Pateld1a89692010-01-11 19:35:55 +0000993</pre>
Devang Pateld1a89692010-01-11 19:35:55 +0000994
995</div>
996
997<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +0000998<h3>
999 <a name="paramattrs">Parameter Attributes</a>
1000</h3>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001001
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001002<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001003
1004<p>The return type and each parameter of a function type may have a set of
1005 <i>parameter attributes</i> associated with them. Parameter attributes are
1006 used to communicate additional information about the result or parameters of
1007 a function. Parameter attributes are considered to be part of the function,
1008 not of the function type, so functions with different parameter attributes
1009 can have the same function type.</p>
1010
1011<p>Parameter attributes are simple keywords that follow the type specified. If
1012 multiple parameter attributes are needed, they are space separated. For
1013 example:</p>
Bill Wendling3716c5d2007-05-29 09:04:49 +00001014
Benjamin Kramer79698be2010-07-13 12:26:09 +00001015<pre class="doc_code">
Nick Lewyckydac78d82009-02-15 23:06:14 +00001016declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerd2597d72008-10-04 18:33:34 +00001017declare i32 @atoi(i8 zeroext)
1018declare signext i8 @returns_signed_char()
Bill Wendling3716c5d2007-05-29 09:04:49 +00001019</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +00001020
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001021<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1022 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001023
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001024<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001025
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001026<dl>
Bill Wendling7f4a3362009-11-02 00:24:16 +00001027 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001028 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarichac106272011-03-16 22:20:18 +00001029 should be zero-extended to the extent required by the target's ABI (which
1030 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1031 parameter) or the callee (for a return value).</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001032
Bill Wendling7f4a3362009-11-02 00:24:16 +00001033 <dt><tt><b>signext</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001034 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarich341c36d2011-03-17 14:21:58 +00001035 should be sign-extended to the extent required by the target's ABI (which
1036 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1037 return value).</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001038
Bill Wendling7f4a3362009-11-02 00:24:16 +00001039 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001040 <dd>This indicates that this parameter or return value should be treated in a
1041 special target-dependent fashion during while emitting code for a function
1042 call or return (usually, by putting it in a register as opposed to memory,
1043 though some targets use it to distinguish between two different kinds of
1044 registers). Use of this attribute is target-specific.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +00001045
Bill Wendling7f4a3362009-11-02 00:24:16 +00001046 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Chris Lattnerd78dbee2010-11-20 23:49:06 +00001047 <dd><p>This indicates that the pointer parameter should really be passed by
1048 value to the function. The attribute implies that a hidden copy of the
1049 pointee
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001050 is made between the caller and the callee, so the callee is unable to
1051 modify the value in the callee. This attribute is only valid on LLVM
1052 pointer arguments. It is generally used to pass structs and arrays by
1053 value, but is also valid on pointers to scalars. The copy is considered
1054 to belong to the caller not the callee (for example,
1055 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1056 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnerd78dbee2010-11-20 23:49:06 +00001057 values.</p>
1058
1059 <p>The byval attribute also supports specifying an alignment with
1060 the align attribute. It indicates the alignment of the stack slot to
1061 form and the known alignment of the pointer specified to the call site. If
1062 the alignment is not specified, then the code generator makes a
1063 target-specific assumption.</p></dd>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001064
Dan Gohman3770af52010-07-02 23:18:08 +00001065 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001066 <dd>This indicates that the pointer parameter specifies the address of a
1067 structure that is the return value of the function in the source program.
1068 This pointer must be guaranteed by the caller to be valid: loads and
1069 stores to the structure may be assumed by the callee to not to trap. This
1070 may only be applied to the first parameter. This is not a valid attribute
1071 for return values. </dd>
1072
Dan Gohman3770af52010-07-02 23:18:08 +00001073 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohmandf12d082010-07-02 18:41:32 +00001074 <dd>This indicates that pointer values
1075 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmande256292010-07-02 23:46:54 +00001076 value do not alias pointer values which are not <i>based</i> on it,
1077 ignoring certain "irrelevant" dependencies.
1078 For a call to the parent function, dependencies between memory
1079 references from before or after the call and from those during the call
1080 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1081 return value used in that call.
Dan Gohmandf12d082010-07-02 18:41:32 +00001082 The caller shares the responsibility with the callee for ensuring that
1083 these requirements are met.
1084 For further details, please see the discussion of the NoAlias response in
Dan Gohman6c858db2010-07-06 15:26:33 +00001085 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1086<br>
John McCall72ed8902010-07-06 21:07:14 +00001087 Note that this definition of <tt>noalias</tt> is intentionally
1088 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner5eff9ca2010-07-06 20:51:35 +00001089 arguments, though it is slightly weaker.
Dan Gohman6c858db2010-07-06 15:26:33 +00001090<br>
1091 For function return values, C99's <tt>restrict</tt> is not meaningful,
1092 while LLVM's <tt>noalias</tt> is.
1093 </dd>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001094
Dan Gohman3770af52010-07-02 23:18:08 +00001095 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001096 <dd>This indicates that the callee does not make any copies of the pointer
1097 that outlive the callee itself. This is not a valid attribute for return
1098 values.</dd>
1099
Dan Gohman3770af52010-07-02 23:18:08 +00001100 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001101 <dd>This indicates that the pointer parameter can be excised using the
1102 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1103 attribute for return values.</dd>
1104</dl>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001105
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001106</div>
1107
1108<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001109<h3>
Gordon Henriksen71183b62007-12-10 03:18:06 +00001110 <a name="gc">Garbage Collector Names</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001111</h3>
Gordon Henriksen71183b62007-12-10 03:18:06 +00001112
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001113<div>
Gordon Henriksen71183b62007-12-10 03:18:06 +00001114
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001115<p>Each function may specify a garbage collector name, which is simply a
1116 string:</p>
1117
Benjamin Kramer79698be2010-07-13 12:26:09 +00001118<pre class="doc_code">
Bill Wendling7f4a3362009-11-02 00:24:16 +00001119define void @f() gc "name" { ... }
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001120</pre>
Gordon Henriksen71183b62007-12-10 03:18:06 +00001121
1122<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001123 collector which will cause the compiler to alter its output in order to
1124 support the named garbage collection algorithm.</p>
1125
Gordon Henriksen71183b62007-12-10 03:18:06 +00001126</div>
1127
1128<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001129<h3>
Devang Patel9eb525d2008-09-26 23:51:19 +00001130 <a name="fnattrs">Function Attributes</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001131</h3>
Devang Patelcaacdba2008-09-04 23:05:13 +00001132
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001133<div>
Devang Patel9eb525d2008-09-26 23:51:19 +00001134
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001135<p>Function attributes are set to communicate additional information about a
1136 function. Function attributes are considered to be part of the function, not
1137 of the function type, so functions with different parameter attributes can
1138 have the same function type.</p>
Devang Patel9eb525d2008-09-26 23:51:19 +00001139
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001140<p>Function attributes are simple keywords that follow the type specified. If
1141 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelcaacdba2008-09-04 23:05:13 +00001142
Benjamin Kramer79698be2010-07-13 12:26:09 +00001143<pre class="doc_code">
Devang Patel9eb525d2008-09-26 23:51:19 +00001144define void @f() noinline { ... }
1145define void @f() alwaysinline { ... }
1146define void @f() alwaysinline optsize { ... }
Bill Wendling7f4a3362009-11-02 00:24:16 +00001147define void @f() optsize { ... }
Bill Wendlingb175fa42008-09-07 10:26:33 +00001148</pre>
Devang Patelcaacdba2008-09-04 23:05:13 +00001149
Bill Wendlingb175fa42008-09-07 10:26:33 +00001150<dl>
Kostya Serebryanya5054ad2012-01-20 17:56:17 +00001151 <dt><tt><b>address_safety</b></tt></dt>
1152 <dd>This attribute indicates that the address safety analysis
1153 is enabled for this function. </dd>
1154
Charles Davisbe5557e2010-02-12 00:31:15 +00001155 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1156 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1157 the backend should forcibly align the stack pointer. Specify the
1158 desired alignment, which must be a power of two, in parentheses.
1159
Bill Wendling7f4a3362009-11-02 00:24:16 +00001160 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001161 <dd>This attribute indicates that the inliner should attempt to inline this
1162 function into callers whenever possible, ignoring any active inlining size
1163 threshold for this caller.</dd>
Bill Wendlingb175fa42008-09-07 10:26:33 +00001164
Dan Gohman8bd11f12011-06-16 16:03:13 +00001165 <dt><tt><b>nonlazybind</b></tt></dt>
1166 <dd>This attribute suppresses lazy symbol binding for the function. This
1167 may make calls to the function faster, at the cost of extra program
1168 startup time if the function is not called during program startup.</dd>
1169
Jakob Stoklund Olesen74bb06c2010-02-06 01:16:28 +00001170 <dt><tt><b>inlinehint</b></tt></dt>
1171 <dd>This attribute indicates that the source code contained a hint that inlining
1172 this function is desirable (such as the "inline" keyword in C/C++). It
1173 is just a hint; it imposes no requirements on the inliner.</dd>
1174
Nick Lewycky14b58da2010-07-06 18:24:09 +00001175 <dt><tt><b>naked</b></tt></dt>
1176 <dd>This attribute disables prologue / epilogue emission for the function.
1177 This can have very system-specific consequences.</dd>
1178
1179 <dt><tt><b>noimplicitfloat</b></tt></dt>
1180 <dd>This attributes disables implicit floating point instructions.</dd>
1181
Bill Wendling7f4a3362009-11-02 00:24:16 +00001182 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001183 <dd>This attribute indicates that the inliner should never inline this
1184 function in any situation. This attribute may not be used together with
1185 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel9eb525d2008-09-26 23:51:19 +00001186
Nick Lewycky14b58da2010-07-06 18:24:09 +00001187 <dt><tt><b>noredzone</b></tt></dt>
1188 <dd>This attribute indicates that the code generator should not use a red
1189 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel9eb525d2008-09-26 23:51:19 +00001190
Bill Wendling7f4a3362009-11-02 00:24:16 +00001191 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001192 <dd>This function attribute indicates that the function never returns
1193 normally. This produces undefined behavior at runtime if the function
1194 ever does dynamically return.</dd>
Bill Wendlinga8130172008-11-13 01:02:51 +00001195
Bill Wendling7f4a3362009-11-02 00:24:16 +00001196 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001197 <dd>This function attribute indicates that the function never returns with an
1198 unwind or exceptional control flow. If the function does unwind, its
1199 runtime behavior is undefined.</dd>
Bill Wendling0f5541e2008-11-26 19:07:40 +00001200
Nick Lewycky14b58da2010-07-06 18:24:09 +00001201 <dt><tt><b>optsize</b></tt></dt>
1202 <dd>This attribute suggests that optimization passes and code generator passes
1203 make choices that keep the code size of this function low, and otherwise
1204 do optimizations specifically to reduce code size.</dd>
1205
Bill Wendling7f4a3362009-11-02 00:24:16 +00001206 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001207 <dd>This attribute indicates that the function computes its result (or decides
1208 to unwind an exception) based strictly on its arguments, without
1209 dereferencing any pointer arguments or otherwise accessing any mutable
1210 state (e.g. memory, control registers, etc) visible to caller functions.
1211 It does not write through any pointer arguments
1212 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1213 changes any state visible to callers. This means that it cannot unwind
Bill Wendling3f6a3a22012-02-06 21:57:33 +00001214 exceptions by calling the <tt>C++</tt> exception throwing methods.</dd>
Devang Patel310fd4a2009-06-12 19:45:19 +00001215
Bill Wendling7f4a3362009-11-02 00:24:16 +00001216 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001217 <dd>This attribute indicates that the function does not write through any
1218 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1219 arguments) or otherwise modify any state (e.g. memory, control registers,
1220 etc) visible to caller functions. It may dereference pointer arguments
1221 and read state that may be set in the caller. A readonly function always
1222 returns the same value (or unwinds an exception identically) when called
1223 with the same set of arguments and global state. It cannot unwind an
Bill Wendling3f6a3a22012-02-06 21:57:33 +00001224 exception by calling the <tt>C++</tt> exception throwing methods.</dd>
Anton Korobeynikovc8ce7b082009-07-17 18:07:26 +00001225
Bill Wendlingb437ab82011-12-05 21:27:54 +00001226 <dt><tt><b><a name="returns_twice">returns_twice</a></b></tt></dt>
1227 <dd>This attribute indicates that this function can return twice. The
1228 C <code>setjmp</code> is an example of such a function. The compiler
1229 disables some optimizations (like tail calls) in the caller of these
1230 functions.</dd>
1231
Bill Wendling7f4a3362009-11-02 00:24:16 +00001232 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001233 <dd>This attribute indicates that the function should emit a stack smashing
1234 protector. It is in the form of a "canary"&mdash;a random value placed on
1235 the stack before the local variables that's checked upon return from the
1236 function to see if it has been overwritten. A heuristic is used to
1237 determine if a function needs stack protectors or not.<br>
1238<br>
1239 If a function that has an <tt>ssp</tt> attribute is inlined into a
1240 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1241 function will have an <tt>ssp</tt> attribute.</dd>
1242
Bill Wendling7f4a3362009-11-02 00:24:16 +00001243 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001244 <dd>This attribute indicates that the function should <em>always</em> emit a
1245 stack smashing protector. This overrides
Bill Wendling30235112009-07-20 02:39:26 +00001246 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1247<br>
1248 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1249 function that doesn't have an <tt>sspreq</tt> attribute or which has
1250 an <tt>ssp</tt> attribute, then the resulting function will have
1251 an <tt>sspreq</tt> attribute.</dd>
Rafael Espindola163d6752011-07-25 15:27:59 +00001252
1253 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1254 <dd>This attribute indicates that the ABI being targeted requires that
1255 an unwind table entry be produce for this function even if we can
1256 show that no exceptions passes by it. This is normally the case for
1257 the ELF x86-64 abi, but it can be disabled for some compilation
1258 units.</dd>
Bill Wendlingb175fa42008-09-07 10:26:33 +00001259</dl>
1260
Devang Patelcaacdba2008-09-04 23:05:13 +00001261</div>
1262
1263<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001264<h3>
Chris Lattner93564892006-04-08 04:40:53 +00001265 <a name="moduleasm">Module-Level Inline Assembly</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001266</h3>
Chris Lattner91c15c42006-01-23 23:23:47 +00001267
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001268<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001269
1270<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1271 the GCC "file scope inline asm" blocks. These blocks are internally
1272 concatenated by LLVM and treated as a single unit, but may be separated in
1273 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner91c15c42006-01-23 23:23:47 +00001274
Benjamin Kramer79698be2010-07-13 12:26:09 +00001275<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +00001276module asm "inline asm code goes here"
1277module asm "more can go here"
1278</pre>
Chris Lattner91c15c42006-01-23 23:23:47 +00001279
1280<p>The strings can contain any character by escaping non-printable characters.
1281 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001282 for the number.</p>
Chris Lattner91c15c42006-01-23 23:23:47 +00001283
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001284<p>The inline asm code is simply printed to the machine code .s file when
1285 assembly code is generated.</p>
1286
Chris Lattner91c15c42006-01-23 23:23:47 +00001287</div>
Chris Lattner6af02f32004-12-09 16:11:40 +00001288
Reid Spencer50c723a2007-02-19 23:54:10 +00001289<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001290<h3>
Reid Spencer50c723a2007-02-19 23:54:10 +00001291 <a name="datalayout">Data Layout</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001292</h3>
Reid Spencer50c723a2007-02-19 23:54:10 +00001293
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001294<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001295
Reid Spencer50c723a2007-02-19 23:54:10 +00001296<p>A module may specify a target specific data layout string that specifies how
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001297 data is to be laid out in memory. The syntax for the data layout is
1298 simply:</p>
1299
Benjamin Kramer79698be2010-07-13 12:26:09 +00001300<pre class="doc_code">
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001301target datalayout = "<i>layout specification</i>"
1302</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001303
1304<p>The <i>layout specification</i> consists of a list of specifications
1305 separated by the minus sign character ('-'). Each specification starts with
1306 a letter and may include other information after the letter to define some
1307 aspect of the data layout. The specifications accepted are as follows:</p>
1308
Reid Spencer50c723a2007-02-19 23:54:10 +00001309<dl>
1310 <dt><tt>E</tt></dt>
1311 <dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001312 bits with the most significance have the lowest address location.</dd>
1313
Reid Spencer50c723a2007-02-19 23:54:10 +00001314 <dt><tt>e</tt></dt>
Chris Lattner67c37d12008-08-05 18:29:16 +00001315 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001316 the bits with the least significance have the lowest address
1317 location.</dd>
1318
Lang Hamesde7ab802011-10-10 23:42:08 +00001319 <dt><tt>S<i>size</i></tt></dt>
1320 <dd>Specifies the natural alignment of the stack in bits. Alignment promotion
1321 of stack variables is limited to the natural stack alignment to avoid
1322 dynamic stack realignment. The stack alignment must be a multiple of
Lang Hamesff2c52c2011-10-11 17:50:14 +00001323 8-bits. If omitted, the natural stack alignment defaults to "unspecified",
1324 which does not prevent any alignment promotions.</dd>
Lang Hamesde7ab802011-10-10 23:42:08 +00001325
Reid Spencer50c723a2007-02-19 23:54:10 +00001326 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher455c5772009-12-05 02:46:03 +00001327 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001328 <i>preferred</i> alignments. All sizes are in bits. Specifying
1329 the <i>pref</i> alignment is optional. If omitted, the
1330 preceding <tt>:</tt> should be omitted too.</dd>
1331
Reid Spencer50c723a2007-02-19 23:54:10 +00001332 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1333 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001334 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1335
Reid Spencer50c723a2007-02-19 23:54:10 +00001336 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher455c5772009-12-05 02:46:03 +00001337 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001338 <i>size</i>.</dd>
1339
Reid Spencer50c723a2007-02-19 23:54:10 +00001340 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher455c5772009-12-05 02:46:03 +00001341 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesence522852010-05-28 18:54:47 +00001342 <i>size</i>. Only values of <i>size</i> that are supported by the target
1343 will work. 32 (float) and 64 (double) are supported on all targets;
1344 80 or 128 (different flavors of long double) are also supported on some
1345 targets.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001346
Reid Spencer50c723a2007-02-19 23:54:10 +00001347 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1348 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001349 <i>size</i>.</dd>
1350
Daniel Dunbar7921a592009-06-08 22:17:53 +00001351 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1352 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001353 <i>size</i>.</dd>
Chris Lattnera381eff2009-11-07 09:35:34 +00001354
1355 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1356 <dd>This specifies a set of native integer widths for the target CPU
1357 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1358 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christopher455c5772009-12-05 02:46:03 +00001359 this set are considered to support most general arithmetic
Chris Lattnera381eff2009-11-07 09:35:34 +00001360 operations efficiently.</dd>
Reid Spencer50c723a2007-02-19 23:54:10 +00001361</dl>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001362
Reid Spencer50c723a2007-02-19 23:54:10 +00001363<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman61110ae2010-04-28 00:36:01 +00001364 default set of specifications which are then (possibly) overridden by the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001365 specifications in the <tt>datalayout</tt> keyword. The default specifications
1366 are given in this list:</p>
1367
Reid Spencer50c723a2007-02-19 23:54:10 +00001368<ul>
1369 <li><tt>E</tt> - big endian</li>
Dan Gohman8ad777d2010-02-23 02:44:03 +00001370 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
Reid Spencer50c723a2007-02-19 23:54:10 +00001371 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1372 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1373 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1374 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner67c37d12008-08-05 18:29:16 +00001375 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Reid Spencer50c723a2007-02-19 23:54:10 +00001376 alignment of 64-bits</li>
1377 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1378 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1379 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1380 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1381 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar7921a592009-06-08 22:17:53 +00001382 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencer50c723a2007-02-19 23:54:10 +00001383</ul>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001384
1385<p>When LLVM is determining the alignment for a given type, it uses the
1386 following rules:</p>
1387
Reid Spencer50c723a2007-02-19 23:54:10 +00001388<ol>
1389 <li>If the type sought is an exact match for one of the specifications, that
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001390 specification is used.</li>
1391
Reid Spencer50c723a2007-02-19 23:54:10 +00001392 <li>If no match is found, and the type sought is an integer type, then the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001393 smallest integer type that is larger than the bitwidth of the sought type
1394 is used. If none of the specifications are larger than the bitwidth then
1395 the the largest integer type is used. For example, given the default
1396 specifications above, the i7 type will use the alignment of i8 (next
1397 largest) while both i65 and i256 will use the alignment of i64 (largest
1398 specified).</li>
1399
Reid Spencer50c723a2007-02-19 23:54:10 +00001400 <li>If no match is found, and the type sought is a vector type, then the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001401 largest vector type that is smaller than the sought vector type will be
1402 used as a fall back. This happens because &lt;128 x double&gt; can be
1403 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencer50c723a2007-02-19 23:54:10 +00001404</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001405
Chris Lattner48797402011-10-11 23:01:39 +00001406<p>The function of the data layout string may not be what you expect. Notably,
1407 this is not a specification from the frontend of what alignment the code
1408 generator should use.</p>
1409
1410<p>Instead, if specified, the target data layout is required to match what the
1411 ultimate <em>code generator</em> expects. This string is used by the
1412 mid-level optimizers to
1413 improve code, and this only works if it matches what the ultimate code
1414 generator uses. If you would like to generate IR that does not embed this
1415 target-specific detail into the IR, then you don't have to specify the
1416 string. This will disable some optimizations that require precise layout
1417 information, but this also prevents those optimizations from introducing
1418 target specificity into the IR.</p>
1419
1420
1421
Reid Spencer50c723a2007-02-19 23:54:10 +00001422</div>
Chris Lattner6af02f32004-12-09 16:11:40 +00001423
Dan Gohman6154a012009-07-27 18:07:55 +00001424<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001425<h3>
Dan Gohman6154a012009-07-27 18:07:55 +00001426 <a name="pointeraliasing">Pointer Aliasing Rules</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001427</h3>
Dan Gohman6154a012009-07-27 18:07:55 +00001428
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001429<div>
Dan Gohman6154a012009-07-27 18:07:55 +00001430
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001431<p>Any memory access must be done through a pointer value associated
Andreas Bolkae39f0332009-07-27 20:37:10 +00001432with an address range of the memory access, otherwise the behavior
Dan Gohman6154a012009-07-27 18:07:55 +00001433is undefined. Pointer values are associated with address ranges
1434according to the following rules:</p>
1435
1436<ul>
Dan Gohmandf12d082010-07-02 18:41:32 +00001437 <li>A pointer value is associated with the addresses associated with
1438 any value it is <i>based</i> on.
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001439 <li>An address of a global variable is associated with the address
Dan Gohman6154a012009-07-27 18:07:55 +00001440 range of the variable's storage.</li>
1441 <li>The result value of an allocation instruction is associated with
1442 the address range of the allocated storage.</li>
1443 <li>A null pointer in the default address-space is associated with
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001444 no address.</li>
Dan Gohman6154a012009-07-27 18:07:55 +00001445 <li>An integer constant other than zero or a pointer value returned
1446 from a function not defined within LLVM may be associated with address
1447 ranges allocated through mechanisms other than those provided by
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001448 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman6154a012009-07-27 18:07:55 +00001449 allocated by mechanisms provided by LLVM.</li>
Dan Gohmandf12d082010-07-02 18:41:32 +00001450</ul>
1451
1452<p>A pointer value is <i>based</i> on another pointer value according
1453 to the following rules:</p>
1454
1455<ul>
1456 <li>A pointer value formed from a
1457 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1458 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1459 <li>The result value of a
1460 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1461 of the <tt>bitcast</tt>.</li>
1462 <li>A pointer value formed by an
1463 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1464 pointer values that contribute (directly or indirectly) to the
1465 computation of the pointer's value.</li>
1466 <li>The "<i>based</i> on" relationship is transitive.</li>
1467</ul>
1468
1469<p>Note that this definition of <i>"based"</i> is intentionally
1470 similar to the definition of <i>"based"</i> in C99, though it is
1471 slightly weaker.</p>
Dan Gohman6154a012009-07-27 18:07:55 +00001472
1473<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001474<tt><a href="#i_load">load</a></tt> merely indicates the size and
1475alignment of the memory from which to load, as well as the
Dan Gohman4eb47192010-06-17 19:23:50 +00001476interpretation of the value. The first operand type of a
Andreas Bolka8ae4e242009-07-29 00:02:05 +00001477<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1478and alignment of the store.</p>
Dan Gohman6154a012009-07-27 18:07:55 +00001479
1480<p>Consequently, type-based alias analysis, aka TBAA, aka
1481<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1482LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1483additional information which specialized optimization passes may use
1484to implement type-based alias analysis.</p>
1485
1486</div>
1487
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00001488<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001489<h3>
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00001490 <a name="volatile">Volatile Memory Accesses</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001491</h3>
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00001492
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001493<div>
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00001494
1495<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1496href="#i_store"><tt>store</tt></a>s, and <a
1497href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1498The optimizers must not change the number of volatile operations or change their
1499order of execution relative to other volatile operations. The optimizers
1500<i>may</i> change the order of volatile operations relative to non-volatile
1501operations. This is not Java's "volatile" and has no cross-thread
1502synchronization behavior.</p>
1503
1504</div>
1505
Eli Friedman35b54aa2011-07-20 21:35:53 +00001506<!-- ======================================================================= -->
1507<h3>
1508 <a name="memmodel">Memory Model for Concurrent Operations</a>
1509</h3>
1510
1511<div>
1512
1513<p>The LLVM IR does not define any way to start parallel threads of execution
1514or to register signal handlers. Nonetheless, there are platform-specific
1515ways to create them, and we define LLVM IR's behavior in their presence. This
1516model is inspired by the C++0x memory model.</p>
1517
Eli Friedman95f69a42011-08-22 21:35:27 +00001518<p>For a more informal introduction to this model, see the
1519<a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.
1520
Eli Friedman35b54aa2011-07-20 21:35:53 +00001521<p>We define a <i>happens-before</i> partial order as the least partial order
1522that</p>
1523<ul>
1524 <li>Is a superset of single-thread program order, and</li>
1525 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1526 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1527 by platform-specific techniques, like pthread locks, thread
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001528 creation, thread joining, etc., and by atomic instructions.
1529 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1530 </li>
Eli Friedman35b54aa2011-07-20 21:35:53 +00001531</ul>
1532
1533<p>Note that program order does not introduce <i>happens-before</i> edges
1534between a thread and signals executing inside that thread.</p>
1535
1536<p>Every (defined) read operation (load instructions, memcpy, atomic
1537loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1538(defined) write operations (store instructions, atomic
Eli Friedmanf12e4e92011-07-22 03:04:45 +00001539stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1540initialized globals are considered to have a write of the initializer which is
1541atomic and happens before any other read or write of the memory in question.
1542For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1543any write to the same byte, except:</p>
Eli Friedman35b54aa2011-07-20 21:35:53 +00001544
1545<ul>
1546 <li>If <var>write<sub>1</sub></var> happens before
1547 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1548 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
Eli Friedmanf12e4e92011-07-22 03:04:45 +00001549 does not see <var>write<sub>1</sub></var>.
Bill Wendling537603b2011-07-31 06:45:03 +00001550 <li>If <var>R<sub>byte</sub></var> happens before
1551 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1552 see <var>write<sub>3</sub></var>.
Eli Friedman35b54aa2011-07-20 21:35:53 +00001553</ul>
1554
1555<p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1556<ul>
Eli Friedman95f69a42011-08-22 21:35:27 +00001557 <li>If <var>R</var> is volatile, the result is target-dependent. (Volatile
1558 is supposed to give guarantees which can support
1559 <code>sig_atomic_t</code> in C/C++, and may be used for accesses to
1560 addresses which do not behave like normal memory. It does not generally
1561 provide cross-thread synchronization.)
1562 <li>Otherwise, if there is no write to the same byte that happens before
Eli Friedman35b54aa2011-07-20 21:35:53 +00001563 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1564 <tt>undef</tt> for that byte.
Eli Friedmanf12e4e92011-07-22 03:04:45 +00001565 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
Eli Friedman35b54aa2011-07-20 21:35:53 +00001566 <var>R<sub>byte</sub></var> returns the value written by that
1567 write.</li>
Eli Friedmanf12e4e92011-07-22 03:04:45 +00001568 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1569 <var>R<sub>byte</sub></var> may see are atomic, it chooses one of the
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001570 values written. See the <a href="#ordering">Atomic Memory Ordering
1571 Constraints</a> section for additional constraints on how the choice
1572 is made.
Eli Friedman35b54aa2011-07-20 21:35:53 +00001573 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1574</ul>
1575
1576<p><var>R</var> returns the value composed of the series of bytes it read.
1577This implies that some bytes within the value may be <tt>undef</tt>
1578<b>without</b> the entire value being <tt>undef</tt>. Note that this only
1579defines the semantics of the operation; it doesn't mean that targets will
1580emit more than one instruction to read the series of bytes.</p>
1581
1582<p>Note that in cases where none of the atomic intrinsics are used, this model
1583places only one restriction on IR transformations on top of what is required
1584for single-threaded execution: introducing a store to a byte which might not
Eli Friedman4bc9f3c2011-08-02 01:15:34 +00001585otherwise be stored is not allowed in general. (Specifically, in the case
1586where another thread might write to and read from an address, introducing a
1587store can change a load that may see exactly one write into a load that may
1588see multiple writes.)</p>
Eli Friedman35b54aa2011-07-20 21:35:53 +00001589
1590<!-- FIXME: This model assumes all targets where concurrency is relevant have
1591a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1592none of the backends currently in the tree fall into this category; however,
1593there might be targets which care. If there are, we want a paragraph
1594like the following:
1595
1596Targets may specify that stores narrower than a certain width are not
1597available; on such a target, for the purposes of this model, treat any
1598non-atomic write with an alignment or width less than the minimum width
1599as if it writes to the relevant surrounding bytes.
1600-->
1601
1602</div>
1603
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001604<!-- ======================================================================= -->
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00001605<h3>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001606 <a name="ordering">Atomic Memory Ordering Constraints</a>
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00001607</h3>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001608
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00001609<div>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001610
1611<p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
Eli Friedman59b66882011-08-09 23:02:53 +00001612<a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1613<a href="#i_fence"><code>fence</code></a>,
1614<a href="#i_load"><code>atomic load</code></a>, and
Eli Friedman75362532011-08-09 23:26:12 +00001615<a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001616that determines which other atomic instructions on the same address they
1617<i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1618but are somewhat more colloquial. If these descriptions aren't precise enough,
Eli Friedman95f69a42011-08-22 21:35:27 +00001619check those specs (see spec references in the
Nick Lewycky75499f52012-01-23 08:47:21 +00001620<a href="Atomics.html#introduction">atomics guide</a>).
Eli Friedman95f69a42011-08-22 21:35:27 +00001621<a href="#i_fence"><code>fence</code></a> instructions
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001622treat these orderings somewhat differently since they don't take an address.
1623See that instruction's documentation for details.</p>
1624
Eli Friedman95f69a42011-08-22 21:35:27 +00001625<p>For a simpler introduction to the ordering constraints, see the
1626<a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.</p>
1627
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001628<dl>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001629<dt><code>unordered</code></dt>
1630<dd>The set of values that can be read is governed by the happens-before
1631partial order. A value cannot be read unless some operation wrote it.
1632This is intended to provide a guarantee strong enough to model Java's
1633non-volatile shared variables. This ordering cannot be specified for
1634read-modify-write operations; it is not strong enough to make them atomic
1635in any interesting way.</dd>
1636<dt><code>monotonic</code></dt>
1637<dd>In addition to the guarantees of <code>unordered</code>, there is a single
1638total order for modifications by <code>monotonic</code> operations on each
1639address. All modification orders must be compatible with the happens-before
1640order. There is no guarantee that the modification orders can be combined to
1641a global total order for the whole program (and this often will not be
1642possible). The read in an atomic read-modify-write operation
1643(<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1644<a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1645reads the value in the modification order immediately before the value it
1646writes. If one atomic read happens before another atomic read of the same
1647address, the later read must see the same value or a later value in the
1648address's modification order. This disallows reordering of
1649<code>monotonic</code> (or stronger) operations on the same address. If an
1650address is written <code>monotonic</code>ally by one thread, and other threads
1651<code>monotonic</code>ally read that address repeatedly, the other threads must
Eli Friedman95f69a42011-08-22 21:35:27 +00001652eventually see the write. This corresponds to the C++0x/C1x
1653<code>memory_order_relaxed</code>.</dd>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001654<dt><code>acquire</code></dt>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001655<dd>In addition to the guarantees of <code>monotonic</code>,
Eli Friedman0cb3b562011-08-24 20:28:39 +00001656a <i>synchronizes-with</i> edge may be formed with a <code>release</code>
1657operation. This is intended to model C++'s <code>memory_order_acquire</code>.</dd>
1658<dt><code>release</code></dt>
1659<dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1660writes a value which is subsequently read by an <code>acquire</code> operation,
1661it <i>synchronizes-with</i> that operation. (This isn't a complete
1662description; see the C++0x definition of a release sequence.) This corresponds
1663to the C++0x/C1x <code>memory_order_release</code>.</dd>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001664<dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
Eli Friedman95f69a42011-08-22 21:35:27 +00001665<code>acquire</code> and <code>release</code> operation on its address.
1666This corresponds to the C++0x/C1x <code>memory_order_acq_rel</code>.</dd>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001667<dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1668<dd>In addition to the guarantees of <code>acq_rel</code>
1669(<code>acquire</code> for an operation which only reads, <code>release</code>
1670for an operation which only writes), there is a global total order on all
1671sequentially-consistent operations on all addresses, which is consistent with
1672the <i>happens-before</i> partial order and with the modification orders of
1673all the affected addresses. Each sequentially-consistent read sees the last
Eli Friedman95f69a42011-08-22 21:35:27 +00001674preceding write to the same address in this global order. This corresponds
1675to the C++0x/C1x <code>memory_order_seq_cst</code> and Java volatile.</dd>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00001676</dl>
1677
1678<p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1679it only <i>synchronizes with</i> or participates in modification and seq_cst
1680total orderings with other operations running in the same thread (for example,
1681in signal handlers).</p>
1682
1683</div>
1684
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001685</div>
1686
Chris Lattner2f7c9632001-06-06 20:29:01 +00001687<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001688<h2><a name="typesystem">Type System</a></h2>
Chris Lattner48b383b02003-11-25 01:02:51 +00001689<!-- *********************************************************************** -->
Chris Lattner6af02f32004-12-09 16:11:40 +00001690
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001691<div>
Chris Lattner6af02f32004-12-09 16:11:40 +00001692
Misha Brukman76307852003-11-08 01:05:38 +00001693<p>The LLVM type system is one of the most important features of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001694 intermediate representation. Being typed enables a number of optimizations
1695 to be performed on the intermediate representation directly, without having
1696 to do extra analyses on the side before the transformation. A strong type
1697 system makes it easier to read the generated code and enables novel analyses
1698 and transformations that are not feasible to perform on normal three address
1699 code representations.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +00001700
Chris Lattner2f7c9632001-06-06 20:29:01 +00001701<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001702<h3>
1703 <a name="t_classifications">Type Classifications</a>
1704</h3>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001705
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001706<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001707
1708<p>The types fall into a few useful classifications:</p>
Misha Brukmanc501f552004-03-01 17:47:27 +00001709
1710<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +00001711 <tbody>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001712 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner48b383b02003-11-25 01:02:51 +00001713 <tr>
Chris Lattner7824d182008-01-04 04:32:38 +00001714 <td><a href="#t_integer">integer</a></td>
Reid Spencer138249b2007-05-16 18:44:01 +00001715 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +00001716 </tr>
1717 <tr>
Chris Lattner7824d182008-01-04 04:32:38 +00001718 <td><a href="#t_floating">floating point</a></td>
Dan Gohman518cda42011-12-17 00:04:22 +00001719 <td><tt>half, float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +00001720 </tr>
1721 <tr>
1722 <td><a name="t_firstclass">first class</a></td>
Chris Lattner7824d182008-01-04 04:32:38 +00001723 <td><a href="#t_integer">integer</a>,
1724 <a href="#t_floating">floating point</a>,
1725 <a href="#t_pointer">pointer</a>,
Dan Gohman08783a882008-06-18 18:42:13 +00001726 <a href="#t_vector">vector</a>,
Dan Gohmanb9d66602008-05-12 23:51:09 +00001727 <a href="#t_struct">structure</a>,
1728 <a href="#t_array">array</a>,
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001729 <a href="#t_label">label</a>,
1730 <a href="#t_metadata">metadata</a>.
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00001731 </td>
Chris Lattner48b383b02003-11-25 01:02:51 +00001732 </tr>
Chris Lattner7824d182008-01-04 04:32:38 +00001733 <tr>
1734 <td><a href="#t_primitive">primitive</a></td>
1735 <td><a href="#t_label">label</a>,
1736 <a href="#t_void">void</a>,
Tobias Grosser4c8c95b2010-12-28 20:29:31 +00001737 <a href="#t_integer">integer</a>,
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001738 <a href="#t_floating">floating point</a>,
Dale Johannesen33e5c352010-10-01 00:48:59 +00001739 <a href="#t_x86mmx">x86mmx</a>,
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001740 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner7824d182008-01-04 04:32:38 +00001741 </tr>
1742 <tr>
1743 <td><a href="#t_derived">derived</a></td>
Chris Lattner392be582010-02-12 20:49:41 +00001744 <td><a href="#t_array">array</a>,
Chris Lattner7824d182008-01-04 04:32:38 +00001745 <a href="#t_function">function</a>,
1746 <a href="#t_pointer">pointer</a>,
1747 <a href="#t_struct">structure</a>,
Chris Lattner7824d182008-01-04 04:32:38 +00001748 <a href="#t_vector">vector</a>,
1749 <a href="#t_opaque">opaque</a>.
Dan Gohman93bf60d2008-10-14 16:32:04 +00001750 </td>
Chris Lattner7824d182008-01-04 04:32:38 +00001751 </tr>
Chris Lattner48b383b02003-11-25 01:02:51 +00001752 </tbody>
Misha Brukman76307852003-11-08 01:05:38 +00001753</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00001754
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001755<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1756 important. Values of these types are the only ones which can be produced by
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001757 instructions.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001758
Misha Brukman76307852003-11-08 01:05:38 +00001759</div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001760
Chris Lattner2f7c9632001-06-06 20:29:01 +00001761<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001762<h3>
1763 <a name="t_primitive">Primitive Types</a>
1764</h3>
Chris Lattner43542b32008-01-04 04:34:14 +00001765
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001766<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001767
Chris Lattner7824d182008-01-04 04:32:38 +00001768<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001769 system.</p>
Chris Lattner7824d182008-01-04 04:32:38 +00001770
1771<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001772<h4>
1773 <a name="t_integer">Integer Type</a>
1774</h4>
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001775
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001776<div>
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001777
1778<h5>Overview:</h5>
1779<p>The integer type is a very simple type that simply specifies an arbitrary
1780 bit width for the integer type desired. Any bit width from 1 bit to
1781 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1782
1783<h5>Syntax:</h5>
1784<pre>
1785 iN
1786</pre>
1787
1788<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1789 value.</p>
1790
1791<h5>Examples:</h5>
1792<table class="layout">
1793 <tr class="layout">
1794 <td class="left"><tt>i1</tt></td>
1795 <td class="left">a single-bit integer.</td>
1796 </tr>
1797 <tr class="layout">
1798 <td class="left"><tt>i32</tt></td>
1799 <td class="left">a 32-bit integer.</td>
1800 </tr>
1801 <tr class="layout">
1802 <td class="left"><tt>i1942652</tt></td>
1803 <td class="left">a really big integer of over 1 million bits.</td>
1804 </tr>
1805</table>
1806
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001807</div>
1808
1809<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001810<h4>
1811 <a name="t_floating">Floating Point Types</a>
1812</h4>
Chris Lattner7824d182008-01-04 04:32:38 +00001813
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001814<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001815
1816<table>
1817 <tbody>
1818 <tr><th>Type</th><th>Description</th></tr>
Dan Gohman518cda42011-12-17 00:04:22 +00001819 <tr><td><tt>half</tt></td><td>16-bit floating point value</td></tr>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001820 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1821 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1822 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1823 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1824 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1825 </tbody>
1826</table>
1827
Chris Lattner7824d182008-01-04 04:32:38 +00001828</div>
1829
1830<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001831<h4>
1832 <a name="t_x86mmx">X86mmx Type</a>
1833</h4>
Dale Johannesen33e5c352010-10-01 00:48:59 +00001834
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001835<div>
Dale Johannesen33e5c352010-10-01 00:48:59 +00001836
1837<h5>Overview:</h5>
1838<p>The x86mmx type represents a value held in an MMX register on an x86 machine. The operations allowed on it are quite limited: parameters and return values, load and store, and bitcast. User-specified MMX instructions are represented as intrinsic or asm calls with arguments and/or results of this type. There are no arrays, vectors or constants of this type.</p>
1839
1840<h5>Syntax:</h5>
1841<pre>
Dale Johannesenb1f0ff12010-10-01 01:07:02 +00001842 x86mmx
Dale Johannesen33e5c352010-10-01 00:48:59 +00001843</pre>
1844
1845</div>
1846
1847<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001848<h4>
1849 <a name="t_void">Void Type</a>
1850</h4>
Chris Lattner7824d182008-01-04 04:32:38 +00001851
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001852<div>
Bill Wendling30235112009-07-20 02:39:26 +00001853
Chris Lattner7824d182008-01-04 04:32:38 +00001854<h5>Overview:</h5>
1855<p>The void type does not represent any value and has no size.</p>
1856
1857<h5>Syntax:</h5>
Chris Lattner7824d182008-01-04 04:32:38 +00001858<pre>
1859 void
1860</pre>
Bill Wendling30235112009-07-20 02:39:26 +00001861
Chris Lattner7824d182008-01-04 04:32:38 +00001862</div>
1863
1864<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001865<h4>
1866 <a name="t_label">Label Type</a>
1867</h4>
Chris Lattner7824d182008-01-04 04:32:38 +00001868
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001869<div>
Bill Wendling30235112009-07-20 02:39:26 +00001870
Chris Lattner7824d182008-01-04 04:32:38 +00001871<h5>Overview:</h5>
1872<p>The label type represents code labels.</p>
1873
1874<h5>Syntax:</h5>
Chris Lattner7824d182008-01-04 04:32:38 +00001875<pre>
1876 label
1877</pre>
Bill Wendling30235112009-07-20 02:39:26 +00001878
Chris Lattner7824d182008-01-04 04:32:38 +00001879</div>
1880
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001881<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001882<h4>
1883 <a name="t_metadata">Metadata Type</a>
1884</h4>
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001885
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001886<div>
Bill Wendling30235112009-07-20 02:39:26 +00001887
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001888<h5>Overview:</h5>
Nick Lewycky93e06a52009-09-27 23:27:42 +00001889<p>The metadata type represents embedded metadata. No derived types may be
1890 created from metadata except for <a href="#t_function">function</a>
1891 arguments.
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001892
1893<h5>Syntax:</h5>
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001894<pre>
1895 metadata
1896</pre>
Bill Wendling30235112009-07-20 02:39:26 +00001897
Nick Lewyckyadbc2842009-05-30 05:06:04 +00001898</div>
1899
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001900</div>
Chris Lattner7824d182008-01-04 04:32:38 +00001901
1902<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001903<h3>
1904 <a name="t_derived">Derived Types</a>
1905</h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00001906
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001907<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001908
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001909<p>The real power in LLVM comes from the derived types in the system. This is
1910 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00001911 useful types. Each of these types contain one or more element types which
1912 may be a primitive type, or another derived type. For example, it is
1913 possible to have a two dimensional array, using an array as the element type
1914 of another array.</p>
Dan Gohman142ccc02009-01-24 15:58:40 +00001915
Chris Lattner392be582010-02-12 20:49:41 +00001916<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001917<h4>
1918 <a name="t_aggregate">Aggregate Types</a>
1919</h4>
Chris Lattner392be582010-02-12 20:49:41 +00001920
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001921<div>
Chris Lattner392be582010-02-12 20:49:41 +00001922
1923<p>Aggregate Types are a subset of derived types that can contain multiple
Duncan Sands9aaec152011-12-14 15:44:20 +00001924 member types. <a href="#t_array">Arrays</a> and
1925 <a href="#t_struct">structs</a> are aggregate types.
1926 <a href="#t_vector">Vectors</a> are not considered to be aggregate types.</p>
Chris Lattner392be582010-02-12 20:49:41 +00001927
1928</div>
1929
Reid Spencer138249b2007-05-16 18:44:01 +00001930<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001931<h4>
1932 <a name="t_array">Array Type</a>
1933</h4>
Chris Lattner74d3f822004-12-09 17:30:23 +00001934
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001935<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00001936
Chris Lattner2f7c9632001-06-06 20:29:01 +00001937<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00001938<p>The array type is a very simple derived type that arranges elements
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001939 sequentially in memory. The array type requires a size (number of elements)
1940 and an underlying data type.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001941
Chris Lattner590645f2002-04-14 06:13:44 +00001942<h5>Syntax:</h5>
Chris Lattner74d3f822004-12-09 17:30:23 +00001943<pre>
1944 [&lt;# elements&gt; x &lt;elementtype&gt;]
1945</pre>
1946
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001947<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1948 be any type with a size.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00001949
Chris Lattner590645f2002-04-14 06:13:44 +00001950<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001951<table class="layout">
1952 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00001953 <td class="left"><tt>[40 x i32]</tt></td>
1954 <td class="left">Array of 40 32-bit integer values.</td>
1955 </tr>
1956 <tr class="layout">
1957 <td class="left"><tt>[41 x i32]</tt></td>
1958 <td class="left">Array of 41 32-bit integer values.</td>
1959 </tr>
1960 <tr class="layout">
1961 <td class="left"><tt>[4 x i8]</tt></td>
1962 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001963 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +00001964</table>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001965<p>Here are some examples of multidimensional arrays:</p>
1966<table class="layout">
1967 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00001968 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1969 <td class="left">3x4 array of 32-bit integer values.</td>
1970 </tr>
1971 <tr class="layout">
1972 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1973 <td class="left">12x10 array of single precision floating point values.</td>
1974 </tr>
1975 <tr class="layout">
1976 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1977 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001978 </tr>
1979</table>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00001980
Dan Gohmanc74bc282009-11-09 19:01:53 +00001981<p>There is no restriction on indexing beyond the end of the array implied by
1982 a static type (though there are restrictions on indexing beyond the bounds
1983 of an allocated object in some cases). This means that single-dimension
1984 'variable sized array' addressing can be implemented in LLVM with a zero
1985 length array type. An implementation of 'pascal style arrays' in LLVM could
1986 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00001987
Misha Brukman76307852003-11-08 01:05:38 +00001988</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00001989
Chris Lattner2f7c9632001-06-06 20:29:01 +00001990<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00001991<h4>
1992 <a name="t_function">Function Type</a>
1993</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001994
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00001995<div>
Chris Lattnerda508ac2008-04-23 04:59:35 +00001996
Chris Lattner2f7c9632001-06-06 20:29:01 +00001997<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00001998<p>The function type can be thought of as a function signature. It consists of
1999 a return type and a list of formal parameter types. The return type of a
Chris Lattner13ee7952010-08-28 04:09:24 +00002000 function type is a first class type or a void type.</p>
Devang Pateld6cff512008-03-10 20:49:15 +00002001
Chris Lattner2f7c9632001-06-06 20:29:01 +00002002<h5>Syntax:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002003<pre>
Nick Lewycky14d1ccc2009-09-27 07:55:32 +00002004 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerda508ac2008-04-23 04:59:35 +00002005</pre>
2006
John Criswell4c0cf7f2005-10-24 16:17:18 +00002007<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002008 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
2009 which indicates that the function takes a variable number of arguments.
2010 Variable argument functions can access their arguments with
2011 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner47f2a832010-03-02 06:36:51 +00002012 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewycky93e06a52009-09-27 23:27:42 +00002013 <a href="#t_label">label</a>.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00002014
Chris Lattner2f7c9632001-06-06 20:29:01 +00002015<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002016<table class="layout">
2017 <tr class="layout">
Reid Spencer58c08712006-12-31 07:18:34 +00002018 <td class="left"><tt>i32 (i32)</tt></td>
2019 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002020 </td>
Reid Spencer58c08712006-12-31 07:18:34 +00002021 </tr><tr class="layout">
Chris Lattner47f2a832010-03-02 06:36:51 +00002022 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencer655dcc62006-12-31 07:20:23 +00002023 </tt></td>
Eric Christopher455c5772009-12-05 02:46:03 +00002024 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner47f2a832010-03-02 06:36:51 +00002025 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
2026 returning <tt>float</tt>.
Reid Spencer58c08712006-12-31 07:18:34 +00002027 </td>
2028 </tr><tr class="layout">
2029 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher455c5772009-12-05 02:46:03 +00002030 <td class="left">A vararg function that takes at least one
2031 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2032 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer58c08712006-12-31 07:18:34 +00002033 LLVM.
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002034 </td>
Devang Patele3dfc1c2008-03-24 05:35:41 +00002035 </tr><tr class="layout">
2036 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky14d1ccc2009-09-27 07:55:32 +00002037 <td class="left">A function taking an <tt>i32</tt>, returning a
2038 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patele3dfc1c2008-03-24 05:35:41 +00002039 </td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002040 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002041</table>
Misha Brukmanc501f552004-03-01 17:47:27 +00002042
Misha Brukman76307852003-11-08 01:05:38 +00002043</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002044
Chris Lattner2f7c9632001-06-06 20:29:01 +00002045<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002046<h4>
2047 <a name="t_struct">Structure Type</a>
2048</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002049
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002050<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002051
Chris Lattner2f7c9632001-06-06 20:29:01 +00002052<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002053<p>The structure type is used to represent a collection of data members together
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002054 in memory. The elements of a structure may be any type that has a size.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002055
Jeffrey Yasskinf991bbb2010-01-11 19:19:26 +00002056<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2057 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2058 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2059 Structures in registers are accessed using the
2060 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2061 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002062
2063<p>Structures may optionally be "packed" structures, which indicate that the
2064 alignment of the struct is one byte, and that there is no padding between
Chris Lattner190552d2011-08-12 17:31:02 +00002065 the elements. In non-packed structs, padding between field types is inserted
2066 as defined by the TargetData string in the module, which is required to match
Chris Lattner7bd0ea32011-10-11 23:02:17 +00002067 what the underlying code generator expects.</p>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002068
Chris Lattner190552d2011-08-12 17:31:02 +00002069<p>Structures can either be "literal" or "identified". A literal structure is
2070 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) whereas identified
2071 types are always defined at the top level with a name. Literal types are
2072 uniqued by their contents and can never be recursive or opaque since there is
Chris Lattner32531732011-08-12 18:12:40 +00002073 no way to write one. Identified types can be recursive, can be opaqued, and are
Chris Lattner190552d2011-08-12 17:31:02 +00002074 never uniqued.
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002075</p>
2076
Chris Lattner2f7c9632001-06-06 20:29:01 +00002077<h5>Syntax:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00002078<pre>
Chris Lattner190552d2011-08-12 17:31:02 +00002079 %T1 = type { &lt;type list&gt; } <i>; Identified normal struct type</i>
2080 %T2 = type &lt;{ &lt;type list&gt; }&gt; <i>; Identified packed struct type</i>
Bill Wendling30235112009-07-20 02:39:26 +00002081</pre>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002082
Chris Lattner2f7c9632001-06-06 20:29:01 +00002083<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002084<table class="layout">
2085 <tr class="layout">
Jeff Cohen5819f182007-04-22 01:17:39 +00002086 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2087 <td class="left">A triple of three <tt>i32</tt> values</td>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002088 </tr>
2089 <tr class="layout">
Jeff Cohen5819f182007-04-22 01:17:39 +00002090 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
2091 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2092 second element is a <a href="#t_pointer">pointer</a> to a
2093 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2094 an <tt>i32</tt>.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002095 </tr>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002096 <tr class="layout">
2097 <td class="left"><tt>&lt;{ i8, i32 }&gt;</tt></td>
2098 <td class="left">A packed struct known to be 5 bytes in size.</td>
2099 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +00002100</table>
Dan Gohman142ccc02009-01-24 15:58:40 +00002101
Misha Brukman76307852003-11-08 01:05:38 +00002102</div>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002103
Chris Lattner2f7c9632001-06-06 20:29:01 +00002104<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002105<h4>
Chris Lattner2a843822011-07-23 19:59:08 +00002106 <a name="t_opaque">Opaque Structure Types</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002107</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002108
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002109<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002110
Andrew Lenharth8df88e22006-12-08 17:13:00 +00002111<h5>Overview:</h5>
Chris Lattner2a843822011-07-23 19:59:08 +00002112<p>Opaque structure types are used to represent named structure types that do
2113 not have a body specified. This corresponds (for example) to the C notion of
2114 a forward declared structure.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002115
Andrew Lenharth8df88e22006-12-08 17:13:00 +00002116<h5>Syntax:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00002117<pre>
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002118 %X = type opaque
2119 %52 = type opaque
Bill Wendling30235112009-07-20 02:39:26 +00002120</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002121
Andrew Lenharth8df88e22006-12-08 17:13:00 +00002122<h5>Examples:</h5>
2123<table class="layout">
2124 <tr class="layout">
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002125 <td class="left"><tt>opaque</tt></td>
2126 <td class="left">An opaque type.</td>
Andrew Lenharth8df88e22006-12-08 17:13:00 +00002127 </tr>
2128</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002129
Andrew Lenharth8df88e22006-12-08 17:13:00 +00002130</div>
2131
Chris Lattnerb1ed91f2011-07-09 17:41:24 +00002132
2133
Andrew Lenharth8df88e22006-12-08 17:13:00 +00002134<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002135<h4>
2136 <a name="t_pointer">Pointer Type</a>
2137</h4>
Chris Lattner4a67c912009-02-08 19:53:29 +00002138
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002139<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002140
2141<h5>Overview:</h5>
Dan Gohman88481112010-02-25 16:50:07 +00002142<p>The pointer type is used to specify memory locations.
2143 Pointers are commonly used to reference objects in memory.</p>
2144
2145<p>Pointer types may have an optional address space attribute defining the
2146 numbered address space where the pointed-to object resides. The default
2147 address space is number zero. The semantics of non-zero address
2148 spaces are target-specific.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002149
2150<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2151 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner4a67c912009-02-08 19:53:29 +00002152
Chris Lattner590645f2002-04-14 06:13:44 +00002153<h5>Syntax:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00002154<pre>
2155 &lt;type&gt; *
2156</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002157
Chris Lattner590645f2002-04-14 06:13:44 +00002158<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002159<table class="layout">
2160 <tr class="layout">
Dan Gohman623806e2009-01-04 23:44:43 +00002161 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner747359f2007-12-19 05:04:11 +00002162 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2163 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2164 </tr>
2165 <tr class="layout">
Dan Gohmanaabfdb32010-05-28 17:13:49 +00002166 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner747359f2007-12-19 05:04:11 +00002167 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00002168 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner747359f2007-12-19 05:04:11 +00002169 <tt>i32</tt>.</td>
2170 </tr>
2171 <tr class="layout">
2172 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2173 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2174 that resides in address space #5.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002175 </tr>
Misha Brukman76307852003-11-08 01:05:38 +00002176</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002177
Misha Brukman76307852003-11-08 01:05:38 +00002178</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002179
Chris Lattnerc8cb6952004-08-12 19:12:28 +00002180<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002181<h4>
2182 <a name="t_vector">Vector Type</a>
2183</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002184
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002185<div>
Chris Lattner37b6b092005-04-25 17:34:15 +00002186
Chris Lattnerc8cb6952004-08-12 19:12:28 +00002187<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002188<p>A vector type is a simple derived type that represents a vector of elements.
2189 Vector types are used when multiple primitive data are operated in parallel
2190 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sands31c0e0e2009-11-27 13:38:03 +00002191 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002192 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +00002193
Chris Lattnerc8cb6952004-08-12 19:12:28 +00002194<h5>Syntax:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +00002195<pre>
2196 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
2197</pre>
2198
Chris Lattnerf11031a2010-10-10 18:20:35 +00002199<p>The number of elements is a constant integer value larger than 0; elementtype
Nadav Rotem3924cb02011-12-05 06:29:09 +00002200 may be any integer or floating point type, or a pointer to these types.
2201 Vectors of size zero are not allowed. </p>
Chris Lattner37b6b092005-04-25 17:34:15 +00002202
Chris Lattnerc8cb6952004-08-12 19:12:28 +00002203<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002204<table class="layout">
2205 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +00002206 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
2207 <td class="left">Vector of 4 32-bit integer values.</td>
2208 </tr>
2209 <tr class="layout">
2210 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
2211 <td class="left">Vector of 8 32-bit floating-point values.</td>
2212 </tr>
2213 <tr class="layout">
2214 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
2215 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002216 </tr>
Nadav Rotem3924cb02011-12-05 06:29:09 +00002217 <tr class="layout">
2218 <td class="left"><tt>&lt;4 x i64*&gt;</tt></td>
2219 <td class="left">Vector of 4 pointers to 64-bit integer values.</td>
2220 </tr>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +00002221</table>
Dan Gohman142ccc02009-01-24 15:58:40 +00002222
Misha Brukman76307852003-11-08 01:05:38 +00002223</div>
2224
Bill Wendlingae8b5ea2011-07-31 06:47:33 +00002225</div>
2226
NAKAMURA Takumia35cdd62011-10-31 13:04:26 +00002227</div>
2228
Chris Lattner74d3f822004-12-09 17:30:23 +00002229<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002230<h2><a name="constants">Constants</a></h2>
Chris Lattner74d3f822004-12-09 17:30:23 +00002231<!-- *********************************************************************** -->
2232
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002233<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002234
2235<p>LLVM has several different basic types of constants. This section describes
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002236 them all and their syntax.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002237
Chris Lattner74d3f822004-12-09 17:30:23 +00002238<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002239<h3>
2240 <a name="simpleconstants">Simple Constants</a>
2241</h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00002242
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002243<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002244
2245<dl>
2246 <dt><b>Boolean constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002247 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00002248 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002249
2250 <dt><b>Integer constants</b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002251 <dd>Standard integers (such as '4') are constants of
2252 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2253 with integer types.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002254
2255 <dt><b>Floating point constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002256 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002257 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2258 notation (see below). The assembler requires the exact decimal value of a
2259 floating-point constant. For example, the assembler accepts 1.25 but
2260 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2261 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002262
2263 <dt><b>Null pointer constants</b></dt>
John Criswelldfe6a862004-12-10 15:51:16 +00002264 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002265 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002266</dl>
2267
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002268<p>The one non-intuitive notation for constants is the hexadecimal form of
2269 floating point constants. For example, the form '<tt>double
2270 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2271 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2272 constants are required (and the only time that they are generated by the
2273 disassembler) is when a floating point constant must be emitted but it cannot
2274 be represented as a decimal floating point number in a reasonable number of
2275 digits. For example, NaN's, infinities, and other special values are
2276 represented in their IEEE hexadecimal format so that assembly and disassembly
2277 do not cause any bits to change in the constants.</p>
2278
Dan Gohman518cda42011-12-17 00:04:22 +00002279<p>When using the hexadecimal form, constants of types half, float, and double are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002280 represented using the 16-digit form shown above (which matches the IEEE754
Dan Gohman518cda42011-12-17 00:04:22 +00002281 representation for double); half and float values must, however, be exactly
2282 representable as IEE754 half and single precision, respectively.
2283 Hexadecimal format is always used
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002284 for long double, and there are three forms of long double. The 80-bit format
2285 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2286 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2287 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2288 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2289 currently supported target uses this format. Long doubles will only work if
2290 they match the long double format on your target. All hexadecimal formats
2291 are big-endian (sign bit at the left).</p>
2292
Dale Johannesen33e5c352010-10-01 00:48:59 +00002293<p>There are no constants of type x86mmx.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002294</div>
2295
2296<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002297<h3>
Bill Wendling972b7202009-07-20 02:32:41 +00002298<a name="aggregateconstants"></a> <!-- old anchor -->
2299<a name="complexconstants">Complex Constants</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002300</h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00002301
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002302<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002303
Chris Lattner361bfcd2009-02-28 18:32:25 +00002304<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002305 constants and smaller complex constants.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002306
2307<dl>
2308 <dt><b>Structure constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002309 <dd>Structure constants are represented with notation similar to structure
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002310 type definitions (a comma separated list of elements, surrounded by braces
2311 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2312 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2313 Structure constants must have <a href="#t_struct">structure type</a>, and
2314 the number and types of elements must match those specified by the
2315 type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002316
2317 <dt><b>Array constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002318 <dd>Array constants are represented with notation similar to array type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002319 definitions (a comma separated list of elements, surrounded by square
2320 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2321 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2322 the number and types of elements must match those specified by the
2323 type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002324
Reid Spencer404a3252007-02-15 03:07:05 +00002325 <dt><b>Vector constants</b></dt>
Reid Spencer404a3252007-02-15 03:07:05 +00002326 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002327 definitions (a comma separated list of elements, surrounded by
2328 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2329 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2330 have <a href="#t_vector">vector type</a>, and the number and types of
2331 elements must match those specified by the type.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002332
2333 <dt><b>Zero initialization</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002334 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattner392be582010-02-12 20:49:41 +00002335 value to zero of <em>any</em> type, including scalar and
2336 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002337 This is often used to avoid having to print large zero initializers
2338 (e.g. for large arrays) and is always exactly equivalent to using explicit
2339 zero initializers.</dd>
Nick Lewycky49f89192009-04-04 07:22:01 +00002340
2341 <dt><b>Metadata node</b></dt>
Nick Lewycky8e2c4f42009-05-30 16:08:30 +00002342 <dd>A metadata node is a structure-like constant with
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002343 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2344 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2345 be interpreted as part of the instruction stream, metadata is a place to
2346 attach additional information such as debug info.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002347</dl>
2348
2349</div>
2350
2351<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002352<h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00002353 <a name="globalconstants">Global Variable and Function Addresses</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002354</h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00002355
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002356<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002357
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002358<p>The addresses of <a href="#globalvars">global variables</a>
2359 and <a href="#functionstructure">functions</a> are always implicitly valid
2360 (link-time) constants. These constants are explicitly referenced when
2361 the <a href="#identifiers">identifier for the global</a> is used and always
2362 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2363 legal LLVM file:</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002364
Benjamin Kramer79698be2010-07-13 12:26:09 +00002365<pre class="doc_code">
Chris Lattner00538a12007-06-06 18:28:13 +00002366@X = global i32 17
2367@Y = global i32 42
2368@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattner74d3f822004-12-09 17:30:23 +00002369</pre>
2370
2371</div>
2372
2373<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002374<h3>
2375 <a name="undefvalues">Undefined Values</a>
2376</h3>
2377
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002378<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002379
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002380<p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
Benjamin Kramer0f420382009-10-12 14:46:08 +00002381 indicates that the user of the value may receive an unspecified bit-pattern.
Bill Wendling6bbe0912010-10-27 01:07:41 +00002382 Undefined values may be of any type (other than '<tt>label</tt>'
2383 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002384
Chris Lattner92ada5d2009-09-11 01:49:31 +00002385<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002386 program is well defined no matter what value is used. This gives the
2387 compiler more freedom to optimize. Here are some examples of (potentially
2388 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002389
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002390
Benjamin Kramer79698be2010-07-13 12:26:09 +00002391<pre class="doc_code">
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002392 %A = add %X, undef
2393 %B = sub %X, undef
2394 %C = xor %X, undef
2395Safe:
2396 %A = undef
2397 %B = undef
2398 %C = undef
2399</pre>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002400
2401<p>This is safe because all of the output bits are affected by the undef bits.
Bill Wendling6bbe0912010-10-27 01:07:41 +00002402 Any output bit can have a zero or one depending on the input bits.</p>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002403
Benjamin Kramer79698be2010-07-13 12:26:09 +00002404<pre class="doc_code">
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002405 %A = or %X, undef
2406 %B = and %X, undef
2407Safe:
2408 %A = -1
2409 %B = 0
2410Unsafe:
2411 %A = undef
2412 %B = undef
2413</pre>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002414
2415<p>These logical operations have bits that are not always affected by the input.
Bill Wendling6bbe0912010-10-27 01:07:41 +00002416 For example, if <tt>%X</tt> has a zero bit, then the output of the
2417 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2418 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2419 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2420 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2421 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2422 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2423 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002424
Benjamin Kramer79698be2010-07-13 12:26:09 +00002425<pre class="doc_code">
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002426 %A = select undef, %X, %Y
2427 %B = select undef, 42, %Y
2428 %C = select %X, %Y, undef
2429Safe:
2430 %A = %X (or %Y)
2431 %B = 42 (or %Y)
2432 %C = %Y
2433Unsafe:
2434 %A = undef
2435 %B = undef
2436 %C = undef
2437</pre>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002438
Bill Wendling6bbe0912010-10-27 01:07:41 +00002439<p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2440 branch) conditions can go <em>either way</em>, but they have to come from one
2441 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2442 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2443 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2444 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2445 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2446 eliminated.</p>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002447
Benjamin Kramer79698be2010-07-13 12:26:09 +00002448<pre class="doc_code">
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002449 %A = xor undef, undef
Eric Christopher455c5772009-12-05 02:46:03 +00002450
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002451 %B = undef
2452 %C = xor %B, %B
2453
2454 %D = undef
2455 %E = icmp lt %D, 4
2456 %F = icmp gte %D, 4
2457
2458Safe:
2459 %A = undef
2460 %B = undef
2461 %C = undef
2462 %D = undef
2463 %E = undef
2464 %F = undef
2465</pre>
Chris Lattnerec72b9b2009-09-07 22:52:39 +00002466
Bill Wendling6bbe0912010-10-27 01:07:41 +00002467<p>This example points out that two '<tt>undef</tt>' operands are not
2468 necessarily the same. This can be surprising to people (and also matches C
2469 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2470 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2471 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2472 its value over its "live range". This is true because the variable doesn't
2473 actually <em>have a live range</em>. Instead, the value is logically read
2474 from arbitrary registers that happen to be around when needed, so the value
2475 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2476 need to have the same semantics or the core LLVM "replace all uses with"
2477 concept would not hold.</p>
Chris Lattnera34a7182009-09-07 23:33:52 +00002478
Benjamin Kramer79698be2010-07-13 12:26:09 +00002479<pre class="doc_code">
Chris Lattnera34a7182009-09-07 23:33:52 +00002480 %A = fdiv undef, %X
2481 %B = fdiv %X, undef
2482Safe:
2483 %A = undef
2484b: unreachable
2485</pre>
Chris Lattnera34a7182009-09-07 23:33:52 +00002486
2487<p>These examples show the crucial difference between an <em>undefined
Bill Wendling6bbe0912010-10-27 01:07:41 +00002488 value</em> and <em>undefined behavior</em>. An undefined value (like
2489 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2490 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2491 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2492 defined on SNaN's. However, in the second example, we can make a more
2493 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2494 arbitrary value, we are allowed to assume that it could be zero. Since a
2495 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2496 the operation does not execute at all. This allows us to delete the divide and
2497 all code after it. Because the undefined operation "can't happen", the
2498 optimizer can assume that it occurs in dead code.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00002499
Benjamin Kramer79698be2010-07-13 12:26:09 +00002500<pre class="doc_code">
Chris Lattnera34a7182009-09-07 23:33:52 +00002501a: store undef -> %X
2502b: store %X -> undef
2503Safe:
2504a: &lt;deleted&gt;
2505b: unreachable
2506</pre>
Chris Lattnera34a7182009-09-07 23:33:52 +00002507
Bill Wendling6bbe0912010-10-27 01:07:41 +00002508<p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2509 undefined value can be assumed to not have any effect; we can assume that the
2510 value is overwritten with bits that happen to match what was already there.
2511 However, a store <em>to</em> an undefined location could clobber arbitrary
2512 memory, therefore, it has undefined behavior.</p>
Chris Lattnera34a7182009-09-07 23:33:52 +00002513
Chris Lattner74d3f822004-12-09 17:30:23 +00002514</div>
2515
2516<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002517<h3>
Dan Gohman9a2a0932011-12-06 03:18:47 +00002518 <a name="poisonvalues">Poison Values</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002519</h3>
2520
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002521<div>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002522
Dan Gohman9a2a0932011-12-06 03:18:47 +00002523<p>Poison values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohman32772f72011-12-06 03:35:58 +00002524 they also represent the fact that an instruction or constant expression which
2525 cannot evoke side effects has nevertheless detected a condition which results
2526 in undefined behavior.</p>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002527
Dan Gohman9a2a0932011-12-06 03:18:47 +00002528<p>There is currently no way of representing a poison value in the IR; they
Dan Gohmanac355aa2010-05-03 14:51:43 +00002529 only exist when produced by operations such as
Dan Gohman2f1ae062010-04-28 00:49:41 +00002530 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002531
Dan Gohman9a2a0932011-12-06 03:18:47 +00002532<p>Poison value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002533
Dan Gohman2f1ae062010-04-28 00:49:41 +00002534<ul>
2535<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2536 their operands.</li>
2537
2538<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2539 to their dynamic predecessor basic block.</li>
2540
2541<li>Function arguments depend on the corresponding actual argument values in
2542 the dynamic callers of their functions.</li>
2543
2544<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2545 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2546 control back to them.</li>
2547
Dan Gohman7292a752010-05-03 14:55:22 +00002548<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
Bill Wendling3f6a3a22012-02-06 21:57:33 +00002549 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_resume"><tt>resume</tt></a>,
Dan Gohman7292a752010-05-03 14:55:22 +00002550 or exception-throwing call instructions that dynamically transfer control
2551 back to them.</li>
2552
Dan Gohman2f1ae062010-04-28 00:49:41 +00002553<li>Non-volatile loads and stores depend on the most recent stores to all of the
2554 referenced memory addresses, following the order in the IR
2555 (including loads and stores implied by intrinsics such as
2556 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2557
Dan Gohman3513ea52010-05-03 14:59:34 +00002558<!-- TODO: In the case of multiple threads, this only applies if the store
2559 "happens-before" the load or store. -->
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002560
Dan Gohman2f1ae062010-04-28 00:49:41 +00002561<!-- TODO: floating-point exception state -->
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002562
Dan Gohman2f1ae062010-04-28 00:49:41 +00002563<li>An instruction with externally visible side effects depends on the most
2564 recent preceding instruction with externally visible side effects, following
Dan Gohman6c858db2010-07-06 15:26:33 +00002565 the order in the IR. (This includes
2566 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002567
Dan Gohman7292a752010-05-03 14:55:22 +00002568<li>An instruction <i>control-depends</i> on a
2569 <a href="#terminators">terminator instruction</a>
2570 if the terminator instruction has multiple successors and the instruction
2571 is always executed when control transfers to one of the successors, and
Chris Lattner0ab5e2c2011-04-15 05:18:47 +00002572 may not be executed when control is transferred to another.</li>
Dan Gohman2f1ae062010-04-28 00:49:41 +00002573
Dan Gohmanc8454ee2011-04-12 23:05:59 +00002574<li>Additionally, an instruction also <i>control-depends</i> on a terminator
2575 instruction if the set of instructions it otherwise depends on would be
Chris Lattner0ab5e2c2011-04-15 05:18:47 +00002576 different if the terminator had transferred control to a different
Dan Gohmanc8454ee2011-04-12 23:05:59 +00002577 successor.</li>
2578
Dan Gohman2f1ae062010-04-28 00:49:41 +00002579<li>Dependence is transitive.</li>
2580
2581</ul>
Dan Gohman2f1ae062010-04-28 00:49:41 +00002582
Dan Gohman32772f72011-12-06 03:35:58 +00002583<p>Poison Values have the same behavior as <a href="#undefvalues">undef values</a>,
2584 with the additional affect that any instruction which has a <i>dependence</i>
2585 on a poison value has undefined behavior.</p>
Dan Gohman2f1ae062010-04-28 00:49:41 +00002586
2587<p>Here are some examples:</p>
Dan Gohman48a25882010-04-26 20:54:53 +00002588
Benjamin Kramer79698be2010-07-13 12:26:09 +00002589<pre class="doc_code">
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002590entry:
Dan Gohman9a2a0932011-12-06 03:18:47 +00002591 %poison = sub nuw i32 0, 1 ; Results in a poison value.
Dan Gohman32772f72011-12-06 03:35:58 +00002592 %still_poison = and i32 %poison, 0 ; 0, but also poison.
Dan Gohman9a2a0932011-12-06 03:18:47 +00002593 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
Dan Gohman32772f72011-12-06 03:35:58 +00002594 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
Dan Gohman2f1ae062010-04-28 00:49:41 +00002595
Dan Gohman32772f72011-12-06 03:35:58 +00002596 store i32 %poison, i32* @g ; Poison value stored to memory.
2597 %poison2 = load i32* @g ; Poison value loaded back from memory.
Dan Gohman2f1ae062010-04-28 00:49:41 +00002598
Dan Gohman9a2a0932011-12-06 03:18:47 +00002599 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
Dan Gohman2f1ae062010-04-28 00:49:41 +00002600
2601 %narrowaddr = bitcast i32* @g to i16*
2602 %wideaddr = bitcast i32* @g to i64*
Dan Gohman9a2a0932011-12-06 03:18:47 +00002603 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2604 %poison4 = load i64* %wideaddr ; Returns a poison value.
Dan Gohman2f1ae062010-04-28 00:49:41 +00002605
Dan Gohman5f115a72011-12-06 03:31:14 +00002606 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2607 br i1 %cmp, label %true, label %end ; Branch to either destination.
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002608
2609true:
Dan Gohman5f115a72011-12-06 03:31:14 +00002610 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2611 ; it has undefined behavior.
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002612 br label %end
2613
2614end:
2615 %p = phi i32 [ 0, %entry ], [ 1, %true ]
Dan Gohman5f115a72011-12-06 03:31:14 +00002616 ; Both edges into this PHI are
2617 ; control-dependent on %cmp, so this
2618 ; always results in a poison value.
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002619
Dan Gohman5f115a72011-12-06 03:31:14 +00002620 store volatile i32 0, i32* @g ; This would depend on the store in %true
2621 ; if %cmp is true, or the store in %entry
2622 ; otherwise, so this is undefined behavior.
Dan Gohmanc8454ee2011-04-12 23:05:59 +00002623
Nick Lewycky9c876bf2011-05-16 19:29:30 +00002624 br i1 %cmp, label %second_true, label %second_end
Dan Gohman5f115a72011-12-06 03:31:14 +00002625 ; The same branch again, but this time the
2626 ; true block doesn't have side effects.
Dan Gohmanc8454ee2011-04-12 23:05:59 +00002627
2628second_true:
2629 ; No side effects!
Nick Lewycky9c876bf2011-05-16 19:29:30 +00002630 ret void
Dan Gohmanc8454ee2011-04-12 23:05:59 +00002631
2632second_end:
Dan Gohman5f115a72011-12-06 03:31:14 +00002633 store volatile i32 0, i32* @g ; This time, the instruction always depends
2634 ; on the store in %end. Also, it is
2635 ; control-equivalent to %end, so this is
Dan Gohman32772f72011-12-06 03:35:58 +00002636 ; well-defined (ignoring earlier undefined
2637 ; behavior in this example).
Dan Gohmanb8b85c12010-04-26 23:36:52 +00002638</pre>
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002639
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00002640</div>
2641
2642<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002643<h3>
2644 <a name="blockaddress">Addresses of Basic Blocks</a>
2645</h3>
2646
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002647<div>
Chris Lattnere4801f72009-10-27 21:01:34 +00002648
Chris Lattneraa99c942009-11-01 01:27:45 +00002649<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnere4801f72009-10-27 21:01:34 +00002650
2651<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner5c5f0ac2009-10-27 21:49:40 +00002652 basic block in the specified function, and always has an i8* type. Taking
Chris Lattneraa99c942009-11-01 01:27:45 +00002653 the address of the entry block is illegal.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00002654
Chris Lattnere4801f72009-10-27 21:01:34 +00002655<p>This value only has defined behavior when used as an operand to the
Bill Wendling6bbe0912010-10-27 01:07:41 +00002656 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2657 comparisons against null. Pointer equality tests between labels addresses
2658 results in undefined behavior &mdash; though, again, comparison against null
2659 is ok, and no label is equal to the null pointer. This may be passed around
2660 as an opaque pointer sized value as long as the bits are not inspected. This
2661 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2662 long as the original value is reconstituted before the <tt>indirectbr</tt>
2663 instruction.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00002664
Bill Wendling6bbe0912010-10-27 01:07:41 +00002665<p>Finally, some targets may provide defined semantics when using the value as
2666 the operand to an inline assembly, but that is target specific.</p>
Chris Lattnere4801f72009-10-27 21:01:34 +00002667
2668</div>
2669
2670
2671<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002672<h3>
2673 <a name="constantexprs">Constant Expressions</a>
2674</h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00002675
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002676<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00002677
2678<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002679 to be used as constants. Constant expressions may be of
2680 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2681 operation that does not have side effects (e.g. load and call are not
Bill Wendling6bbe0912010-10-27 01:07:41 +00002682 supported). The following is the syntax for constant expressions:</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00002683
2684<dl>
Dan Gohmand6a6f612010-05-28 17:07:41 +00002685 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002686 <dd>Truncate a constant to another type. The bit size of CST must be larger
2687 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002688
Dan Gohmand6a6f612010-05-28 17:07:41 +00002689 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002690 <dd>Zero extend a constant to another type. The bit size of CST must be
Duncan Sandsa522e562010-07-13 12:06:14 +00002691 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002692
Dan Gohmand6a6f612010-05-28 17:07:41 +00002693 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002694 <dd>Sign extend a constant to another type. The bit size of CST must be
Duncan Sandsa522e562010-07-13 12:06:14 +00002695 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002696
Dan Gohmand6a6f612010-05-28 17:07:41 +00002697 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002698 <dd>Truncate a floating point constant to another floating point type. The
2699 size of CST must be larger than the size of TYPE. Both types must be
2700 floating point.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002701
Dan Gohmand6a6f612010-05-28 17:07:41 +00002702 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002703 <dd>Floating point extend a constant to another type. The size of CST must be
2704 smaller or equal to the size of TYPE. Both types must be floating
2705 point.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002706
Dan Gohmand6a6f612010-05-28 17:07:41 +00002707 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002708 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002709 constant. TYPE must be a scalar or vector integer type. CST must be of
2710 scalar or vector floating point type. Both CST and TYPE must be scalars,
2711 or vectors of the same number of elements. If the value won't fit in the
2712 integer type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002713
Dan Gohmand6a6f612010-05-28 17:07:41 +00002714 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002715 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002716 constant. TYPE must be a scalar or vector integer type. CST must be of
2717 scalar or vector floating point type. Both CST and TYPE must be scalars,
2718 or vectors of the same number of elements. If the value won't fit in the
2719 integer type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002720
Dan Gohmand6a6f612010-05-28 17:07:41 +00002721 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002722 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002723 constant. TYPE must be a scalar or vector floating point type. CST must be
2724 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2725 vectors of the same number of elements. If the value won't fit in the
2726 floating point type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002727
Dan Gohmand6a6f612010-05-28 17:07:41 +00002728 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002729 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002730 constant. TYPE must be a scalar or vector floating point type. CST must be
2731 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2732 vectors of the same number of elements. If the value won't fit in the
2733 floating point type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00002734
Dan Gohmand6a6f612010-05-28 17:07:41 +00002735 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5b950642006-11-11 23:08:07 +00002736 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002737 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2738 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2739 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5b950642006-11-11 23:08:07 +00002740
Dan Gohmand6a6f612010-05-28 17:07:41 +00002741 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002742 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2743 type. CST must be of integer type. The CST value is zero extended,
2744 truncated, or unchanged to make it fit in a pointer size. This one is
2745 <i>really</i> dangerous!</dd>
Reid Spencer5b950642006-11-11 23:08:07 +00002746
Dan Gohmand6a6f612010-05-28 17:07:41 +00002747 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner789dee32009-02-28 18:27:03 +00002748 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2749 are the same as those for the <a href="#i_bitcast">bitcast
2750 instruction</a>.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002751
Dan Gohmand6a6f612010-05-28 17:07:41 +00002752 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2753 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +00002754 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002755 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2756 instruction, the index list may have zero or more indexes, which are
2757 required to make sense for the type of "CSTPTR".</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002758
Dan Gohmand6a6f612010-05-28 17:07:41 +00002759 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002760 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer9965ee72006-12-04 19:23:19 +00002761
Dan Gohmand6a6f612010-05-28 17:07:41 +00002762 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer9965ee72006-12-04 19:23:19 +00002763 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2764
Dan Gohmand6a6f612010-05-28 17:07:41 +00002765 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer9965ee72006-12-04 19:23:19 +00002766 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino7e97a6d2006-01-10 19:31:34 +00002767
Dan Gohmand6a6f612010-05-28 17:07:41 +00002768 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002769 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2770 constants.</dd>
Robert Bocchino7e97a6d2006-01-10 19:31:34 +00002771
Dan Gohmand6a6f612010-05-28 17:07:41 +00002772 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002773 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2774 constants.</dd>
Chris Lattner016a0e52006-04-08 00:13:41 +00002775
Dan Gohmand6a6f612010-05-28 17:07:41 +00002776 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002777 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2778 constants.</dd>
Chris Lattner016a0e52006-04-08 00:13:41 +00002779
Nick Lewycky9ab9a7f2010-05-29 06:44:15 +00002780 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2781 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2782 constants. The index list is interpreted in a similar manner as indices in
2783 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2784 index value must be specified.</dd>
2785
2786 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2787 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2788 constants. The index list is interpreted in a similar manner as indices in
2789 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2790 index value must be specified.</dd>
2791
Dan Gohmand6a6f612010-05-28 17:07:41 +00002792 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002793 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2794 be any of the <a href="#binaryops">binary</a>
2795 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2796 on operands are the same as those for the corresponding instruction
2797 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +00002798</dl>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002799
Chris Lattner74d3f822004-12-09 17:30:23 +00002800</div>
Chris Lattnerb1652612004-03-08 16:49:10 +00002801
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002802</div>
2803
Chris Lattner2f7c9632001-06-06 20:29:01 +00002804<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002805<h2><a name="othervalues">Other Values</a></h2>
Chris Lattner98f013c2006-01-25 23:47:57 +00002806<!-- *********************************************************************** -->
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002807<div>
Chris Lattner98f013c2006-01-25 23:47:57 +00002808<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002809<h3>
Chris Lattner98f013c2006-01-25 23:47:57 +00002810<a name="inlineasm">Inline Assembler Expressions</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002811</h3>
Chris Lattner98f013c2006-01-25 23:47:57 +00002812
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002813<div>
Chris Lattner98f013c2006-01-25 23:47:57 +00002814
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002815<p>LLVM supports inline assembler expressions (as opposed
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002816 to <a href="#moduleasm">Module-Level Inline Assembly</a>) through the use of
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002817 a special value. This value represents the inline assembler as a string
2818 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002819 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen1cfb9582009-10-21 23:28:00 +00002820 expression has side effects, and a flag indicating whether the function
2821 containing the asm needs to align its stack conservatively. An example
2822 inline assembler expression is:</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002823
Benjamin Kramer79698be2010-07-13 12:26:09 +00002824<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +00002825i32 (i32) asm "bswap $0", "=r,r"
Chris Lattner98f013c2006-01-25 23:47:57 +00002826</pre>
2827
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002828<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2829 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2830 have:</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002831
Benjamin Kramer79698be2010-07-13 12:26:09 +00002832<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +00002833%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattner98f013c2006-01-25 23:47:57 +00002834</pre>
2835
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002836<p>Inline asms with side effects not visible in the constraint list must be
2837 marked as having side effects. This is done through the use of the
2838 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002839
Benjamin Kramer79698be2010-07-13 12:26:09 +00002840<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +00002841call void asm sideeffect "eieio", ""()
Chris Lattner98f013c2006-01-25 23:47:57 +00002842</pre>
2843
Dale Johannesen1cfb9582009-10-21 23:28:00 +00002844<p>In some cases inline asms will contain code that will not work unless the
2845 stack is aligned in some way, such as calls or SSE instructions on x86,
2846 yet will not contain code that does that alignment within the asm.
2847 The compiler should make conservative assumptions about what the asm might
2848 contain and should generate its usual stack alignment code in the prologue
2849 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002850
Benjamin Kramer79698be2010-07-13 12:26:09 +00002851<pre class="doc_code">
Dale Johannesen1cfb9582009-10-21 23:28:00 +00002852call void asm alignstack "eieio", ""()
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002853</pre>
Dale Johannesen63c94fe2009-10-13 21:56:55 +00002854
2855<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2856 first.</p>
2857
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002858<!--
Chris Lattner98f013c2006-01-25 23:47:57 +00002859<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00002860 documented here. Constraints on what can be done (e.g. duplication, moving,
2861 etc need to be documented). This is probably best done by reference to
2862 another document that covers inline asm from a holistic perspective.</p>
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002863 -->
Chris Lattner51065562010-04-07 05:38:05 +00002864
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002865<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002866<h4>
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002867 <a name="inlineasm_md">Inline Asm Metadata</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002868</h4>
Chris Lattner51065562010-04-07 05:38:05 +00002869
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002870<div>
Chris Lattner51065562010-04-07 05:38:05 +00002871
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002872<p>The call instructions that wrap inline asm nodes may have a
2873 "<tt>!srcloc</tt>" MDNode attached to it that contains a list of constant
2874 integers. If present, the code generator will use the integer as the
2875 location cookie value when report errors through the <tt>LLVMContext</tt>
2876 error reporting mechanisms. This allows a front-end to correlate backend
2877 errors that occur with inline asm back to the source code that produced it.
2878 For example:</p>
Chris Lattner51065562010-04-07 05:38:05 +00002879
Benjamin Kramer79698be2010-07-13 12:26:09 +00002880<pre class="doc_code">
Chris Lattner51065562010-04-07 05:38:05 +00002881call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2882...
2883!42 = !{ i32 1234567 }
2884</pre>
Chris Lattner51065562010-04-07 05:38:05 +00002885
2886<p>It is up to the front-end to make sense of the magic numbers it places in the
Bill Wendlingad8b58b2011-11-30 21:52:43 +00002887 IR. If the MDNode contains multiple constants, the code generator will use
Chris Lattner79ffdc72010-11-17 08:20:42 +00002888 the one that corresponds to the line of the asm that the error occurs on.</p>
Chris Lattner98f013c2006-01-25 23:47:57 +00002889
2890</div>
2891
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002892</div>
2893
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002894<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00002895<h3>
2896 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2897</h3>
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002898
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00002899<div>
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002900
2901<p>LLVM IR allows metadata to be attached to instructions in the program that
2902 can convey extra information about the code to the optimizers and code
2903 generator. One example application of metadata is source-level debug
2904 information. There are two metadata primitives: strings and nodes. All
2905 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2906 preceding exclamation point ('<tt>!</tt>').</p>
2907
2908<p>A metadata string is a string surrounded by double quotes. It can contain
Bill Wendlingb6c22202011-11-30 21:43:43 +00002909 any character by escaping non-printable characters with "<tt>\xx</tt>" where
2910 "<tt>xx</tt>" is the two digit hex code. For example:
2911 "<tt>!"test\00"</tt>".</p>
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002912
2913<p>Metadata nodes are represented with notation similar to structure constants
2914 (a comma separated list of elements, surrounded by braces and preceded by an
Bill Wendlingb6c22202011-11-30 21:43:43 +00002915 exclamation point). Metadata nodes can have any values as their operand. For
2916 example:</p>
2917
2918<div class="doc_code">
2919<pre>
2920!{ metadata !"test\00", i32 10}
2921</pre>
2922</div>
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002923
2924<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2925 metadata nodes, which can be looked up in the module symbol table. For
Bill Wendlingb6c22202011-11-30 21:43:43 +00002926 example:</p>
2927
2928<div class="doc_code">
2929<pre>
2930!foo = metadata !{!4, !3}
2931</pre>
2932</div>
Chris Lattnerc2f8f162010-01-15 21:50:19 +00002933
Devang Patel9984bd62010-03-04 23:44:48 +00002934<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
Bill Wendlingb6c22202011-11-30 21:43:43 +00002935 function is using two metadata arguments:</p>
Devang Patel9984bd62010-03-04 23:44:48 +00002936
Bill Wendlingc0e10672011-03-02 02:17:11 +00002937<div class="doc_code">
2938<pre>
2939call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2940</pre>
2941</div>
Devang Patel9984bd62010-03-04 23:44:48 +00002942
2943<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
Bill Wendlingb6c22202011-11-30 21:43:43 +00002944 attached to the <tt>add</tt> instruction using the <tt>!dbg</tt>
2945 identifier:</p>
Devang Patel9984bd62010-03-04 23:44:48 +00002946
Bill Wendlingc0e10672011-03-02 02:17:11 +00002947<div class="doc_code">
2948<pre>
2949%indvar.next = add i64 %indvar, 1, !dbg !21
2950</pre>
2951</div>
2952
Peter Collingbourneec9ff672011-10-27 19:19:07 +00002953<p>More information about specific metadata nodes recognized by the optimizers
2954 and code generator is found below.</p>
2955
Bill Wendlingb6c22202011-11-30 21:43:43 +00002956<!-- _______________________________________________________________________ -->
Peter Collingbourneec9ff672011-10-27 19:19:07 +00002957<h4>
2958 <a name="tbaa">'<tt>tbaa</tt>' Metadata</a>
2959</h4>
2960
2961<div>
2962
2963<p>In LLVM IR, memory does not have types, so LLVM's own type system is not
2964 suitable for doing TBAA. Instead, metadata is added to the IR to describe
2965 a type system of a higher level language. This can be used to implement
2966 typical C/C++ TBAA, but it can also be used to implement custom alias
2967 analysis behavior for other languages.</p>
2968
2969<p>The current metadata format is very simple. TBAA metadata nodes have up to
2970 three fields, e.g.:</p>
2971
2972<div class="doc_code">
2973<pre>
2974!0 = metadata !{ metadata !"an example type tree" }
2975!1 = metadata !{ metadata !"int", metadata !0 }
2976!2 = metadata !{ metadata !"float", metadata !0 }
2977!3 = metadata !{ metadata !"const float", metadata !2, i64 1 }
2978</pre>
2979</div>
2980
2981<p>The first field is an identity field. It can be any value, usually
2982 a metadata string, which uniquely identifies the type. The most important
2983 name in the tree is the name of the root node. Two trees with
2984 different root node names are entirely disjoint, even if they
2985 have leaves with common names.</p>
2986
2987<p>The second field identifies the type's parent node in the tree, or
2988 is null or omitted for a root node. A type is considered to alias
2989 all of its descendants and all of its ancestors in the tree. Also,
2990 a type is considered to alias all types in other trees, so that
2991 bitcode produced from multiple front-ends is handled conservatively.</p>
2992
2993<p>If the third field is present, it's an integer which if equal to 1
2994 indicates that the type is "constant" (meaning
2995 <tt>pointsToConstantMemory</tt> should return true; see
2996 <a href="AliasAnalysis.html#OtherItfs">other useful
2997 <tt>AliasAnalysis</tt> methods</a>).</p>
2998
2999</div>
3000
Bill Wendlingb6c22202011-11-30 21:43:43 +00003001<!-- _______________________________________________________________________ -->
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003002<h4>
Duncan Sands34bd91a2012-04-14 12:36:06 +00003003 <a name="fpmath">'<tt>fpmath</tt>' Metadata</a>
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003004</h4>
3005
3006<div>
3007
Duncan Sands34bd91a2012-04-14 12:36:06 +00003008<p><tt>fpmath</tt> metadata may be attached to any instruction of floating point
Duncan Sands05f4df82012-04-16 16:28:59 +00003009 type. It can be used to express the maximum acceptable error in the result of
3010 that instruction, in ULPs, thus potentially allowing the compiler to use a
Duncan Sands9af62982012-04-16 19:39:33 +00003011 more efficient but less accurate method of computing it. ULP is defined as
3012 follows:</p>
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003013
Bill Wendling302d7ce2011-11-09 19:33:56 +00003014<blockquote>
3015
3016<p>If <tt>x</tt> is a real number that lies between two finite consecutive
3017 floating-point numbers <tt>a</tt> and <tt>b</tt>, without being equal to one
3018 of them, then <tt>ulp(x) = |b - a|</tt>, otherwise <tt>ulp(x)</tt> is the
3019 distance between the two non-equal finite floating-point numbers nearest
3020 <tt>x</tt>. Moreover, <tt>ulp(NaN)</tt> is <tt>NaN</tt>.</p>
3021
3022</blockquote>
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003023
Duncan Sands05f4df82012-04-16 16:28:59 +00003024<p>The metadata node shall consist of a single positive floating point number
Duncan Sands9af62982012-04-16 19:39:33 +00003025 representing the maximum relative error, for example:</p>
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003026
3027<div class="doc_code">
3028<pre>
Duncan Sands05f4df82012-04-16 16:28:59 +00003029!0 = metadata !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003030</pre>
3031</div>
3032
NAKAMURA Takumic9d9b922012-03-27 11:25:16 +00003033</div>
3034
Rafael Espindolaef9f5502012-03-24 00:14:51 +00003035<!-- _______________________________________________________________________ -->
3036<h4>
3037 <a name="range">'<tt>range</tt>' Metadata</a>
3038</h4>
3039
3040<div>
3041<p><tt>range</tt> metadata may be attached only to loads of integer types. It
3042 expresses the possible ranges the loaded value is in. The ranges are
3043 represented with a flattened list of integers. The loaded value is known to
3044 be in the union of the ranges defined by each consecutive pair. Each pair
3045 has the following properties:</p>
3046<ul>
3047 <li>The type must match the type loaded by the instruction.</li>
3048 <li>The pair <tt>a,b</tt> represents the range <tt>[a,b)</tt>.</li>
3049 <li>Both <tt>a</tt> and <tt>b</tt> are constants.</li>
3050 <li>The range is allowed to wrap.</li>
3051 <li>The range should not represent the full or empty set. That is,
3052 <tt>a!=b</tt>. </li>
3053</ul>
3054
3055<p>Examples:</p>
3056<div class="doc_code">
3057<pre>
3058 %a = load i8* %x, align 1, !range !0 ; Can only be 0 or 1
3059 %b = load i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1
3060 %c = load i8* %z, align 1, !range !2 ; Can only be 0, 1, 3, 4 or 5
3061...
3062!0 = metadata !{ i8 0, i8 2 }
3063!1 = metadata !{ i8 255, i8 2 }
3064!2 = metadata !{ i8 0, i8 2, i8 3, i8 6 }
3065</pre>
3066</div>
3067</div>
Peter Collingbournef7d1e7b2011-10-27 19:19:14 +00003068</div>
3069
Chris Lattnerc2f8f162010-01-15 21:50:19 +00003070</div>
3071
Chris Lattnerae76db52009-07-20 05:55:19 +00003072<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003073<h2>
Bill Wendling911fdf42012-02-11 11:59:36 +00003074 <a name="module_flags">Module Flags Metadata</a>
3075</h2>
3076<!-- *********************************************************************** -->
3077
3078<div>
3079
3080<p>Information about the module as a whole is difficult to convey to LLVM's
3081 subsystems. The LLVM IR isn't sufficient to transmit this
3082 information. The <tt>llvm.module.flags</tt> named metadata exists in order to
3083 facilitate this. These flags are in the form of key / value pairs &mdash;
3084 much like a dictionary &mdash; making it easy for any subsystem who cares
3085 about a flag to look it up.</p>
3086
3087<p>The <tt>llvm.module.flags</tt> metadata contains a list of metadata
3088 triplets. Each triplet has the following form:</p>
3089
3090<ul>
3091 <li>The first element is a <i>behavior</i> flag, which specifies the behavior
3092 when two (or more) modules are merged together, and it encounters two (or
3093 more) metadata with the same ID. The supported behaviors are described
3094 below.</li>
3095
3096 <li>The second element is a metadata string that is a unique ID for the
3097 metadata. How each ID is interpreted is documented below.</li>
3098
3099 <li>The third element is the value of the flag.</li>
3100</ul>
3101
3102<p>When two (or more) modules are merged together, the resulting
3103 <tt>llvm.module.flags</tt> metadata is the union of the
3104 modules' <tt>llvm.module.flags</tt> metadata. The only exception being a flag
3105 with the <i>Override</i> behavior, which may override another flag's value
3106 (see below).</p>
3107
3108<p>The following behaviors are supported:</p>
3109
3110<table border="1" cellspacing="0" cellpadding="4">
3111 <tbody>
3112 <tr>
3113 <th>Value</th>
3114 <th>Behavior</th>
3115 </tr>
3116 <tr>
3117 <td>1</td>
3118 <td align="left">
Bill Wendlingd672d9c2012-03-06 09:17:04 +00003119 <dl>
3120 <dt><b>Error</b></dt>
3121 <dd>Emits an error if two values disagree. It is an error to have an ID
3122 with both an Error and a Warning behavior.</dd>
3123 </dl>
Bill Wendling911fdf42012-02-11 11:59:36 +00003124 </td>
3125 </tr>
3126 <tr>
3127 <td>2</td>
3128 <td align="left">
Bill Wendlingd672d9c2012-03-06 09:17:04 +00003129 <dl>
3130 <dt><b>Warning</b></dt>
3131 <dd>Emits a warning if two values disagree.</dd>
3132 </dl>
Bill Wendling911fdf42012-02-11 11:59:36 +00003133 </td>
3134 </tr>
3135 <tr>
3136 <td>3</td>
3137 <td align="left">
Bill Wendlingd672d9c2012-03-06 09:17:04 +00003138 <dl>
3139 <dt><b>Require</b></dt>
3140 <dd>Emits an error when the specified value is not present or doesn't
3141 have the specified value. It is an error for two (or more)
3142 <tt>llvm.module.flags</tt> with the same ID to have the Require
3143 behavior but different values. There may be multiple Require flags
3144 per ID.</dd>
3145 </dl>
Bill Wendling911fdf42012-02-11 11:59:36 +00003146 </td>
3147 </tr>
3148 <tr>
3149 <td>4</td>
3150 <td align="left">
Bill Wendlingd672d9c2012-03-06 09:17:04 +00003151 <dl>
3152 <dt><b>Override</b></dt>
3153 <dd>Uses the specified value if the two values disagree. It is an
3154 error for two (or more) <tt>llvm.module.flags</tt> with the same
3155 ID to have the Override behavior but different values.</dd>
3156 </dl>
Bill Wendling911fdf42012-02-11 11:59:36 +00003157 </td>
3158 </tr>
3159 </tbody>
3160</table>
3161
3162<p>An example of module flags:</p>
3163
3164<pre class="doc_code">
3165!0 = metadata !{ i32 1, metadata !"foo", i32 1 }
3166!1 = metadata !{ i32 4, metadata !"bar", i32 37 }
3167!2 = metadata !{ i32 2, metadata !"qux", i32 42 }
3168!3 = metadata !{ i32 3, metadata !"qux",
3169 metadata !{
3170 metadata !"foo", i32 1
3171 }
3172}
3173!llvm.module.flags = !{ !0, !1, !2, !3 }
3174</pre>
3175
3176<ul>
3177 <li><p>Metadata <tt>!0</tt> has the ID <tt>!"foo"</tt> and the value '1'. The
3178 behavior if two or more <tt>!"foo"</tt> flags are seen is to emit an
3179 error if their values are not equal.</p></li>
3180
3181 <li><p>Metadata <tt>!1</tt> has the ID <tt>!"bar"</tt> and the value '37'. The
3182 behavior if two or more <tt>!"bar"</tt> flags are seen is to use the
3183 value '37' if their values are not equal.</p></li>
3184
3185 <li><p>Metadata <tt>!2</tt> has the ID <tt>!"qux"</tt> and the value '42'. The
3186 behavior if two or more <tt>!"qux"</tt> flags are seen is to emit a
3187 warning if their values are not equal.</p></li>
3188
3189 <li><p>Metadata <tt>!3</tt> has the ID <tt>!"qux"</tt> and the value:</p>
3190
3191<pre class="doc_code">
3192metadata !{ metadata !"foo", i32 1 }
3193</pre>
Bill Wendling73462772012-02-16 01:10:50 +00003194
Bill Wendling911fdf42012-02-11 11:59:36 +00003195 <p>The behavior is to emit an error if the <tt>llvm.module.flags</tt> does
3196 not contain a flag with the ID <tt>!"foo"</tt> that has the value
3197 '1'. If two or more <tt>!"qux"</tt> flags exist, then they must have
3198 the same value or an error will be issued.</p></li>
3199</ul>
3200
Bill Wendling73462772012-02-16 01:10:50 +00003201
3202<!-- ======================================================================= -->
3203<h3>
3204<a name="objc_gc_flags">Objective-C Garbage Collection Module Flags Metadata</a>
3205</h3>
3206
3207<div>
3208
3209<p>On the Mach-O platform, Objective-C stores metadata about garbage collection
3210 in a special section called "image info". The metadata consists of a version
3211 number and a bitmask specifying what types of garbage collection are
3212 supported (if any) by the file. If two or more modules are linked together
3213 their garbage collection metadata needs to be merged rather than appended
3214 together.</p>
3215
3216<p>The Objective-C garbage collection module flags metadata consists of the
3217 following key-value pairs:</p>
3218
3219<table border="1" cellspacing="0" cellpadding="4">
Bill Wendling4fa13cc2012-03-06 09:23:25 +00003220 <col width="30%">
Bill Wendling73462772012-02-16 01:10:50 +00003221 <tbody>
3222 <tr>
Bill Wendlingd672d9c2012-03-06 09:17:04 +00003223 <th>Key</th>
Bill Wendling73462772012-02-16 01:10:50 +00003224 <th>Value</th>
3225 </tr>
3226 <tr>
3227 <td><tt>Objective-C&nbsp;Version</tt></td>
3228 <td align="left"><b>[Required]</b> &mdash; The Objective-C ABI
3229 version. Valid values are 1 and 2.</td>
3230 </tr>
3231 <tr>
3232 <td><tt>Objective-C&nbsp;Image&nbsp;Info&nbsp;Version</tt></td>
3233 <td align="left"><b>[Required]</b> &mdash; The version of the image info
3234 section. Currently always 0.</td>
3235 </tr>
3236 <tr>
3237 <td><tt>Objective-C&nbsp;Image&nbsp;Info&nbsp;Section</tt></td>
3238 <td align="left"><b>[Required]</b> &mdash; The section to place the
3239 metadata. Valid values are <tt>"__OBJC, __image_info, regular"</tt> for
3240 Objective-C ABI version 1, and <tt>"__DATA,__objc_imageinfo, regular,
3241 no_dead_strip"</tt> for Objective-C ABI version 2.</td>
3242 </tr>
3243 <tr>
3244 <td><tt>Objective-C&nbsp;Garbage&nbsp;Collection</tt></td>
3245 <td align="left"><b>[Required]</b> &mdash; Specifies whether garbage
3246 collection is supported or not. Valid values are 0, for no garbage
3247 collection, and 2, for garbage collection supported.</td>
3248 </tr>
3249 <tr>
3250 <td><tt>Objective-C&nbsp;GC&nbsp;Only</tt></td>
3251 <td align="left"><b>[Optional]</b> &mdash; Specifies that only garbage
3252 collection is supported. If present, its value must be 6. This flag
3253 requires that the <tt>Objective-C Garbage Collection</tt> flag have the
3254 value 2.</td>
3255 </tr>
3256 </tbody>
3257</table>
3258
3259<p>Some important flag interactions:</p>
3260
3261<ul>
3262 <li>If a module with <tt>Objective-C Garbage Collection</tt> set to 0 is
3263 merged with a module with <tt>Objective-C Garbage Collection</tt> set to
3264 2, then the resulting module has the <tt>Objective-C Garbage
3265 Collection</tt> flag set to 0.</li>
3266
3267 <li>A module with <tt>Objective-C Garbage Collection</tt> set to 0 cannot be
3268 merged with a module with <tt>Objective-C GC Only</tt> set to 6.</li>
3269</ul>
3270
3271</div>
3272
Bill Wendling911fdf42012-02-11 11:59:36 +00003273</div>
3274
3275<!-- *********************************************************************** -->
3276<h2>
Chris Lattnerae76db52009-07-20 05:55:19 +00003277 <a name="intrinsic_globals">Intrinsic Global Variables</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003278</h2>
Chris Lattnerae76db52009-07-20 05:55:19 +00003279<!-- *********************************************************************** -->
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003280<div>
Chris Lattnerae76db52009-07-20 05:55:19 +00003281<p>LLVM has a number of "magic" global variables that contain data that affect
3282code generation or other IR semantics. These are documented here. All globals
Chris Lattner58f9bb22009-07-20 06:14:25 +00003283of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
3284section and all globals that start with "<tt>llvm.</tt>" are reserved for use
3285by LLVM.</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00003286
3287<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003288<h3>
Chris Lattnerae76db52009-07-20 05:55:19 +00003289<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003290</h3>
Chris Lattnerae76db52009-07-20 05:55:19 +00003291
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003292<div>
Chris Lattnerae76db52009-07-20 05:55:19 +00003293
3294<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
3295href="#linkage_appending">appending linkage</a>. This array contains a list of
3296pointers to global variables and functions which may optionally have a pointer
3297cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
3298
Bill Wendling1654bb22011-11-08 00:32:45 +00003299<div class="doc_code">
Chris Lattnerae76db52009-07-20 05:55:19 +00003300<pre>
Bill Wendling1654bb22011-11-08 00:32:45 +00003301@X = global i8 4
3302@Y = global i32 123
Chris Lattnerae76db52009-07-20 05:55:19 +00003303
Bill Wendling1654bb22011-11-08 00:32:45 +00003304@llvm.used = appending global [2 x i8*] [
3305 i8* @X,
3306 i8* bitcast (i32* @Y to i8*)
3307], section "llvm.metadata"
Chris Lattnerae76db52009-07-20 05:55:19 +00003308</pre>
Bill Wendling1654bb22011-11-08 00:32:45 +00003309</div>
Chris Lattnerae76db52009-07-20 05:55:19 +00003310
3311<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
Bill Wendling1654bb22011-11-08 00:32:45 +00003312 compiler, assembler, and linker are required to treat the symbol as if there
3313 is a reference to the global that it cannot see. For example, if a variable
3314 has internal linkage and no references other than that from
3315 the <tt>@llvm.used</tt> list, it cannot be deleted. This is commonly used to
3316 represent references from inline asms and other things the compiler cannot
3317 "see", and corresponds to "<tt>attribute((used))</tt>" in GNU C.</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00003318
3319<p>On some targets, the code generator must emit a directive to the assembler or
Bill Wendling1654bb22011-11-08 00:32:45 +00003320 object file to prevent the assembler and linker from molesting the
3321 symbol.</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00003322
3323</div>
3324
3325<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003326<h3>
3327 <a name="intg_compiler_used">
3328 The '<tt>llvm.compiler.used</tt>' Global Variable
3329 </a>
3330</h3>
Chris Lattner58f9bb22009-07-20 06:14:25 +00003331
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003332<div>
Chris Lattner58f9bb22009-07-20 06:14:25 +00003333
3334<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
Bill Wendling1654bb22011-11-08 00:32:45 +00003335 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
3336 touching the symbol. On targets that support it, this allows an intelligent
3337 linker to optimize references to the symbol without being impeded as it would
3338 be by <tt>@llvm.used</tt>.</p>
Chris Lattner58f9bb22009-07-20 06:14:25 +00003339
3340<p>This is a rare construct that should only be used in rare circumstances, and
Bill Wendling1654bb22011-11-08 00:32:45 +00003341 should not be exposed to source languages.</p>
Chris Lattner58f9bb22009-07-20 06:14:25 +00003342
3343</div>
3344
3345<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003346<h3>
Chris Lattnerae76db52009-07-20 05:55:19 +00003347<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003348</h3>
Chris Lattnerae76db52009-07-20 05:55:19 +00003349
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003350<div>
Bill Wendling1654bb22011-11-08 00:32:45 +00003351
3352<div class="doc_code">
David Chisnalla9d4a6f2010-04-30 19:23:49 +00003353<pre>
3354%0 = type { i32, void ()* }
David Chisnallb492b812010-04-30 19:27:35 +00003355@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalla9d4a6f2010-04-30 19:23:49 +00003356</pre>
Bill Wendling1654bb22011-11-08 00:32:45 +00003357</div>
3358
3359<p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor
3360 functions and associated priorities. The functions referenced by this array
3361 will be called in ascending order of priority (i.e. lowest first) when the
3362 module is loaded. The order of functions with the same priority is not
3363 defined.</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00003364
3365</div>
3366
3367<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003368<h3>
Chris Lattnerae76db52009-07-20 05:55:19 +00003369<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003370</h3>
Chris Lattnerae76db52009-07-20 05:55:19 +00003371
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003372<div>
Bill Wendling1654bb22011-11-08 00:32:45 +00003373
3374<div class="doc_code">
David Chisnalla9d4a6f2010-04-30 19:23:49 +00003375<pre>
3376%0 = type { i32, void ()* }
David Chisnallb492b812010-04-30 19:27:35 +00003377@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalla9d4a6f2010-04-30 19:23:49 +00003378</pre>
Bill Wendling1654bb22011-11-08 00:32:45 +00003379</div>
Chris Lattnerae76db52009-07-20 05:55:19 +00003380
Bill Wendling1654bb22011-11-08 00:32:45 +00003381<p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions
3382 and associated priorities. The functions referenced by this array will be
3383 called in descending order of priority (i.e. highest first) when the module
3384 is loaded. The order of functions with the same priority is not defined.</p>
Chris Lattnerae76db52009-07-20 05:55:19 +00003385
3386</div>
3387
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003388</div>
Chris Lattnerae76db52009-07-20 05:55:19 +00003389
Chris Lattner98f013c2006-01-25 23:47:57 +00003390<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003391<h2><a name="instref">Instruction Reference</a></h2>
Chris Lattner48b383b02003-11-25 01:02:51 +00003392<!-- *********************************************************************** -->
Chris Lattner74d3f822004-12-09 17:30:23 +00003393
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003394<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00003395
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003396<p>The LLVM instruction set consists of several different classifications of
3397 instructions: <a href="#terminators">terminator
3398 instructions</a>, <a href="#binaryops">binary instructions</a>,
3399 <a href="#bitwiseops">bitwise binary instructions</a>,
3400 <a href="#memoryops">memory instructions</a>, and
3401 <a href="#otherops">other instructions</a>.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00003402
Chris Lattner2f7c9632001-06-06 20:29:01 +00003403<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003404<h3>
3405 <a name="terminators">Terminator Instructions</a>
3406</h3>
Chris Lattner74d3f822004-12-09 17:30:23 +00003407
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003408<div>
Chris Lattner74d3f822004-12-09 17:30:23 +00003409
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003410<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3411 in a program ends with a "Terminator" instruction, which indicates which
3412 block should be executed after the current block is finished. These
3413 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3414 control flow, not values (the one exception being the
3415 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3416
Chris Lattnerd3d65ab2011-08-02 20:29:13 +00003417<p>The terminator instructions are:
3418 '<a href="#i_ret"><tt>ret</tt></a>',
3419 '<a href="#i_br"><tt>br</tt></a>',
3420 '<a href="#i_switch"><tt>switch</tt></a>',
3421 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3422 '<a href="#i_invoke"><tt>invoke</tt></a>',
Chris Lattnerd3d65ab2011-08-02 20:29:13 +00003423 '<a href="#i_resume"><tt>resume</tt></a>', and
3424 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +00003425
Chris Lattner2f7c9632001-06-06 20:29:01 +00003426<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003427<h4>
3428 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3429</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003430
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003431<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003432
Chris Lattner2f7c9632001-06-06 20:29:01 +00003433<h5>Syntax:</h5>
Dan Gohmancc3132e2008-10-04 19:00:07 +00003434<pre>
3435 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner590645f2002-04-14 06:13:44 +00003436 ret void <i>; Return from void function</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003437</pre>
Chris Lattnerda508ac2008-04-23 04:59:35 +00003438
Chris Lattner2f7c9632001-06-06 20:29:01 +00003439<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003440<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3441 a value) from a function back to the caller.</p>
3442
3443<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3444 value and then causes control flow, and one that just causes control flow to
3445 occur.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00003446
Chris Lattner2f7c9632001-06-06 20:29:01 +00003447<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003448<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3449 return value. The type of the return value must be a
3450 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmancc3132e2008-10-04 19:00:07 +00003451
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003452<p>A function is not <a href="#wellformed">well formed</a> if it it has a
3453 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3454 value or a return value with a type that does not match its type, or if it
3455 has a void return type and contains a '<tt>ret</tt>' instruction with a
3456 return value.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00003457
Chris Lattner2f7c9632001-06-06 20:29:01 +00003458<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003459<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3460 the calling function's context. If the caller is a
3461 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3462 instruction after the call. If the caller was an
3463 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3464 the beginning of the "normal" destination block. If the instruction returns
3465 a value, that value shall set the call or invoke instruction's return
3466 value.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +00003467
Chris Lattner2f7c9632001-06-06 20:29:01 +00003468<h5>Example:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +00003469<pre>
3470 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner590645f2002-04-14 06:13:44 +00003471 ret void <i>; Return from a void function</i>
Bill Wendling050ee8f2009-02-28 22:12:54 +00003472 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003473</pre>
Dan Gohman3065b612009-01-12 23:12:39 +00003474
Misha Brukman76307852003-11-08 01:05:38 +00003475</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003476<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003477<h4>
3478 <a name="i_br">'<tt>br</tt>' Instruction</a>
3479</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003480
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003481<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003482
Chris Lattner2f7c9632001-06-06 20:29:01 +00003483<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003484<pre>
Bill Wendling16b86742011-07-26 10:41:15 +00003485 br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;
3486 br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003487</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003488
Chris Lattner2f7c9632001-06-06 20:29:01 +00003489<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003490<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3491 different basic block in the current function. There are two forms of this
3492 instruction, corresponding to a conditional branch and an unconditional
3493 branch.</p>
3494
Chris Lattner2f7c9632001-06-06 20:29:01 +00003495<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003496<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3497 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3498 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3499 target.</p>
3500
Chris Lattner2f7c9632001-06-06 20:29:01 +00003501<h5>Semantics:</h5>
Reid Spencer36a15422007-01-12 03:35:51 +00003502<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003503 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3504 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3505 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3506
Chris Lattner2f7c9632001-06-06 20:29:01 +00003507<h5>Example:</h5>
Bill Wendling30235112009-07-20 02:39:26 +00003508<pre>
3509Test:
3510 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3511 br i1 %cond, label %IfEqual, label %IfUnequal
3512IfEqual:
3513 <a href="#i_ret">ret</a> i32 1
3514IfUnequal:
3515 <a href="#i_ret">ret</a> i32 0
3516</pre>
3517
Misha Brukman76307852003-11-08 01:05:38 +00003518</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003519
Chris Lattner2f7c9632001-06-06 20:29:01 +00003520<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003521<h4>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003522 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003523</h4>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003524
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003525<div>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003526
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003527<h5>Syntax:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003528<pre>
3529 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
3530</pre>
3531
Chris Lattner2f7c9632001-06-06 20:29:01 +00003532<h5>Overview:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003533<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003534 several different places. It is a generalization of the '<tt>br</tt>'
3535 instruction, allowing a branch to occur to one of many possible
3536 destinations.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003537
Chris Lattner2f7c9632001-06-06 20:29:01 +00003538<h5>Arguments:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003539<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003540 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3541 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3542 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003543
Chris Lattner2f7c9632001-06-06 20:29:01 +00003544<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +00003545<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003546 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3547 is searched for the given value. If the value is found, control flow is
Benjamin Kramer0f420382009-10-12 14:46:08 +00003548 transferred to the corresponding destination; otherwise, control flow is
3549 transferred to the default destination.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003550
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003551<h5>Implementation:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003552<p>Depending on properties of the target machine and the particular
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003553 <tt>switch</tt> instruction, this instruction may be code generated in
3554 different ways. For example, it could be generated as a series of chained
3555 conditional branches or with a lookup table.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003556
3557<h5>Example:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003558<pre>
3559 <i>; Emulate a conditional br instruction</i>
Reid Spencer36a15422007-01-12 03:35:51 +00003560 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman623806e2009-01-04 23:44:43 +00003561 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003562
3563 <i>; Emulate an unconditional br instruction</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00003564 switch i32 0, label %dest [ ]
Chris Lattnercf96c6c2004-02-24 04:54:45 +00003565
3566 <i>; Implement a jump table:</i>
Dan Gohman623806e2009-01-04 23:44:43 +00003567 switch i32 %val, label %otherwise [ i32 0, label %onzero
3568 i32 1, label %onone
3569 i32 2, label %ontwo ]
Chris Lattner2f7c9632001-06-06 20:29:01 +00003570</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003571
Misha Brukman76307852003-11-08 01:05:38 +00003572</div>
Chris Lattner0132aff2005-05-06 22:57:40 +00003573
Chris Lattner3ed871f2009-10-27 19:13:16 +00003574
3575<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003576<h4>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003577 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003578</h4>
Chris Lattner3ed871f2009-10-27 19:13:16 +00003579
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003580<div>
Chris Lattner3ed871f2009-10-27 19:13:16 +00003581
3582<h5>Syntax:</h5>
3583<pre>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003584 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattner3ed871f2009-10-27 19:13:16 +00003585</pre>
3586
3587<h5>Overview:</h5>
3588
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003589<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattner3ed871f2009-10-27 19:13:16 +00003590 within the current function, whose address is specified by
Chris Lattnere4801f72009-10-27 21:01:34 +00003591 "<tt>address</tt>". Address must be derived from a <a
3592 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattner3ed871f2009-10-27 19:13:16 +00003593
3594<h5>Arguments:</h5>
3595
3596<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3597 rest of the arguments indicate the full set of possible destinations that the
3598 address may point to. Blocks are allowed to occur multiple times in the
3599 destination list, though this isn't particularly useful.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00003600
Chris Lattner3ed871f2009-10-27 19:13:16 +00003601<p>This destination list is required so that dataflow analysis has an accurate
3602 understanding of the CFG.</p>
3603
3604<h5>Semantics:</h5>
3605
3606<p>Control transfers to the block specified in the address argument. All
3607 possible destination blocks must be listed in the label list, otherwise this
3608 instruction has undefined behavior. This implies that jumps to labels
3609 defined in other functions have undefined behavior as well.</p>
3610
3611<h5>Implementation:</h5>
3612
3613<p>This is typically implemented with a jump through a register.</p>
3614
3615<h5>Example:</h5>
3616<pre>
Chris Lattnerd04cb6d2009-10-28 00:19:10 +00003617 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattner3ed871f2009-10-27 19:13:16 +00003618</pre>
3619
3620</div>
3621
3622
Chris Lattner2f7c9632001-06-06 20:29:01 +00003623<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003624<h4>
Chris Lattner0132aff2005-05-06 22:57:40 +00003625 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003626</h4>
Chris Lattner0132aff2005-05-06 22:57:40 +00003627
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003628<div>
Chris Lattner0132aff2005-05-06 22:57:40 +00003629
Chris Lattner2f7c9632001-06-06 20:29:01 +00003630<h5>Syntax:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00003631<pre>
Devang Patel02256232008-10-07 17:48:33 +00003632 &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 +00003633 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattner0132aff2005-05-06 22:57:40 +00003634</pre>
3635
Chris Lattnera8292f32002-05-06 22:08:29 +00003636<h5>Overview:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00003637<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003638 function, with the possibility of control flow transfer to either the
3639 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3640 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3641 control flow will return to the "normal" label. If the callee (or any
Bill Wendling3f6a3a22012-02-06 21:57:33 +00003642 indirect callees) returns via the "<a href="#i_resume"><tt>resume</tt></a>"
3643 instruction or other exception handling mechanism, control is interrupted and
3644 continued at the dynamically nearest "exception" label.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003645
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00003646<p>The '<tt>exception</tt>' label is a
3647 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3648 exception. As such, '<tt>exception</tt>' label is required to have the
3649 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
Chad Rosierc28f3e92011-12-09 02:00:44 +00003650 the information about the behavior of the program after unwinding
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00003651 happens, as its first non-PHI instruction. The restrictions on the
3652 "<tt>landingpad</tt>" instruction's tightly couples it to the
3653 "<tt>invoke</tt>" instruction, so that the important information contained
3654 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3655 code motion.</p>
3656
Chris Lattner2f7c9632001-06-06 20:29:01 +00003657<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003658<p>This instruction requires several arguments:</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003659
Chris Lattner2f7c9632001-06-06 20:29:01 +00003660<ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003661 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3662 convention</a> the call should use. If none is specified, the call
3663 defaults to using C calling conventions.</li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00003664
3665 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003666 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3667 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00003668
Chris Lattner0132aff2005-05-06 22:57:40 +00003669 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003670 function value being invoked. In most cases, this is a direct function
3671 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3672 off an arbitrary pointer to function value.</li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003673
3674 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003675 function to be invoked. </li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003676
3677 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner47f2a832010-03-02 06:36:51 +00003678 signature argument types and parameter attributes. All arguments must be
3679 of <a href="#t_firstclass">first class</a> type. If the function
3680 signature indicates the function accepts a variable number of arguments,
3681 the extra arguments can be specified.</li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003682
3683 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003684 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003685
Bill Wendling3f6a3a22012-02-06 21:57:33 +00003686 <li>'<tt>exception label</tt>': the label reached when a callee returns via
3687 the <a href="#i_resume"><tt>resume</tt></a> instruction or other exception
3688 handling mechanism.</li>
Chris Lattner0132aff2005-05-06 22:57:40 +00003689
Devang Patel02256232008-10-07 17:48:33 +00003690 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003691 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3692 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003693</ol>
Chris Lattner0132aff2005-05-06 22:57:40 +00003694
Chris Lattner2f7c9632001-06-06 20:29:01 +00003695<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003696<p>This instruction is designed to operate as a standard
3697 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3698 primary difference is that it establishes an association with a label, which
3699 is used by the runtime library to unwind the stack.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003700
3701<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003702 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3703 exception. Additionally, this is important for implementation of
3704 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +00003705
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003706<p>For the purposes of the SSA form, the definition of the value returned by the
3707 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3708 block to the "normal" label. If the callee unwinds then no return value is
3709 available.</p>
Dan Gohman9069d892009-05-22 21:47:08 +00003710
Chris Lattner2f7c9632001-06-06 20:29:01 +00003711<h5>Example:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +00003712<pre>
Nick Lewycky084ab472008-03-16 07:18:12 +00003713 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen5819f182007-04-22 01:17:39 +00003714 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewycky084ab472008-03-16 07:18:12 +00003715 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen5819f182007-04-22 01:17:39 +00003716 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003717</pre>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003718
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003719</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003720
Bill Wendlingf891bf82011-07-31 06:30:59 +00003721 <!-- _______________________________________________________________________ -->
3722
3723<h4>
3724 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3725</h4>
3726
3727<div>
3728
3729<h5>Syntax:</h5>
3730<pre>
3731 resume &lt;type&gt; &lt;value&gt;
3732</pre>
3733
3734<h5>Overview:</h5>
3735<p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3736 successors.</p>
3737
3738<h5>Arguments:</h5>
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00003739<p>The '<tt>resume</tt>' instruction requires one argument, which must have the
Bill Wendlingc5a13612011-08-03 18:37:32 +00003740 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3741 function.</p>
Bill Wendlingf891bf82011-07-31 06:30:59 +00003742
3743<h5>Semantics:</h5>
3744<p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3745 (in-flight) exception whose unwinding was interrupted with
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00003746 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
Bill Wendlingf891bf82011-07-31 06:30:59 +00003747
3748<h5>Example:</h5>
3749<pre>
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00003750 resume { i8*, i32 } %exn
Bill Wendlingf891bf82011-07-31 06:30:59 +00003751</pre>
3752
3753</div>
3754
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003755<!-- _______________________________________________________________________ -->
3756
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003757<h4>
3758 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3759</h4>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003760
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003761<div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003762
3763<h5>Syntax:</h5>
3764<pre>
3765 unreachable
3766</pre>
3767
3768<h5>Overview:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003769<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003770 instruction is used to inform the optimizer that a particular portion of the
3771 code is not reachable. This can be used to indicate that the code after a
3772 no-return function cannot be reached, and other facts.</p>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003773
3774<h5>Semantics:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003775<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003776
Chris Lattner08b7d5b2004-10-16 18:04:13 +00003777</div>
3778
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003779</div>
3780
Chris Lattner2f7c9632001-06-06 20:29:01 +00003781<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003782<h3>
3783 <a name="binaryops">Binary Operations</a>
3784</h3>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003785
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003786<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003787
3788<p>Binary operators are used to do most of the computation in a program. They
3789 require two operands of the same type, execute an operation on them, and
3790 produce a single value. The operands might represent multiple data, as is
3791 the case with the <a href="#t_vector">vector</a> data type. The result value
3792 has the same type as its operands.</p>
3793
Misha Brukman76307852003-11-08 01:05:38 +00003794<p>There are several different binary operators:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003795
Chris Lattner2f7c9632001-06-06 20:29:01 +00003796<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003797<h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003798 <a name="i_add">'<tt>add</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003799</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003800
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003801<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003802
Chris Lattner2f7c9632001-06-06 20:29:01 +00003803<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003804<pre>
Dan Gohmanb07de442009-07-20 22:41:19 +00003805 &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 +00003806 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3807 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3808 &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 +00003809</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003810
Chris Lattner2f7c9632001-06-06 20:29:01 +00003811<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003812<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003813
Chris Lattner2f7c9632001-06-06 20:29:01 +00003814<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003815<p>The two arguments to the '<tt>add</tt>' instruction must
3816 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3817 integer values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003818
Chris Lattner2f7c9632001-06-06 20:29:01 +00003819<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003820<p>The value produced is the integer sum of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003821
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003822<p>If the sum has unsigned overflow, the result returned is the mathematical
3823 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003824
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003825<p>Because LLVM integers use a two's complement representation, this instruction
3826 is appropriate for both signed and unsigned integers.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003827
Dan Gohman902dfff2009-07-22 22:44:56 +00003828<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3829 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3830 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohman9a2a0932011-12-06 03:18:47 +00003831 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00003832 respectively, occurs.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00003833
Chris Lattner2f7c9632001-06-06 20:29:01 +00003834<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003835<pre>
3836 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003837</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003838
Misha Brukman76307852003-11-08 01:05:38 +00003839</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003840
Chris Lattner2f7c9632001-06-06 20:29:01 +00003841<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003842<h4>
Dan Gohmana5b96452009-06-04 22:49:04 +00003843 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003844</h4>
Dan Gohmana5b96452009-06-04 22:49:04 +00003845
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003846<div>
Dan Gohmana5b96452009-06-04 22:49:04 +00003847
3848<h5>Syntax:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003849<pre>
3850 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3851</pre>
3852
3853<h5>Overview:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003854<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3855
3856<h5>Arguments:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003857<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003858 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3859 floating point values. Both arguments must have identical types.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003860
3861<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003862<p>The value produced is the floating point sum of the two operands.</p>
3863
3864<h5>Example:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003865<pre>
3866 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3867</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003868
Dan Gohmana5b96452009-06-04 22:49:04 +00003869</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003870
Dan Gohmana5b96452009-06-04 22:49:04 +00003871<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003872<h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003873 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003874</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003875
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003876<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003877
Chris Lattner2f7c9632001-06-06 20:29:01 +00003878<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003879<pre>
Dan Gohman902dfff2009-07-22 22:44:56 +00003880 &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 +00003881 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3882 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3883 &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 +00003884</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003885
Chris Lattner2f7c9632001-06-06 20:29:01 +00003886<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00003887<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003888 operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003889
3890<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003891 '<tt>neg</tt>' instruction present in most other intermediate
3892 representations.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003893
Chris Lattner2f7c9632001-06-06 20:29:01 +00003894<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003895<p>The two arguments to the '<tt>sub</tt>' instruction must
3896 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3897 integer values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003898
Chris Lattner2f7c9632001-06-06 20:29:01 +00003899<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003900<p>The value produced is the integer difference of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003901
Dan Gohmana5b96452009-06-04 22:49:04 +00003902<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003903 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3904 result.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003905
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003906<p>Because LLVM integers use a two's complement representation, this instruction
3907 is appropriate for both signed and unsigned integers.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003908
Dan Gohman902dfff2009-07-22 22:44:56 +00003909<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3910 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3911 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohman9a2a0932011-12-06 03:18:47 +00003912 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00003913 respectively, occurs.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00003914
Chris Lattner2f7c9632001-06-06 20:29:01 +00003915<h5>Example:</h5>
Bill Wendling2d8b9a82007-05-29 09:42:13 +00003916<pre>
3917 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00003918 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00003919</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003920
Misha Brukman76307852003-11-08 01:05:38 +00003921</div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003922
Chris Lattner2f7c9632001-06-06 20:29:01 +00003923<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003924<h4>
Dan Gohmana5b96452009-06-04 22:49:04 +00003925 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003926</h4>
Dan Gohmana5b96452009-06-04 22:49:04 +00003927
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003928<div>
Dan Gohmana5b96452009-06-04 22:49:04 +00003929
3930<h5>Syntax:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003931<pre>
3932 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3933</pre>
3934
3935<h5>Overview:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003936<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003937 operands.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003938
3939<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003940 '<tt>fneg</tt>' instruction present in most other intermediate
3941 representations.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003942
3943<h5>Arguments:</h5>
Bill Wendling972b7202009-07-20 02:32:41 +00003944<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003945 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3946 floating point values. Both arguments must have identical types.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00003947
3948<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003949<p>The value produced is the floating point difference of the two operands.</p>
3950
3951<h5>Example:</h5>
3952<pre>
3953 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3954 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3955</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003956
Dan Gohmana5b96452009-06-04 22:49:04 +00003957</div>
3958
3959<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003960<h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003961 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00003962</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003963
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00003964<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003965
Chris Lattner2f7c9632001-06-06 20:29:01 +00003966<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003967<pre>
Dan Gohman902dfff2009-07-22 22:44:56 +00003968 &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 +00003969 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3970 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3971 &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 +00003972</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003973
Chris Lattner2f7c9632001-06-06 20:29:01 +00003974<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003975<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003976
Chris Lattner2f7c9632001-06-06 20:29:01 +00003977<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003978<p>The two arguments to the '<tt>mul</tt>' instruction must
3979 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3980 integer values. Both arguments must have identical types.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00003981
Chris Lattner2f7c9632001-06-06 20:29:01 +00003982<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00003983<p>The value produced is the integer product of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00003984
Bill Wendlingd9a66f72009-07-20 02:29:24 +00003985<p>If the result of the multiplication has unsigned overflow, the result
3986 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3987 width of the result.</p>
3988
3989<p>Because LLVM integers use a two's complement representation, and the result
3990 is the same width as the operands, this instruction returns the correct
3991 result for both signed and unsigned integers. If a full product
3992 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3993 be sign-extended or zero-extended as appropriate to the width of the full
3994 product.</p>
3995
Dan Gohman902dfff2009-07-22 22:44:56 +00003996<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3997 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3998 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohman9a2a0932011-12-06 03:18:47 +00003999 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanffc9a6b2010-04-22 23:14:21 +00004000 respectively, occurs.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00004001
Chris Lattner2f7c9632001-06-06 20:29:01 +00004002<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004003<pre>
4004 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004005</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004006
Misha Brukman76307852003-11-08 01:05:38 +00004007</div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004008
Chris Lattner2f7c9632001-06-06 20:29:01 +00004009<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004010<h4>
Dan Gohmana5b96452009-06-04 22:49:04 +00004011 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004012</h4>
Dan Gohmana5b96452009-06-04 22:49:04 +00004013
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004014<div>
Dan Gohmana5b96452009-06-04 22:49:04 +00004015
4016<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004017<pre>
4018 &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 +00004019</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004020
Dan Gohmana5b96452009-06-04 22:49:04 +00004021<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004022<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00004023
4024<h5>Arguments:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00004025<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004026 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4027 floating point values. Both arguments must have identical types.</p>
Dan Gohmana5b96452009-06-04 22:49:04 +00004028
4029<h5>Semantics:</h5>
Dan Gohmana5b96452009-06-04 22:49:04 +00004030<p>The value produced is the floating point product of the two operands.</p>
4031
4032<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004033<pre>
4034 &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohmana5b96452009-06-04 22:49:04 +00004035</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004036
Dan Gohmana5b96452009-06-04 22:49:04 +00004037</div>
4038
4039<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004040<h4>
4041 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
4042</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004043
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004044<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004045
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004046<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004047<pre>
Chris Lattner35315d02011-02-06 21:44:57 +00004048 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4049 &lt;result&gt; = udiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004050</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004051
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004052<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004053<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004054
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004055<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004056<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004057 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4058 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004059
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004060<h5>Semantics:</h5>
Chris Lattner2f2427e2008-01-28 00:36:27 +00004061<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004062
Chris Lattner2f2427e2008-01-28 00:36:27 +00004063<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004064 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
4065
Chris Lattner2f2427e2008-01-28 00:36:27 +00004066<p>Division by zero leads to undefined behavior.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004067
Chris Lattner35315d02011-02-06 21:44:57 +00004068<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman9a2a0932011-12-06 03:18:47 +00004069 <tt>udiv</tt> is a <a href="#poisonvalues">poison value</a> if %op1 is not a
Chris Lattner35315d02011-02-06 21:44:57 +00004070 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
4071
4072
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004073<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004074<pre>
4075 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004076</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004077
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004078</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004079
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004080<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004081<h4>
4082 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
4083</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004084
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004085<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004086
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004087<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004088<pre>
Dan Gohmanb07de442009-07-20 22:41:19 +00004089 &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 +00004090 &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 +00004091</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004092
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004093<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004094<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004095
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004096<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004097<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004098 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4099 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004100
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004101<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004102<p>The value produced is the signed integer quotient of the two operands rounded
4103 towards zero.</p>
4104
Chris Lattner2f2427e2008-01-28 00:36:27 +00004105<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004106 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
4107
Chris Lattner2f2427e2008-01-28 00:36:27 +00004108<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004109 undefined behavior; this is a rare case, but can occur, for example, by doing
4110 a 32-bit division of -2147483648 by -1.</p>
4111
Dan Gohman71dfd782009-07-22 00:04:19 +00004112<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman9a2a0932011-12-06 03:18:47 +00004113 <tt>sdiv</tt> is a <a href="#poisonvalues">poison value</a> if the result would
Dan Gohmane501ff72010-07-11 00:08:34 +00004114 be rounded.</p>
Dan Gohmanb07de442009-07-20 22:41:19 +00004115
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004116<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004117<pre>
4118 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004119</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004120
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004121</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004122
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004123<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004124<h4>
4125 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
4126</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004127
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004128<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004129
Chris Lattner2f7c9632001-06-06 20:29:01 +00004130<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004131<pre>
Gabor Greif0f75ad02008-08-07 21:46:00 +00004132 &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 +00004133</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004134
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004135<h5>Overview:</h5>
4136<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004137
Chris Lattner48b383b02003-11-25 01:02:51 +00004138<h5>Arguments:</h5>
Jeff Cohen5819f182007-04-22 01:17:39 +00004139<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004140 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4141 floating point values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004142
Chris Lattner48b383b02003-11-25 01:02:51 +00004143<h5>Semantics:</h5>
Reid Spencer7e80b0b2006-10-26 06:15:43 +00004144<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004145
Chris Lattner48b383b02003-11-25 01:02:51 +00004146<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004147<pre>
4148 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner48b383b02003-11-25 01:02:51 +00004149</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004150
Chris Lattner48b383b02003-11-25 01:02:51 +00004151</div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004152
Chris Lattner48b383b02003-11-25 01:02:51 +00004153<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004154<h4>
4155 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
4156</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004157
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004158<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004159
Reid Spencer7eb55b32006-11-02 01:53:59 +00004160<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004161<pre>
4162 &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 +00004163</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004164
Reid Spencer7eb55b32006-11-02 01:53:59 +00004165<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004166<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
4167 division of its two arguments.</p>
4168
Reid Spencer7eb55b32006-11-02 01:53:59 +00004169<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004170<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004171 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4172 values. Both arguments must have identical types.</p>
4173
Reid Spencer7eb55b32006-11-02 01:53:59 +00004174<h5>Semantics:</h5>
4175<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004176 This instruction always performs an unsigned division to get the
4177 remainder.</p>
4178
Chris Lattner2f2427e2008-01-28 00:36:27 +00004179<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004180 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
4181
Chris Lattner2f2427e2008-01-28 00:36:27 +00004182<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004183
Reid Spencer7eb55b32006-11-02 01:53:59 +00004184<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004185<pre>
4186 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +00004187</pre>
4188
4189</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004190
Reid Spencer7eb55b32006-11-02 01:53:59 +00004191<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004192<h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004193 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004194</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004195
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004196<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004197
Chris Lattner48b383b02003-11-25 01:02:51 +00004198<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004199<pre>
Gabor Greif0f75ad02008-08-07 21:46:00 +00004200 &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 +00004201</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004202
Chris Lattner48b383b02003-11-25 01:02:51 +00004203<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004204<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
4205 division of its two operands. This instruction can also take
4206 <a href="#t_vector">vector</a> versions of the values in which case the
4207 elements must be integers.</p>
Chris Lattnerb8f816e2008-01-04 04:33:49 +00004208
Chris Lattner48b383b02003-11-25 01:02:51 +00004209<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004210<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004211 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4212 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004213
Chris Lattner48b383b02003-11-25 01:02:51 +00004214<h5>Semantics:</h5>
Reid Spencer7eb55b32006-11-02 01:53:59 +00004215<p>This instruction returns the <i>remainder</i> of a division (where the result
Duncan Sands2769c6e2011-03-07 09:12:24 +00004216 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
4217 <i>modulo</i> operator (where the result is either zero or has the same sign
4218 as the divisor, <tt>op2</tt>) of a value.
4219 For more information about the difference,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004220 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
4221 Math Forum</a>. For a table of how this is implemented in various languages,
4222 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
4223 Wikipedia: modulo operation</a>.</p>
4224
Chris Lattner2f2427e2008-01-28 00:36:27 +00004225<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004226 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
4227
Chris Lattner2f2427e2008-01-28 00:36:27 +00004228<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004229 Overflow also leads to undefined behavior; this is a rare case, but can
4230 occur, for example, by taking the remainder of a 32-bit division of
4231 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
4232 lets srem be implemented using instructions that return both the result of
4233 the division and the remainder.)</p>
4234
Chris Lattner48b383b02003-11-25 01:02:51 +00004235<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004236<pre>
4237 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +00004238</pre>
4239
4240</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004241
Reid Spencer7eb55b32006-11-02 01:53:59 +00004242<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004243<h4>
4244 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
4245</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004246
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004247<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004248
Reid Spencer7eb55b32006-11-02 01:53:59 +00004249<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004250<pre>
4251 &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 +00004252</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004253
Reid Spencer7eb55b32006-11-02 01:53:59 +00004254<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004255<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
4256 its two operands.</p>
4257
Reid Spencer7eb55b32006-11-02 01:53:59 +00004258<h5>Arguments:</h5>
4259<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004260 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4261 floating point values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004262
Reid Spencer7eb55b32006-11-02 01:53:59 +00004263<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004264<p>This instruction returns the <i>remainder</i> of a division. The remainder
4265 has the same sign as the dividend.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004266
Reid Spencer7eb55b32006-11-02 01:53:59 +00004267<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004268<pre>
4269 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner48b383b02003-11-25 01:02:51 +00004270</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004271
Misha Brukman76307852003-11-08 01:05:38 +00004272</div>
Robert Bocchino820bc75b2006-02-17 21:18:08 +00004273
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004274</div>
4275
Reid Spencer2ab01932007-02-02 13:57:07 +00004276<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004277<h3>
4278 <a name="bitwiseops">Bitwise Binary Operations</a>
4279</h3>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004280
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004281<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004282
4283<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
4284 program. They are generally very efficient instructions and can commonly be
4285 strength reduced from other instructions. They require two operands of the
4286 same type, execute an operation on them, and produce a single value. The
4287 resulting value is the same type as its operands.</p>
4288
Reid Spencer04e259b2007-01-31 21:39:12 +00004289<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004290<h4>
4291 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
4292</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004293
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004294<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004295
Reid Spencer04e259b2007-01-31 21:39:12 +00004296<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004297<pre>
Chris Lattnera676c0f2011-02-07 16:40:21 +00004298 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4299 &lt;result&gt; = shl nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4300 &lt;result&gt; = shl nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4301 &lt;result&gt; = shl nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer04e259b2007-01-31 21:39:12 +00004302</pre>
Chris Lattnerf0e50112007-10-03 21:01:14 +00004303
Reid Spencer04e259b2007-01-31 21:39:12 +00004304<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004305<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
4306 a specified number of bits.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +00004307
Reid Spencer04e259b2007-01-31 21:39:12 +00004308<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004309<p>Both arguments to the '<tt>shl</tt>' instruction must be the
4310 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
4311 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher455c5772009-12-05 02:46:03 +00004312
Reid Spencer04e259b2007-01-31 21:39:12 +00004313<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004314<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
4315 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
4316 is (statically or dynamically) negative or equal to or larger than the number
4317 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4318 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4319 shift amount in <tt>op2</tt>.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +00004320
Chris Lattnera676c0f2011-02-07 16:40:21 +00004321<p>If the <tt>nuw</tt> keyword is present, then the shift produces a
Dan Gohman9a2a0932011-12-06 03:18:47 +00004322 <a href="#poisonvalues">poison value</a> if it shifts out any non-zero bits. If
Chris Lattnerf10dfdc2011-02-09 16:44:44 +00004323 the <tt>nsw</tt> keyword is present, then the shift produces a
Dan Gohman9a2a0932011-12-06 03:18:47 +00004324 <a href="#poisonvalues">poison value</a> if it shifts out any bits that disagree
Chris Lattnera676c0f2011-02-07 16:40:21 +00004325 with the resultant sign bit. As such, NUW/NSW have the same semantics as
4326 they would if the shift were expressed as a mul instruction with the same
4327 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
4328
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004329<h5>Example:</h5>
4330<pre>
Reid Spencer04e259b2007-01-31 21:39:12 +00004331 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
4332 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
4333 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerf0e50112007-10-03 21:01:14 +00004334 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang4dd832d2008-12-09 05:46:39 +00004335 &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 +00004336</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004337
Reid Spencer04e259b2007-01-31 21:39:12 +00004338</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004339
Reid Spencer04e259b2007-01-31 21:39:12 +00004340<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004341<h4>
4342 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
4343</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004344
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004345<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004346
Reid Spencer04e259b2007-01-31 21:39:12 +00004347<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004348<pre>
Chris Lattnera676c0f2011-02-07 16:40:21 +00004349 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4350 &lt;result&gt; = lshr exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer04e259b2007-01-31 21:39:12 +00004351</pre>
4352
4353<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004354<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4355 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00004356
4357<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004358<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004359 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4360 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00004361
4362<h5>Semantics:</h5>
4363<p>This instruction always performs a logical shift right operation. The most
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004364 significant bits of the result will be filled with zero bits after the shift.
4365 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4366 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4367 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4368 shift amount in <tt>op2</tt>.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00004369
Chris Lattnera676c0f2011-02-07 16:40:21 +00004370<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman9a2a0932011-12-06 03:18:47 +00004371 <tt>lshr</tt> is a <a href="#poisonvalues">poison value</a> if any of the bits
Chris Lattnera676c0f2011-02-07 16:40:21 +00004372 shifted out are non-zero.</p>
4373
4374
Reid Spencer04e259b2007-01-31 21:39:12 +00004375<h5>Example:</h5>
4376<pre>
4377 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4378 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4379 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4380 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerf0e50112007-10-03 21:01:14 +00004381 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang4dd832d2008-12-09 05:46:39 +00004382 &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 +00004383</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004384
Reid Spencer04e259b2007-01-31 21:39:12 +00004385</div>
4386
Reid Spencer2ab01932007-02-02 13:57:07 +00004387<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004388<h4>
4389 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4390</h4>
4391
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004392<div>
Reid Spencer04e259b2007-01-31 21:39:12 +00004393
4394<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004395<pre>
Chris Lattnera676c0f2011-02-07 16:40:21 +00004396 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4397 &lt;result&gt; = ashr exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer04e259b2007-01-31 21:39:12 +00004398</pre>
4399
4400<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004401<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4402 operand shifted to the right a specified number of bits with sign
4403 extension.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00004404
4405<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004406<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004407 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4408 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00004409
4410<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004411<p>This instruction always performs an arithmetic shift right operation, The
4412 most significant bits of the result will be filled with the sign bit
4413 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4414 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4415 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4416 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +00004417
Chris Lattnera676c0f2011-02-07 16:40:21 +00004418<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman9a2a0932011-12-06 03:18:47 +00004419 <tt>ashr</tt> is a <a href="#poisonvalues">poison value</a> if any of the bits
Chris Lattnera676c0f2011-02-07 16:40:21 +00004420 shifted out are non-zero.</p>
4421
Reid Spencer04e259b2007-01-31 21:39:12 +00004422<h5>Example:</h5>
4423<pre>
4424 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4425 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4426 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4427 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerf0e50112007-10-03 21:01:14 +00004428 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang4dd832d2008-12-09 05:46:39 +00004429 &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 +00004430</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004431
Reid Spencer04e259b2007-01-31 21:39:12 +00004432</div>
4433
Chris Lattner2f7c9632001-06-06 20:29:01 +00004434<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004435<h4>
4436 <a name="i_and">'<tt>and</tt>' Instruction</a>
4437</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004438
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004439<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004440
Chris Lattner2f7c9632001-06-06 20:29:01 +00004441<h5>Syntax:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004442<pre>
Gabor Greif0f75ad02008-08-07 21:46:00 +00004443 &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 +00004444</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004445
Chris Lattner2f7c9632001-06-06 20:29:01 +00004446<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004447<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4448 operands.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004449
Chris Lattner2f7c9632001-06-06 20:29:01 +00004450<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004451<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004452 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4453 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004454
Chris Lattner2f7c9632001-06-06 20:29:01 +00004455<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004456<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004457
Misha Brukman76307852003-11-08 01:05:38 +00004458<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +00004459 <tbody>
4460 <tr>
Bill Wendling4517fe52011-12-09 22:41:40 +00004461 <th>In0</th>
4462 <th>In1</th>
4463 <th>Out</th>
Chris Lattner48b383b02003-11-25 01:02:51 +00004464 </tr>
4465 <tr>
4466 <td>0</td>
4467 <td>0</td>
4468 <td>0</td>
4469 </tr>
4470 <tr>
4471 <td>0</td>
4472 <td>1</td>
4473 <td>0</td>
4474 </tr>
4475 <tr>
4476 <td>1</td>
4477 <td>0</td>
4478 <td>0</td>
4479 </tr>
4480 <tr>
4481 <td>1</td>
4482 <td>1</td>
4483 <td>1</td>
4484 </tr>
4485 </tbody>
4486</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004487
Chris Lattner2f7c9632001-06-06 20:29:01 +00004488<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004489<pre>
4490 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004491 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4492 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004493</pre>
Misha Brukman76307852003-11-08 01:05:38 +00004494</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004495<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004496<h4>
4497 <a name="i_or">'<tt>or</tt>' Instruction</a>
4498</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004499
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004500<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004501
4502<h5>Syntax:</h5>
4503<pre>
4504 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4505</pre>
4506
4507<h5>Overview:</h5>
4508<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4509 two operands.</p>
4510
4511<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004512<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004513 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4514 values. Both arguments must have identical types.</p>
4515
Chris Lattner2f7c9632001-06-06 20:29:01 +00004516<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004517<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004518
Chris Lattner48b383b02003-11-25 01:02:51 +00004519<table border="1" cellspacing="0" cellpadding="4">
4520 <tbody>
4521 <tr>
Bill Wendling4517fe52011-12-09 22:41:40 +00004522 <th>In0</th>
4523 <th>In1</th>
4524 <th>Out</th>
Chris Lattner48b383b02003-11-25 01:02:51 +00004525 </tr>
4526 <tr>
4527 <td>0</td>
4528 <td>0</td>
4529 <td>0</td>
4530 </tr>
4531 <tr>
4532 <td>0</td>
4533 <td>1</td>
4534 <td>1</td>
4535 </tr>
4536 <tr>
4537 <td>1</td>
4538 <td>0</td>
4539 <td>1</td>
4540 </tr>
4541 <tr>
4542 <td>1</td>
4543 <td>1</td>
4544 <td>1</td>
4545 </tr>
4546 </tbody>
4547</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004548
Chris Lattner2f7c9632001-06-06 20:29:01 +00004549<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004550<pre>
4551 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004552 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4553 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004554</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004555
Misha Brukman76307852003-11-08 01:05:38 +00004556</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004557
Chris Lattner2f7c9632001-06-06 20:29:01 +00004558<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004559<h4>
4560 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4561</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004562
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004563<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004564
Chris Lattner2f7c9632001-06-06 20:29:01 +00004565<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004566<pre>
4567 &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 +00004568</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004569
Chris Lattner2f7c9632001-06-06 20:29:01 +00004570<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004571<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4572 its two operands. The <tt>xor</tt> is used to implement the "one's
4573 complement" operation, which is the "~" operator in C.</p>
4574
Chris Lattner2f7c9632001-06-06 20:29:01 +00004575<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00004576<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004577 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4578 values. Both arguments must have identical types.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00004579
Chris Lattner2f7c9632001-06-06 20:29:01 +00004580<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004581<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004582
Chris Lattner48b383b02003-11-25 01:02:51 +00004583<table border="1" cellspacing="0" cellpadding="4">
4584 <tbody>
4585 <tr>
Bill Wendling4517fe52011-12-09 22:41:40 +00004586 <th>In0</th>
4587 <th>In1</th>
4588 <th>Out</th>
Chris Lattner48b383b02003-11-25 01:02:51 +00004589 </tr>
4590 <tr>
4591 <td>0</td>
4592 <td>0</td>
4593 <td>0</td>
4594 </tr>
4595 <tr>
4596 <td>0</td>
4597 <td>1</td>
4598 <td>1</td>
4599 </tr>
4600 <tr>
4601 <td>1</td>
4602 <td>0</td>
4603 <td>1</td>
4604 </tr>
4605 <tr>
4606 <td>1</td>
4607 <td>1</td>
4608 <td>0</td>
4609 </tr>
4610 </tbody>
4611</table>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004612
Chris Lattner2f7c9632001-06-06 20:29:01 +00004613<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004614<pre>
4615 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004616 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4617 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4618 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004619</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004620
Misha Brukman76307852003-11-08 01:05:38 +00004621</div>
Chris Lattner54611b42005-11-06 08:02:57 +00004622
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004623</div>
4624
Chris Lattner2f7c9632001-06-06 20:29:01 +00004625<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004626<h3>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004627 <a name="vectorops">Vector Operations</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004628</h3>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004629
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004630<div>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004631
4632<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004633 target-independent manner. These instructions cover the element-access and
4634 vector-specific operations needed to process vectors effectively. While LLVM
4635 does directly support these vector operations, many sophisticated algorithms
4636 will want to use target-specific intrinsics to take full advantage of a
4637 specific target.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004638
Chris Lattnerce83bff2006-04-08 23:07:04 +00004639<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004640<h4>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004641 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004642</h4>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004643
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004644<div>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004645
4646<h5>Syntax:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004647<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004648 &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 +00004649</pre>
4650
4651<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004652<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4653 from a vector at a specified index.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004654
4655
4656<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004657<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4658 of <a href="#t_vector">vector</a> type. The second operand is an index
4659 indicating the position from which to extract the element. The index may be
4660 a variable.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004661
4662<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004663<p>The result is a scalar of the same type as the element type of
4664 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4665 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4666 results are undefined.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004667
4668<h5>Example:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004669<pre>
Gabor Greif03ab4dc2009-10-28 13:14:50 +00004670 &lt;result&gt; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004671</pre>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004672
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004673</div>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004674
4675<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004676<h4>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004677 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004678</h4>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004679
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004680<div>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004681
4682<h5>Syntax:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004683<pre>
Dan Gohman43ba0672008-05-12 23:38:42 +00004684 &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 +00004685</pre>
4686
4687<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004688<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4689 vector at a specified index.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004690
4691<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004692<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4693 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4694 whose type must equal the element type of the first operand. The third
4695 operand is an index indicating the position at which to insert the value.
4696 The index may be a variable.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004697
4698<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004699<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4700 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4701 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4702 results are undefined.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004703
4704<h5>Example:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004705<pre>
Gabor Greif03ab4dc2009-10-28 13:14:50 +00004706 &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 +00004707</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004708
Chris Lattnerce83bff2006-04-08 23:07:04 +00004709</div>
4710
4711<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004712<h4>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004713 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004714</h4>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004715
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004716<div>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004717
4718<h5>Syntax:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004719<pre>
Mon P Wang25f01062008-11-10 04:46:22 +00004720 &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 +00004721</pre>
4722
4723<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004724<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4725 from two input vectors, returning a vector with the same element type as the
4726 input and length that is the same as the shuffle mask.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004727
4728<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004729<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4730 with types that match each other. The third argument is a shuffle mask whose
4731 element type is always 'i32'. The result of the instruction is a vector
4732 whose length is the same as the shuffle mask and whose element type is the
4733 same as the element type of the first two operands.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004734
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004735<p>The shuffle mask operand is required to be a constant vector with either
4736 constant integer or undef values.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004737
4738<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004739<p>The elements of the two input vectors are numbered from left to right across
4740 both of the vectors. The shuffle mask operand specifies, for each element of
4741 the result vector, which element of the two input vectors the result element
4742 gets. The element selector may be undef (meaning "don't care") and the
4743 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004744
4745<h5>Example:</h5>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004746<pre>
Eric Christopher455c5772009-12-05 02:46:03 +00004747 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen5819f182007-04-22 01:17:39 +00004748 &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 +00004749 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004750 &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 +00004751 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wang25f01062008-11-10 04:46:22 +00004752 &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 +00004753 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wang25f01062008-11-10 04:46:22 +00004754 &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 +00004755</pre>
Chris Lattnerce83bff2006-04-08 23:07:04 +00004756
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004757</div>
Tanya Lattnerb138bbe2006-04-14 19:24:33 +00004758
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004759</div>
4760
Chris Lattnerce83bff2006-04-08 23:07:04 +00004761<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004762<h3>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004763 <a name="aggregateops">Aggregate Operations</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004764</h3>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004765
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004766<div>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004767
Chris Lattner392be582010-02-12 20:49:41 +00004768<p>LLVM supports several instructions for working with
4769 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004770
Dan Gohmanb9d66602008-05-12 23:51:09 +00004771<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004772<h4>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004773 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004774</h4>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004775
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004776<div>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004777
4778<h5>Syntax:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004779<pre>
4780 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4781</pre>
4782
4783<h5>Overview:</h5>
Chris Lattner392be582010-02-12 20:49:41 +00004784<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4785 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004786
4787<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004788<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattner13ee7952010-08-28 04:09:24 +00004789 of <a href="#t_struct">struct</a> or
Chris Lattner392be582010-02-12 20:49:41 +00004790 <a href="#t_array">array</a> type. The operands are constant indices to
4791 specify which value to extract in a similar manner as indices in a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004792 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Frits van Bommel7cf63ac2010-12-05 20:54:38 +00004793 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4794 <ul>
4795 <li>Since the value being indexed is not a pointer, the first index is
4796 omitted and assumed to be zero.</li>
4797 <li>At least one index must be specified.</li>
4798 <li>Not only struct indices but also array indices must be in
4799 bounds.</li>
4800 </ul>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004801
4802<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004803<p>The result is the value at the position in the aggregate specified by the
4804 index operands.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004805
4806<h5>Example:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004807<pre>
Gabor Greif03ab4dc2009-10-28 13:14:50 +00004808 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004809</pre>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004810
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004811</div>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004812
4813<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004814<h4>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004815 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004816</h4>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004817
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004818<div>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004819
4820<h5>Syntax:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004821<pre>
Bill Wendlingf6a91cf2011-07-26 20:42:28 +00004822 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, &lt;idx&gt;{, &lt;idx&gt;}* <i>; yields &lt;aggregate type&gt;</i>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004823</pre>
4824
4825<h5>Overview:</h5>
Chris Lattner392be582010-02-12 20:49:41 +00004826<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4827 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004828
4829<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004830<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattner13ee7952010-08-28 04:09:24 +00004831 of <a href="#t_struct">struct</a> or
Chris Lattner392be582010-02-12 20:49:41 +00004832 <a href="#t_array">array</a> type. The second operand is a first-class
4833 value to insert. The following operands are constant indices indicating
4834 the position at which to insert the value in a similar manner as indices in a
Frits van Bommel7cf63ac2010-12-05 20:54:38 +00004835 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004836 value to insert must have the same type as the value identified by the
4837 indices.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004838
4839<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004840<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4841 that of <tt>val</tt> except that the value at the position specified by the
4842 indices is that of <tt>elt</tt>.</p>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004843
4844<h5>Example:</h5>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004845<pre>
Chris Lattnerc2e85402011-05-22 07:18:08 +00004846 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4847 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4848 %agg3 = insertvalue {i32, {float}} %agg1, float %val, 1, 0 <i>; yields {i32 1, float %val}</i>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004849</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004850
Dan Gohmanb9d66602008-05-12 23:51:09 +00004851</div>
4852
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004853</div>
Dan Gohmanb9d66602008-05-12 23:51:09 +00004854
4855<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004856<h3>
Chris Lattner6ab66722006-08-15 00:45:58 +00004857 <a name="memoryops">Memory Access and Addressing Operations</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004858</h3>
Chris Lattner54611b42005-11-06 08:02:57 +00004859
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004860<div>
Chris Lattner54611b42005-11-06 08:02:57 +00004861
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004862<p>A key design point of an SSA-based representation is how it represents
4863 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandeza70c6df2009-10-26 23:44:29 +00004864 very simple. This section describes how to read, write, and allocate
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004865 memory in LLVM.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004866
Chris Lattner2f7c9632001-06-06 20:29:01 +00004867<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004868<h4>
Chris Lattner54611b42005-11-06 08:02:57 +00004869 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004870</h4>
Chris Lattner54611b42005-11-06 08:02:57 +00004871
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004872<div>
Chris Lattner54611b42005-11-06 08:02:57 +00004873
Chris Lattner2f7c9632001-06-06 20:29:01 +00004874<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00004875<pre>
Dan Gohman2140a742010-05-28 01:14:11 +00004876 &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 +00004877</pre>
Chris Lattner54611b42005-11-06 08:02:57 +00004878
Chris Lattner2f7c9632001-06-06 20:29:01 +00004879<h5>Overview:</h5>
Jeff Cohen5819f182007-04-22 01:17:39 +00004880<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004881 currently executing function, to be automatically released when this function
4882 returns to its caller. The object is always allocated in the generic address
4883 space (address space zero).</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004884
Chris Lattner2f7c9632001-06-06 20:29:01 +00004885<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004886<p>The '<tt>alloca</tt>' instruction
4887 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4888 runtime stack, returning a pointer of the appropriate type to the program.
4889 If "NumElements" is specified, it is the number of elements allocated,
4890 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4891 specified, the value result of the allocation is guaranteed to be aligned to
4892 at least that boundary. If not specified, or if zero, the target can choose
4893 to align the allocation on any convenient boundary compatible with the
4894 type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004895
Misha Brukman76307852003-11-08 01:05:38 +00004896<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +00004897
Chris Lattner2f7c9632001-06-06 20:29:01 +00004898<h5>Semantics:</h5>
Bill Wendling9ee6a312009-05-08 20:49:29 +00004899<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004900 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4901 memory is automatically released when the function returns. The
4902 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4903 variables that must have an address available. When the function returns
4904 (either with the <tt><a href="#i_ret">ret</a></tt>
Bill Wendling3f6a3a22012-02-06 21:57:33 +00004905 or <tt><a href="#i_resume">resume</a></tt> instructions), the memory is
Nick Lewyckyefe5e2e2012-02-29 08:26:44 +00004906 reclaimed. Allocating zero bytes is legal, but the result is undefined.
4907 The order in which memory is allocated (ie., which way the stack grows) is
Nick Lewyckyf70a2bd2012-03-18 09:35:50 +00004908 not specified.</p>
Nick Lewyckyefe5e2e2012-02-29 08:26:44 +00004909
4910<p>
Chris Lattner54611b42005-11-06 08:02:57 +00004911
Chris Lattner2f7c9632001-06-06 20:29:01 +00004912<h5>Example:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +00004913<pre>
Dan Gohman7a5acb52009-01-04 23:49:44 +00004914 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4915 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4916 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4917 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +00004918</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004919
Misha Brukman76307852003-11-08 01:05:38 +00004920</div>
Chris Lattner54611b42005-11-06 08:02:57 +00004921
Chris Lattner2f7c9632001-06-06 20:29:01 +00004922<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00004923<h4>
4924 <a name="i_load">'<tt>load</tt>' Instruction</a>
4925</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004926
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00004927<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004928
Chris Lattner095735d2002-05-06 03:03:22 +00004929<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004930<pre>
Pete Cooper13e082d2012-02-10 18:13:54 +00004931 &lt;result&gt; = load [volatile] &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;][, !invariant.load !&lt;index&gt;]
Eli Friedman02e737b2011-08-12 22:50:01 +00004932 &lt;result&gt; = load atomic [volatile] &lt;ty&gt;* &lt;pointer&gt; [singlethread] &lt;ordering&gt;, align &lt;alignment&gt;
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004933 !&lt;index&gt; = !{ i32 1 }
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004934</pre>
4935
Chris Lattner095735d2002-05-06 03:03:22 +00004936<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00004937<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004938
Chris Lattner095735d2002-05-06 03:03:22 +00004939<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004940<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4941 from which to load. The pointer must point to
4942 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4943 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00004944 number or order of execution of this <tt>load</tt> with other <a
4945 href="#volatile">volatile operations</a>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004946
Eli Friedman59b66882011-08-09 23:02:53 +00004947<p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
4948 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4949 argument. The <code>release</code> and <code>acq_rel</code> orderings are
4950 not valid on <code>load</code> instructions. Atomic loads produce <a
4951 href="#memorymodel">defined</a> results when they may see multiple atomic
4952 stores. The type of the pointee must be an integer type whose bit width
4953 is a power of two greater than or equal to eight and less than or equal
4954 to a target-specific size limit. <code>align</code> must be explicitly
4955 specified on atomic loads, and the load has undefined behavior if the
4956 alignment is not set to a value which is at least the size in bytes of
4957 the pointee. <code>!nontemporal</code> does not have any defined semantics
4958 for atomic loads.</p>
4959
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004960<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004961 operation (that is, the alignment of the memory address). A value of 0 or an
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004962 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004963 alignment for the target. It is the responsibility of the code emitter to
4964 ensure that the alignment information is correct. Overestimating the
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004965 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004966 produce less efficient code. An alignment of 1 is always safe.</p>
4967
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004968<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4969 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmana269a0a2010-03-01 17:41:39 +00004970 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendlingbc4024f2010-02-25 21:23:24 +00004971 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4972 and code generator that this load is not expected to be reused in the cache.
4973 The code generator may select special instructions to save cache bandwidth,
Dan Gohmana269a0a2010-03-01 17:41:39 +00004974 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene9641d062010-02-16 20:50:18 +00004975
Pete Cooper13e082d2012-02-10 18:13:54 +00004976<p>The optional <tt>!invariant.load</tt> metadata must reference a single
4977 metatadata name &lt;index&gt; corresponding to a metadata node with no
4978 entries. The existence of the <tt>!invariant.load</tt> metatadata on the
4979 instruction tells the optimizer and code generator that this load address
4980 points to memory which does not change value during program execution.
4981 The optimizer may then move this load around, for example, by hoisting it
4982 out of loops using loop invariant code motion.</p>
4983
Chris Lattner095735d2002-05-06 03:03:22 +00004984<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004985<p>The location of memory pointed to is loaded. If the value being loaded is of
4986 scalar type then the number of bytes read does not exceed the minimum number
4987 of bytes needed to hold all bits of the type. For example, loading an
4988 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4989 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4990 is undefined if the value was not originally written using a store of the
4991 same type.</p>
4992
Chris Lattner095735d2002-05-06 03:03:22 +00004993<h5>Examples:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004994<pre>
4995 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4996 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00004997 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner095735d2002-05-06 03:03:22 +00004998</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00004999
Misha Brukman76307852003-11-08 01:05:38 +00005000</div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005001
Chris Lattner095735d2002-05-06 03:03:22 +00005002<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005003<h4>
5004 <a name="i_store">'<tt>store</tt>' Instruction</a>
5005</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005006
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005007<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005008
Chris Lattner095735d2002-05-06 03:03:22 +00005009<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005010<pre>
Bill Wendling4517fe52011-12-09 22:41:40 +00005011 store [volatile] &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;] <i>; yields {void}</i>
5012 store atomic [volatile] &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; [singlethread] &lt;ordering&gt;, align &lt;alignment&gt; <i>; yields {void}</i>
Chris Lattner095735d2002-05-06 03:03:22 +00005013</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005014
Chris Lattner095735d2002-05-06 03:03:22 +00005015<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00005016<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005017
Chris Lattner095735d2002-05-06 03:03:22 +00005018<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005019<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
5020 and an address at which to store it. The type of the
5021 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
5022 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00005023 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
5024 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
5025 order of execution of this <tt>store</tt> with other <a
5026 href="#volatile">volatile operations</a>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005027
Eli Friedman59b66882011-08-09 23:02:53 +00005028<p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
5029 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
5030 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
5031 valid on <code>store</code> instructions. Atomic loads produce <a
5032 href="#memorymodel">defined</a> results when they may see multiple atomic
5033 stores. The type of the pointee must be an integer type whose bit width
5034 is a power of two greater than or equal to eight and less than or equal
5035 to a target-specific size limit. <code>align</code> must be explicitly
5036 specified on atomic stores, and the store has undefined behavior if the
5037 alignment is not set to a value which is at least the size in bytes of
5038 the pointee. <code>!nontemporal</code> does not have any defined semantics
5039 for atomic stores.</p>
5040
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005041<p>The optional constant "align" argument specifies the alignment of the
5042 operation (that is, the alignment of the memory address). A value of 0 or an
5043 omitted "align" argument means that the operation has the preferential
5044 alignment for the target. It is the responsibility of the code emitter to
5045 ensure that the alignment information is correct. Overestimating the
5046 alignment results in an undefined behavior. Underestimating the alignment may
5047 produce less efficient code. An alignment of 1 is always safe.</p>
5048
David Greene9641d062010-02-16 20:50:18 +00005049<p>The optional !nontemporal metadata must reference a single metatadata
Benjamin Kramer79698be2010-07-13 12:26:09 +00005050 name &lt;index&gt; corresponding to a metadata node with one i32 entry of
Dan Gohmana269a0a2010-03-01 17:41:39 +00005051 value 1. The existence of the !nontemporal metatadata on the
David Greene9641d062010-02-16 20:50:18 +00005052 instruction tells the optimizer and code generator that this load is
5053 not expected to be reused in the cache. The code generator may
5054 select special instructions to save cache bandwidth, such as the
Dan Gohmana269a0a2010-03-01 17:41:39 +00005055 MOVNT instruction on x86.</p>
David Greene9641d062010-02-16 20:50:18 +00005056
5057
Chris Lattner48b383b02003-11-25 01:02:51 +00005058<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005059<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
5060 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
5061 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
5062 does not exceed the minimum number of bytes needed to hold all bits of the
5063 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
5064 writing a value of a type like <tt>i20</tt> with a size that is not an
5065 integral number of bytes, it is unspecified what happens to the extra bits
5066 that do not belong to the type, but they will typically be overwritten.</p>
5067
Chris Lattner095735d2002-05-06 03:03:22 +00005068<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005069<pre>
5070 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8830ffe2007-10-22 05:10:05 +00005071 store i32 3, i32* %ptr <i>; yields {void}</i>
5072 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner095735d2002-05-06 03:03:22 +00005073</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005074
Reid Spencer443460a2006-11-09 21:15:49 +00005075</div>
5076
Chris Lattner095735d2002-05-06 03:03:22 +00005077<!-- _______________________________________________________________________ -->
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00005078<h4>
5079<a name="i_fence">'<tt>fence</tt>' Instruction</a>
5080</h4>
Eli Friedmanfee02c62011-07-25 23:16:38 +00005081
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00005082<div>
Eli Friedmanfee02c62011-07-25 23:16:38 +00005083
5084<h5>Syntax:</h5>
5085<pre>
5086 fence [singlethread] &lt;ordering&gt; <i>; yields {void}</i>
5087</pre>
5088
5089<h5>Overview:</h5>
5090<p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
5091between operations.</p>
5092
5093<h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
5094href="#ordering">ordering</a> argument which defines what
5095<i>synchronizes-with</i> edges they add. They can only be given
5096<code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
5097<code>seq_cst</code> orderings.</p>
5098
5099<h5>Semantics:</h5>
5100<p>A fence <var>A</var> which has (at least) <code>release</code> ordering
5101semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
5102<code>acquire</code> ordering semantics if and only if there exist atomic
5103operations <var>X</var> and <var>Y</var>, both operating on some atomic object
5104<var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
5105<var>X</var> modifies <var>M</var> (either directly or through some side effect
5106of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
5107<var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
5108<i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
5109than an explicit <code>fence</code>, one (but not both) of the atomic operations
5110<var>X</var> or <var>Y</var> might provide a <code>release</code> or
5111<code>acquire</code> (resp.) ordering constraint and still
5112<i>synchronize-with</i> the explicit <code>fence</code> and establish the
5113<i>happens-before</i> edge.</p>
5114
5115<p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
5116having both <code>acquire</code> and <code>release</code> semantics specified
5117above, participates in the global program order of other <code>seq_cst</code>
5118operations and/or fences.</p>
5119
5120<p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
5121specifies that the fence only synchronizes with other fences in the same
5122thread. (This is useful for interacting with signal handlers.)</p>
5123
Eli Friedmanfee02c62011-07-25 23:16:38 +00005124<h5>Example:</h5>
5125<pre>
5126 fence acquire <i>; yields {void}</i>
5127 fence singlethread seq_cst <i>; yields {void}</i>
5128</pre>
5129
5130</div>
5131
5132<!-- _______________________________________________________________________ -->
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00005133<h4>
5134<a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
5135</h4>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005136
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00005137<div>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005138
5139<h5>Syntax:</h5>
5140<pre>
Bill Wendling4517fe52011-12-09 22:41:40 +00005141 cmpxchg [volatile] &lt;ty&gt;* &lt;pointer&gt;, &lt;ty&gt; &lt;cmp&gt;, &lt;ty&gt; &lt;new&gt; [singlethread] &lt;ordering&gt; <i>; yields {ty}</i>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005142</pre>
5143
5144<h5>Overview:</h5>
5145<p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
5146It loads a value in memory and compares it to a given value. If they are
5147equal, it stores a new value into the memory.</p>
5148
5149<h5>Arguments:</h5>
5150<p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
5151address to operate on, a value to compare to the value currently be at that
5152address, and a new value to place at that address if the compared values are
5153equal. The type of '<var>&lt;cmp&gt;</var>' must be an integer type whose
5154bit width is a power of two greater than or equal to eight and less than
5155or equal to a target-specific size limit. '<var>&lt;cmp&gt;</var>' and
5156'<var>&lt;new&gt;</var>' must have the same type, and the type of
5157'<var>&lt;pointer&gt;</var>' must be a pointer to that type. If the
5158<code>cmpxchg</code> is marked as <code>volatile</code>, then the
5159optimizer is not allowed to modify the number or order of execution
5160of this <code>cmpxchg</code> with other <a href="#volatile">volatile
5161operations</a>.</p>
5162
5163<!-- FIXME: Extend allowed types. -->
5164
5165<p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
5166<code>cmpxchg</code> synchronizes with other atomic operations.</p>
5167
5168<p>The optional "<code>singlethread</code>" argument declares that the
5169<code>cmpxchg</code> is only atomic with respect to code (usually signal
5170handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
5171cmpxchg is atomic with respect to all other code in the system.</p>
5172
5173<p>The pointer passed into cmpxchg must have alignment greater than or equal to
5174the size in memory of the operand.
5175
5176<h5>Semantics:</h5>
5177<p>The contents of memory at the location specified by the
5178'<tt>&lt;pointer&gt;</tt>' operand is read and compared to
5179'<tt>&lt;cmp&gt;</tt>'; if the read value is the equal,
5180'<tt>&lt;new&gt;</tt>' is written. The original value at the location
5181is returned.
5182
5183<p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
5184purpose of identifying <a href="#release_sequence">release sequences</a>. A
5185failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
5186parameter determined by dropping any <code>release</code> part of the
5187<code>cmpxchg</code>'s ordering.</p>
5188
5189<!--
5190FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
5191optimization work on ARM.)
5192
5193FIXME: Is a weaker ordering constraint on failure helpful in practice?
5194-->
5195
5196<h5>Example:</h5>
5197<pre>
5198entry:
Bill Wendling4517fe52011-12-09 22:41:40 +00005199 %orig = atomic <a href="#i_load">load</a> i32* %ptr unordered <i>; yields {i32}</i>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005200 <a href="#i_br">br</a> label %loop
5201
5202loop:
5203 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
5204 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
Bill Wendling4517fe52011-12-09 22:41:40 +00005205 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005206 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
5207 <a href="#i_br">br</a> i1 %success, label %done, label %loop
5208
5209done:
5210 ...
5211</pre>
5212
5213</div>
5214
5215<!-- _______________________________________________________________________ -->
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00005216<h4>
5217<a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
5218</h4>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005219
NAKAMURA Takumi0300d882011-08-12 06:17:17 +00005220<div>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005221
5222<h5>Syntax:</h5>
5223<pre>
Eli Friedman02e737b2011-08-12 22:50:01 +00005224 atomicrmw [volatile] &lt;operation&gt; &lt;ty&gt;* &lt;pointer&gt;, &lt;ty&gt; &lt;value&gt; [singlethread] &lt;ordering&gt; <i>; yields {ty}</i>
Eli Friedmanc9a551e2011-07-28 21:48:00 +00005225</pre>
5226
5227<h5>Overview:</h5>
5228<p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
5229
5230<h5>Arguments:</h5>
5231<p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
5232operation to apply, an address whose value to modify, an argument to the
5233operation. The operation must be one of the following keywords:</p>
5234<ul>
5235 <li>xchg</li>
5236 <li>add</li>
5237 <li>sub</li>
5238 <li>and</li>
5239 <li>nand</li>
5240 <li>or</li>
5241 <li>xor</li>
5242 <li>max</li>
5243 <li>min</li>
5244 <li>umax</li>
5245 <li>umin</li>
5246</ul>
5247
5248<p>The type of '<var>&lt;value&gt;</var>' must be an integer type whose
5249bit width is a power of two greater than or equal to eight and less than
5250or equal to a target-specific size limit. The type of the
5251'<code>&lt;pointer&gt;</code>' operand must be a pointer to that type.
5252If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
5253optimizer is not allowed to modify the number or order of execution of this
5254<code>atomicrmw</code> with other <a href="#volatile">volatile
5255 operations</a>.</p>
5256
5257<!-- FIXME: Extend allowed types. -->
5258
5259<h5>Semantics:</h5>
5260<p>The contents of memory at the location specified by the
5261'<tt>&lt;pointer&gt;</tt>' operand are atomically read, modified, and written
5262back. The original value at the location is returned. The modification is
5263specified by the <var>operation</var> argument:</p>
5264
5265<ul>
5266 <li>xchg: <code>*ptr = val</code></li>
5267 <li>add: <code>*ptr = *ptr + val</code></li>
5268 <li>sub: <code>*ptr = *ptr - val</code></li>
5269 <li>and: <code>*ptr = *ptr &amp; val</code></li>
5270 <li>nand: <code>*ptr = ~(*ptr &amp; val)</code></li>
5271 <li>or: <code>*ptr = *ptr | val</code></li>
5272 <li>xor: <code>*ptr = *ptr ^ val</code></li>
5273 <li>max: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using a signed comparison)</li>
5274 <li>min: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using a signed comparison)</li>
5275 <li>umax: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using an unsigned comparison)</li>
5276 <li>umin: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using an unsigned comparison)</li>
5277</ul>
5278
5279<h5>Example:</h5>
5280<pre>
5281 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
5282</pre>
5283
5284</div>
5285
5286<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005287<h4>
Chris Lattner33fd7022004-04-05 01:30:49 +00005288 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005289</h4>
Chris Lattner33fd7022004-04-05 01:30:49 +00005290
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005291<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005292
Chris Lattner590645f2002-04-14 06:13:44 +00005293<h5>Syntax:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00005294<pre>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00005295 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohman1639c392009-07-27 21:53:46 +00005296 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Nadav Rotem3924cb02011-12-05 06:29:09 +00005297 &lt;result&gt; = getelementptr &lt;ptr vector&gt; ptrval, &lt;vector index type&gt; idx
Chris Lattner33fd7022004-04-05 01:30:49 +00005298</pre>
5299
Chris Lattner590645f2002-04-14 06:13:44 +00005300<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005301<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattner392be582010-02-12 20:49:41 +00005302 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
5303 It performs address calculation only and does not access memory.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00005304
Chris Lattner590645f2002-04-14 06:13:44 +00005305<h5>Arguments:</h5>
Nadav Rotem3924cb02011-12-05 06:29:09 +00005306<p>The first argument is always a pointer or a vector of pointers,
5307 and forms the basis of the
Chris Lattnera40b9122009-07-29 06:44:13 +00005308 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005309 elements of the aggregate object are indexed. The interpretation of each
5310 index is dependent on the type being indexed into. The first index always
5311 indexes the pointer value given as the first argument, the second index
5312 indexes a value of the type pointed to (not necessarily the value directly
5313 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattner392be582010-02-12 20:49:41 +00005314 indexed into must be a pointer value, subsequent types can be arrays,
Chris Lattner13ee7952010-08-28 04:09:24 +00005315 vectors, and structs. Note that subsequent types being indexed into
Chris Lattner392be582010-02-12 20:49:41 +00005316 can never be pointers, since that would require loading the pointer before
5317 continuing calculation.</p>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00005318
5319<p>The type of each index argument depends on the type it is indexing into.
Chris Lattner13ee7952010-08-28 04:09:24 +00005320 When indexing into a (optionally packed) structure, only <tt>i32</tt>
Chris Lattner392be582010-02-12 20:49:41 +00005321 integer <b>constants</b> are allowed. When indexing into an array, pointer
5322 or vector, integers of any width are allowed, and they are not required to be
Eli Friedmand8874dc2011-08-12 23:37:55 +00005323 constant. These integers are treated as signed values where relevant.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00005324
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005325<p>For example, let's consider a C code fragment and how it gets compiled to
5326 LLVM:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00005327
Benjamin Kramer79698be2010-07-13 12:26:09 +00005328<pre class="doc_code">
Bill Wendling3716c5d2007-05-29 09:04:49 +00005329struct RT {
5330 char A;
Chris Lattnera446f1b2007-05-29 15:43:56 +00005331 int B[10][20];
Bill Wendling3716c5d2007-05-29 09:04:49 +00005332 char C;
5333};
5334struct ST {
Chris Lattnera446f1b2007-05-29 15:43:56 +00005335 int X;
Bill Wendling3716c5d2007-05-29 09:04:49 +00005336 double Y;
5337 struct RT Z;
5338};
Chris Lattner33fd7022004-04-05 01:30:49 +00005339
Chris Lattnera446f1b2007-05-29 15:43:56 +00005340int *foo(struct ST *s) {
Bill Wendling3716c5d2007-05-29 09:04:49 +00005341 return &amp;s[1].Z.B[5][13];
5342}
Chris Lattner33fd7022004-04-05 01:30:49 +00005343</pre>
5344
Bill Wendling7ad1f362011-12-13 01:07:07 +00005345<p>The LLVM code generated by Clang is:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00005346
Benjamin Kramer79698be2010-07-13 12:26:09 +00005347<pre class="doc_code">
Bill Wendling7ad1f362011-12-13 01:07:07 +00005348%struct.RT = <a href="#namedtypes">type</a> { i8, [10 x [20 x i32]], i8 }
5349%struct.ST = <a href="#namedtypes">type</a> { i32, double, %struct.RT }
Chris Lattner33fd7022004-04-05 01:30:49 +00005350
Bill Wendling7ad1f362011-12-13 01:07:07 +00005351define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
Bill Wendling3716c5d2007-05-29 09:04:49 +00005352entry:
Bill Wendling7ad1f362011-12-13 01:07:07 +00005353 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5354 ret i32* %arrayidx
Bill Wendling3716c5d2007-05-29 09:04:49 +00005355}
Chris Lattner33fd7022004-04-05 01:30:49 +00005356</pre>
5357
Chris Lattner590645f2002-04-14 06:13:44 +00005358<h5>Semantics:</h5>
Bill Wendling7ad1f362011-12-13 01:07:07 +00005359<p>In the example above, the first index is indexing into the
5360 '<tt>%struct.ST*</tt>' type, which is a pointer, yielding a
5361 '<tt>%struct.ST</tt>' = '<tt>{ i32, double, %struct.RT }</tt>' type, a
5362 structure. The second index indexes into the third element of the structure,
5363 yielding a '<tt>%struct.RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]], i8 }</tt>'
5364 type, another structure. The third index indexes into the second element of
5365 the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an array. The
5366 two dimensions of the array are subscripted into, yielding an '<tt>i32</tt>'
5367 type. The '<tt>getelementptr</tt>' instruction returns a pointer to this
5368 element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00005369
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005370<p>Note that it is perfectly legal to index partially through a structure,
5371 returning a pointer to an inner element. Because of this, the LLVM code for
5372 the given testcase is equivalent to:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +00005373
Bill Wendling7ad1f362011-12-13 01:07:07 +00005374<pre class="doc_code">
5375define i32* @foo(%struct.ST* %s) {
5376 %t1 = getelementptr %struct.ST* %s, i32 1 <i>; yields %struct.ST*:%t1</i>
5377 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 <i>; yields %struct.RT*:%t2</i>
5378 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
5379 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5380 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5381 ret i32* %t5
5382}
Chris Lattnera8292f32002-05-06 22:08:29 +00005383</pre>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +00005384
Dan Gohman1639c392009-07-27 21:53:46 +00005385<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman9a2a0932011-12-06 03:18:47 +00005386 <tt>getelementptr</tt> is a <a href="#poisonvalues">poison value</a> if the
Dan Gohman57255802010-04-23 15:23:32 +00005387 base pointer is not an <i>in bounds</i> address of an allocated object,
5388 or if any of the addresses that would be formed by successive addition of
5389 the offsets implied by the indices to the base address with infinitely
Eli Friedmand8874dc2011-08-12 23:37:55 +00005390 precise signed arithmetic are not an <i>in bounds</i> address of that
5391 allocated object. The <i>in bounds</i> addresses for an allocated object
5392 are all the addresses that point into the object, plus the address one
Nadav Rotem3924cb02011-12-05 06:29:09 +00005393 byte past the end.
5394 In cases where the base is a vector of pointers the <tt>inbounds</tt> keyword
5395 applies to each of the computations element-wise. </p>
Dan Gohman1639c392009-07-27 21:53:46 +00005396
5397<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
Eli Friedmand8874dc2011-08-12 23:37:55 +00005398 the base address with silently-wrapping two's complement arithmetic. If the
5399 offsets have a different width from the pointer, they are sign-extended or
5400 truncated to the width of the pointer. The result value of the
5401 <tt>getelementptr</tt> may be outside the object pointed to by the base
5402 pointer. The result value may not necessarily be used to access memory
5403 though, even if it happens to point into allocated storage. See the
5404 <a href="#pointeraliasing">Pointer Aliasing Rules</a> section for more
5405 information.</p>
Dan Gohman1639c392009-07-27 21:53:46 +00005406
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005407<p>The getelementptr instruction is often confusing. For some more insight into
5408 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner6ab66722006-08-15 00:45:58 +00005409
Chris Lattner590645f2002-04-14 06:13:44 +00005410<h5>Example:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +00005411<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005412 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman0e268272008-10-13 13:44:15 +00005413 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5414 <i>; yields i8*:vptr</i>
Dan Gohmanef9462f2008-10-14 16:51:45 +00005415 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijman0e268272008-10-13 13:44:15 +00005416 <i>; yields i8*:eptr</i>
5417 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta0c155e62009-04-25 07:27:44 +00005418 <i>; yields i32*:iptr</i>
Sanjiv Gupta77abea02009-04-24 16:38:13 +00005419 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattner33fd7022004-04-05 01:30:49 +00005420</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005421
Nadav Rotem3924cb02011-12-05 06:29:09 +00005422<p>In cases where the pointer argument is a vector of pointers, only a
5423 single index may be used, and the number of vector elements has to be
5424 the same. For example: </p>
5425<pre class="doc_code">
5426 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
5427</pre>
5428
Chris Lattner33fd7022004-04-05 01:30:49 +00005429</div>
Reid Spencer443460a2006-11-09 21:15:49 +00005430
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005431</div>
5432
Chris Lattner2f7c9632001-06-06 20:29:01 +00005433<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005434<h3>
5435 <a name="convertops">Conversion Operations</a>
5436</h3>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005437
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005438<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005439
Reid Spencer97c5fa42006-11-08 01:18:52 +00005440<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005441 which all take a single operand and a type. They perform various bit
5442 conversions on the operand.</p>
5443
Chris Lattnera8292f32002-05-06 22:08:29 +00005444<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005445<h4>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005446 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005447</h4>
5448
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005449<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005450
5451<h5>Syntax:</h5>
5452<pre>
5453 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5454</pre>
5455
5456<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005457<p>The '<tt>trunc</tt>' instruction truncates its operand to the
5458 type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005459
5460<h5>Arguments:</h5>
Nadav Rotem502f1b92011-02-24 21:01:34 +00005461<p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5462 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5463 of the same number of integers.
5464 The bit size of the <tt>value</tt> must be larger than
5465 the bit size of the destination type, <tt>ty2</tt>.
5466 Equal sized types are not allowed.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005467
5468<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005469<p>The '<tt>trunc</tt>' instruction truncates the high order bits
5470 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5471 source size must be larger than the destination size, <tt>trunc</tt> cannot
5472 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005473
5474<h5>Example:</h5>
5475<pre>
Nadav Rotem502f1b92011-02-24 21:01:34 +00005476 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5477 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5478 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5479 %W = trunc &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i8&gt; <i>; yields &lt;i8 8, i8 7&gt;</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005480</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005481
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005482</div>
5483
5484<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005485<h4>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005486 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005487</h4>
5488
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005489<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005490
5491<h5>Syntax:</h5>
5492<pre>
5493 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5494</pre>
5495
5496<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00005497<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005498 <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005499
5500
5501<h5>Arguments:</h5>
Nadav Rotem25f2ac92011-02-20 12:37:50 +00005502<p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5503 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5504 of the same number of integers.
5505 The bit size of the <tt>value</tt> must be smaller than
5506 the bit size of the destination type,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005507 <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005508
5509<h5>Semantics:</h5>
5510<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005511 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005512
Reid Spencer07c9c682007-01-12 15:46:11 +00005513<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005514
5515<h5>Example:</h5>
5516<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005517 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencer36a15422007-01-12 03:35:51 +00005518 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Nadav Rotem25f2ac92011-02-20 12:37:50 +00005519 %Z = zext &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i32&gt; <i>; yields &lt;i32 8, i32 7&gt;</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005520</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005521
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005522</div>
5523
5524<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005525<h4>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005526 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005527</h4>
5528
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005529<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005530
5531<h5>Syntax:</h5>
5532<pre>
5533 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5534</pre>
5535
5536<h5>Overview:</h5>
5537<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5538
5539<h5>Arguments:</h5>
Nadav Rotem502f1b92011-02-24 21:01:34 +00005540<p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5541 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5542 of the same number of integers.
5543 The bit size of the <tt>value</tt> must be smaller than
5544 the bit size of the destination type,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005545 <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005546
5547<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005548<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5549 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5550 of the type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005551
Reid Spencer36a15422007-01-12 03:35:51 +00005552<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005553
5554<h5>Example:</h5>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005555<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005556 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencer36a15422007-01-12 03:35:51 +00005557 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Nadav Rotem502f1b92011-02-24 21:01:34 +00005558 %Z = sext &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i32&gt; <i>; yields &lt;i32 8, i32 7&gt;</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005559</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005560
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005561</div>
5562
5563<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005564<h4>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005565 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005566</h4>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005567
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005568<div>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005569
5570<h5>Syntax:</h5>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005571<pre>
5572 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5573</pre>
5574
5575<h5>Overview:</h5>
5576<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005577 <tt>ty2</tt>.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005578
5579<h5>Arguments:</h5>
5580<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005581 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5582 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christopher455c5772009-12-05 02:46:03 +00005583 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005584 <i>no-op cast</i>.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005585
5586<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005587<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher455c5772009-12-05 02:46:03 +00005588 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005589 <a href="#t_floating">floating point</a> type. If the value cannot fit
5590 within the destination type, <tt>ty2</tt>, then the results are
5591 undefined.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +00005592
5593<h5>Example:</h5>
5594<pre>
5595 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5596 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5597</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005598
Reid Spencer2e2740d2006-11-09 21:48:10 +00005599</div>
5600
5601<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005602<h4>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005603 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005604</h4>
5605
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005606<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005607
5608<h5>Syntax:</h5>
5609<pre>
5610 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5611</pre>
5612
5613<h5>Overview:</h5>
5614<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005615 floating point value.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005616
5617<h5>Arguments:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00005618<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005619 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5620 a <a href="#t_floating">floating point</a> type to cast it to. The source
5621 type must be smaller than the destination type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005622
5623<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +00005624<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005625 <a href="#t_floating">floating point</a> type to a larger
5626 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5627 used to make a <i>no-op cast</i> because it always changes bits. Use
5628 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005629
5630<h5>Example:</h5>
5631<pre>
Nick Lewycky9feca672011-03-31 18:20:19 +00005632 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5633 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005634</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005635
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005636</div>
5637
5638<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005639<h4>
Reid Spencer2eadb532007-01-21 00:29:26 +00005640 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005641</h4>
5642
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005643<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005644
5645<h5>Syntax:</h5>
5646<pre>
Reid Spencer753163d2007-07-31 14:40:14 +00005647 &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 +00005648</pre>
5649
5650<h5>Overview:</h5>
Reid Spencer753163d2007-07-31 14:40:14 +00005651<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005652 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005653
5654<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005655<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5656 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5657 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5658 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5659 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005660
5661<h5>Semantics:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00005662<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005663 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5664 towards zero) unsigned integer value. If the value cannot fit
5665 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005666
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005667<h5>Example:</h5>
5668<pre>
Reid Spencer753163d2007-07-31 14:40:14 +00005669 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner5b95a172007-09-22 03:17:52 +00005670 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00005671 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005672</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005673
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005674</div>
5675
5676<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005677<h4>
Reid Spencer51b07252006-11-09 23:03:26 +00005678 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005679</h4>
5680
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005681<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005682
5683<h5>Syntax:</h5>
5684<pre>
Reid Spencer51b07252006-11-09 23:03:26 +00005685 &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 +00005686</pre>
5687
5688<h5>Overview:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00005689<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005690 <a href="#t_floating">floating point</a> <tt>value</tt> to
5691 type <tt>ty2</tt>.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005692
Chris Lattnera8292f32002-05-06 22:08:29 +00005693<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005694<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5695 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5696 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5697 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5698 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005699
Chris Lattnera8292f32002-05-06 22:08:29 +00005700<h5>Semantics:</h5>
Eric Christopher455c5772009-12-05 02:46:03 +00005701<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005702 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5703 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5704 the results are undefined.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005705
Chris Lattner70de6632001-07-09 00:26:23 +00005706<h5>Example:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005707<pre>
Reid Spencer36a15422007-01-12 03:35:51 +00005708 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner5b95a172007-09-22 03:17:52 +00005709 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00005710 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005711</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005712
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005713</div>
5714
5715<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005716<h4>
Reid Spencer51b07252006-11-09 23:03:26 +00005717 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005718</h4>
5719
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005720<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005721
5722<h5>Syntax:</h5>
5723<pre>
Reid Spencer51b07252006-11-09 23:03:26 +00005724 &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 +00005725</pre>
5726
5727<h5>Overview:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +00005728<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005729 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005730
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005731<h5>Arguments:</h5>
Nate Begemand4d45c22007-11-17 03:58:34 +00005732<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005733 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5734 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5735 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5736 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005737
5738<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +00005739<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005740 integer quantity and converts it to the corresponding floating point
5741 value. If the value cannot fit in the floating point value, the results are
5742 undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005743
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005744<h5>Example:</h5>
5745<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005746 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohmanef9462f2008-10-14 16:51:45 +00005747 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005748</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005749
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005750</div>
5751
5752<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005753<h4>
Reid Spencer51b07252006-11-09 23:03:26 +00005754 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005755</h4>
5756
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005757<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005758
5759<h5>Syntax:</h5>
5760<pre>
Reid Spencer51b07252006-11-09 23:03:26 +00005761 &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 +00005762</pre>
5763
5764<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005765<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5766 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005767
5768<h5>Arguments:</h5>
Nate Begemand4d45c22007-11-17 03:58:34 +00005769<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005770 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5771 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5772 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5773 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005774
5775<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005776<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5777 quantity and converts it to the corresponding floating point value. If the
5778 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005779
5780<h5>Example:</h5>
5781<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005782 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohmanef9462f2008-10-14 16:51:45 +00005783 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005784</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005785
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005786</div>
5787
5788<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005789<h4>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005790 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005791</h4>
5792
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005793<div>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005794
5795<h5>Syntax:</h5>
5796<pre>
5797 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5798</pre>
5799
5800<h5>Overview:</h5>
Nadav Rotem3924cb02011-12-05 06:29:09 +00005801<p>The '<tt>ptrtoint</tt>' instruction converts the pointer or a vector of
5802 pointers <tt>value</tt> to
5803 the integer (or vector of integers) type <tt>ty2</tt>.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005804
5805<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005806<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
Nadav Rotem3924cb02011-12-05 06:29:09 +00005807 must be a a value of type <a href="#t_pointer">pointer</a> or a vector of
5808 pointers, and a type to cast it to
5809 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> or a vector
5810 of integers type.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005811
5812<h5>Semantics:</h5>
5813<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005814 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5815 truncating or zero extending that value to the size of the integer type. If
5816 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5817 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5818 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5819 change.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005820
5821<h5>Example:</h5>
5822<pre>
Nadav Rotem3924cb02011-12-05 06:29:09 +00005823 %X = ptrtoint i32* %P to i8 <i>; yields truncation on 32-bit architecture</i>
5824 %Y = ptrtoint i32* %P to i64 <i>; yields zero extension on 32-bit architecture</i>
5825 %Z = ptrtoint &lt;4 x i32*&gt; %P to &lt;4 x i64&gt;<i>; yields vector zero extension for a vector of addresses on 32-bit architecture</i>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005826</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005827
Reid Spencerb7344ff2006-11-11 21:00:47 +00005828</div>
5829
5830<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005831<h4>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005832 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005833</h4>
5834
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005835<div>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005836
5837<h5>Syntax:</h5>
5838<pre>
5839 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5840</pre>
5841
5842<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005843<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5844 pointer type, <tt>ty2</tt>.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005845
5846<h5>Arguments:</h5>
Duncan Sands16f122e2007-03-30 12:22:09 +00005847<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005848 value to cast, and a type to cast it to, which must be a
5849 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005850
5851<h5>Semantics:</h5>
5852<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005853 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5854 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5855 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5856 than the size of a pointer then a zero extension is done. If they are the
5857 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005858
5859<h5>Example:</h5>
5860<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005861 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greiff50fd572009-10-28 09:21:30 +00005862 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5863 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Nadav Rotem3924cb02011-12-05 06:29:09 +00005864 %Z = inttoptr &lt;4 x i32&gt; %G to &lt;4 x i8*&gt;<i>; yields truncation of vector G to four pointers</i>
Reid Spencerb7344ff2006-11-11 21:00:47 +00005865</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005866
Reid Spencerb7344ff2006-11-11 21:00:47 +00005867</div>
5868
5869<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005870<h4>
Reid Spencer5b950642006-11-11 23:08:07 +00005871 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005872</h4>
5873
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005874<div>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005875
5876<h5>Syntax:</h5>
5877<pre>
Reid Spencer5b950642006-11-11 23:08:07 +00005878 &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 +00005879</pre>
5880
5881<h5>Overview:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +00005882<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005883 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005884
5885<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005886<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5887 non-aggregate first class value, and a type to cast it to, which must also be
5888 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5889 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5890 identical. If the source type is a pointer, the destination type must also be
5891 a pointer. This instruction supports bitwise conversion of vectors to
5892 integers and to vectors of other types (as long as they have the same
5893 size).</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005894
5895<h5>Semantics:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +00005896<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005897 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5898 this conversion. The conversion is done as if the <tt>value</tt> had been
Nadav Rotem3924cb02011-12-05 06:29:09 +00005899 stored to memory and read back as type <tt>ty2</tt>.
5900 Pointer (or vector of pointers) types may only be converted to other pointer
5901 (or vector of pointers) types with this instruction. To convert
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005902 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5903 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +00005904
5905<h5>Example:</h5>
5906<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +00005907 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00005908 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Nadav Rotem3924cb02011-12-05 06:29:09 +00005909 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
5910 %Z = bitcast &lt;2 x i32*&gt; %V to &lt;2 x i64*&gt; <i>; yields &lt;2 x i64*&gt;</i>
Chris Lattner70de6632001-07-09 00:26:23 +00005911</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005912
Misha Brukman76307852003-11-08 01:05:38 +00005913</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00005914
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005915</div>
5916
Reid Spencer97c5fa42006-11-08 01:18:52 +00005917<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005918<h3>
5919 <a name="otherops">Other Operations</a>
5920</h3>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005921
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005922<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005923
5924<p>The instructions in this category are the "miscellaneous" instructions, which
5925 defy better classification.</p>
5926
Reid Spencerc828a0e2006-11-18 21:50:54 +00005927<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00005928<h4>
5929 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5930</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005931
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00005932<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005933
Reid Spencerc828a0e2006-11-18 21:50:54 +00005934<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005935<pre>
5936 &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 +00005937</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005938
Reid Spencerc828a0e2006-11-18 21:50:54 +00005939<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005940<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
Nadav Rotem3924cb02011-12-05 06:29:09 +00005941 boolean values based on comparison of its two integer, integer vector,
5942 pointer, or pointer vector operands.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005943
Reid Spencerc828a0e2006-11-18 21:50:54 +00005944<h5>Arguments:</h5>
5945<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005946 the condition code indicating the kind of comparison to perform. It is not a
5947 value, just a keyword. The possible condition code are:</p>
5948
Reid Spencerc828a0e2006-11-18 21:50:54 +00005949<ol>
5950 <li><tt>eq</tt>: equal</li>
5951 <li><tt>ne</tt>: not equal </li>
5952 <li><tt>ugt</tt>: unsigned greater than</li>
5953 <li><tt>uge</tt>: unsigned greater or equal</li>
5954 <li><tt>ult</tt>: unsigned less than</li>
5955 <li><tt>ule</tt>: unsigned less or equal</li>
5956 <li><tt>sgt</tt>: signed greater than</li>
5957 <li><tt>sge</tt>: signed greater or equal</li>
5958 <li><tt>slt</tt>: signed less than</li>
5959 <li><tt>sle</tt>: signed less or equal</li>
5960</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005961
Chris Lattnerc0f423a2007-01-15 01:54:13 +00005962<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005963 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5964 typed. They must also be identical types.</p>
5965
Reid Spencerc828a0e2006-11-18 21:50:54 +00005966<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005967<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5968 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00005969 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005970 result, as follows:</p>
5971
Reid Spencerc828a0e2006-11-18 21:50:54 +00005972<ol>
Eric Christopher455c5772009-12-05 02:46:03 +00005973 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005974 <tt>false</tt> otherwise. No sign interpretation is necessary or
5975 performed.</li>
5976
Eric Christopher455c5772009-12-05 02:46:03 +00005977 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005978 <tt>false</tt> otherwise. No sign interpretation is necessary or
5979 performed.</li>
5980
Reid Spencerc828a0e2006-11-18 21:50:54 +00005981 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005982 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5983
Reid Spencerc828a0e2006-11-18 21:50:54 +00005984 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005985 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5986 to <tt>op2</tt>.</li>
5987
Reid Spencerc828a0e2006-11-18 21:50:54 +00005988 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005989 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5990
Reid Spencerc828a0e2006-11-18 21:50:54 +00005991 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005992 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5993
Reid Spencerc828a0e2006-11-18 21:50:54 +00005994 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005995 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5996
Reid Spencerc828a0e2006-11-18 21:50:54 +00005997 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00005998 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5999 to <tt>op2</tt>.</li>
6000
Reid Spencerc828a0e2006-11-18 21:50:54 +00006001 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006002 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
6003
Reid Spencerc828a0e2006-11-18 21:50:54 +00006004 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006005 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006006</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006007
Reid Spencerc828a0e2006-11-18 21:50:54 +00006008<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006009 values are compared as if they were integers.</p>
6010
6011<p>If the operands are integer vectors, then they are compared element by
6012 element. The result is an <tt>i1</tt> vector with the same number of elements
6013 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006014
6015<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006016<pre>
6017 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00006018 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
6019 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
6020 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
6021 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
6022 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006023</pre>
Dan Gohmana5127ab2009-01-22 01:39:38 +00006024
6025<p>Note that the code generator does not yet support vector types with
6026 the <tt>icmp</tt> instruction.</p>
6027
Reid Spencerc828a0e2006-11-18 21:50:54 +00006028</div>
6029
6030<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006031<h4>
6032 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
6033</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006034
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006035<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006036
Reid Spencerc828a0e2006-11-18 21:50:54 +00006037<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006038<pre>
6039 &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 +00006040</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006041
Reid Spencerc828a0e2006-11-18 21:50:54 +00006042<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006043<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
6044 values based on comparison of its operands.</p>
6045
6046<p>If the operands are floating point scalars, then the result type is a boolean
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00006047(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006048
6049<p>If the operands are floating point vectors, then the result type is a vector
6050 of boolean with the same number of elements as the operands being
6051 compared.</p>
6052
Reid Spencerc828a0e2006-11-18 21:50:54 +00006053<h5>Arguments:</h5>
6054<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006055 the condition code indicating the kind of comparison to perform. It is not a
6056 value, just a keyword. The possible condition code are:</p>
6057
Reid Spencerc828a0e2006-11-18 21:50:54 +00006058<ol>
Reid Spencerf69acf32006-11-19 03:00:14 +00006059 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006060 <li><tt>oeq</tt>: ordered and equal</li>
6061 <li><tt>ogt</tt>: ordered and greater than </li>
6062 <li><tt>oge</tt>: ordered and greater than or equal</li>
6063 <li><tt>olt</tt>: ordered and less than </li>
6064 <li><tt>ole</tt>: ordered and less than or equal</li>
6065 <li><tt>one</tt>: ordered and not equal</li>
6066 <li><tt>ord</tt>: ordered (no nans)</li>
6067 <li><tt>ueq</tt>: unordered or equal</li>
6068 <li><tt>ugt</tt>: unordered or greater than </li>
6069 <li><tt>uge</tt>: unordered or greater than or equal</li>
6070 <li><tt>ult</tt>: unordered or less than </li>
6071 <li><tt>ule</tt>: unordered or less than or equal</li>
6072 <li><tt>une</tt>: unordered or not equal</li>
6073 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerf69acf32006-11-19 03:00:14 +00006074 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006075</ol>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006076
Jeff Cohen222a8a42007-04-29 01:07:00 +00006077<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006078 <i>unordered</i> means that either operand may be a QNAN.</p>
6079
6080<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
6081 a <a href="#t_floating">floating point</a> type or
6082 a <a href="#t_vector">vector</a> of floating point type. They must have
6083 identical types.</p>
6084
Reid Spencerc828a0e2006-11-18 21:50:54 +00006085<h5>Semantics:</h5>
Gabor Greif0f75ad02008-08-07 21:46:00 +00006086<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006087 according to the condition code given as <tt>cond</tt>. If the operands are
6088 vectors, then the vectors are compared element by element. Each comparison
Nick Lewycky84a1eeb2009-09-27 00:45:11 +00006089 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006090 follows:</p>
6091
Reid Spencerc828a0e2006-11-18 21:50:54 +00006092<ol>
6093 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006094
Eric Christopher455c5772009-12-05 02:46:03 +00006095 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006096 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
6097
Reid Spencerf69acf32006-11-19 03:00:14 +00006098 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Dan Gohmana269a0a2010-03-01 17:41:39 +00006099 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006100
Eric Christopher455c5772009-12-05 02:46:03 +00006101 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006102 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
6103
Eric Christopher455c5772009-12-05 02:46:03 +00006104 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006105 <tt>op1</tt> is less than <tt>op2</tt>.</li>
6106
Eric Christopher455c5772009-12-05 02:46:03 +00006107 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006108 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
6109
Eric Christopher455c5772009-12-05 02:46:03 +00006110 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006111 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
6112
Reid Spencerf69acf32006-11-19 03:00:14 +00006113 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006114
Eric Christopher455c5772009-12-05 02:46:03 +00006115 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006116 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
6117
Eric Christopher455c5772009-12-05 02:46:03 +00006118 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006119 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
6120
Eric Christopher455c5772009-12-05 02:46:03 +00006121 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006122 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
6123
Eric Christopher455c5772009-12-05 02:46:03 +00006124 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006125 <tt>op1</tt> is less than <tt>op2</tt>.</li>
6126
Eric Christopher455c5772009-12-05 02:46:03 +00006127 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006128 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
6129
Eric Christopher455c5772009-12-05 02:46:03 +00006130 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006131 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
6132
Reid Spencerf69acf32006-11-19 03:00:14 +00006133 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006134
Reid Spencerc828a0e2006-11-18 21:50:54 +00006135 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
6136</ol>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006137
6138<h5>Example:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006139<pre>
6140 &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanc579d972008-09-09 01:02:47 +00006141 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
6142 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
6143 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +00006144</pre>
Dan Gohmana5127ab2009-01-22 01:39:38 +00006145
6146<p>Note that the code generator does not yet support vector types with
6147 the <tt>fcmp</tt> instruction.</p>
6148
Reid Spencerc828a0e2006-11-18 21:50:54 +00006149</div>
6150
Reid Spencer97c5fa42006-11-08 01:18:52 +00006151<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006152<h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006153 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006154</h4>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006155
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006156<div>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006157
Reid Spencer97c5fa42006-11-08 01:18:52 +00006158<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006159<pre>
6160 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
6161</pre>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006162
Reid Spencer97c5fa42006-11-08 01:18:52 +00006163<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006164<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
6165 SSA graph representing the function.</p>
6166
Reid Spencer97c5fa42006-11-08 01:18:52 +00006167<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006168<p>The type of the incoming values is specified with the first type field. After
6169 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
6170 one pair for each predecessor basic block of the current block. Only values
6171 of <a href="#t_firstclass">first class</a> type may be used as the value
6172 arguments to the PHI node. Only labels may be used as the label
6173 arguments.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006174
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006175<p>There must be no non-phi instructions between the start of a basic block and
6176 the PHI instructions: i.e. PHI instructions must be first in a basic
6177 block.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006178
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006179<p>For the purposes of the SSA form, the use of each incoming value is deemed to
6180 occur on the edge from the corresponding predecessor block to the current
6181 block (but after any definition of an '<tt>invoke</tt>' instruction's return
6182 value on the same edge).</p>
Jay Foad1a4eea52009-06-03 10:20:10 +00006183
Reid Spencer97c5fa42006-11-08 01:18:52 +00006184<h5>Semantics:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00006185<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006186 specified by the pair corresponding to the predecessor basic block that
6187 executed just prior to the current block.</p>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006188
Reid Spencer97c5fa42006-11-08 01:18:52 +00006189<h5>Example:</h5>
Chris Lattnerb54c30f2008-05-20 20:48:21 +00006190<pre>
6191Loop: ; Infinite loop that counts from 0 on up...
6192 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
6193 %nextindvar = add i32 %indvar, 1
6194 br label %Loop
6195</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006196
Reid Spencer97c5fa42006-11-08 01:18:52 +00006197</div>
6198
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006199<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006200<h4>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006201 <a name="i_select">'<tt>select</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006202</h4>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006203
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006204<div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006205
6206<h5>Syntax:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006207<pre>
Dan Gohmanc579d972008-09-09 01:02:47 +00006208 &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>
6209
Dan Gohmanef9462f2008-10-14 16:51:45 +00006210 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006211</pre>
6212
6213<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006214<p>The '<tt>select</tt>' instruction is used to choose one value based on a
6215 condition, without branching.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006216
6217
6218<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006219<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
6220 values indicating the condition, and two values of the
6221 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
6222 vectors and the condition is a scalar, then entire vectors are selected, not
6223 individual elements.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006224
6225<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006226<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
6227 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006228
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006229<p>If the condition is a vector of i1, then the value arguments must be vectors
6230 of the same size, and the selection is done element by element.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006231
6232<h5>Example:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006233<pre>
Reid Spencer36a15422007-01-12 03:35:51 +00006234 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006235</pre>
Dan Gohmana5127ab2009-01-22 01:39:38 +00006236
Chris Lattnerb53c28d2004-03-12 05:50:16 +00006237</div>
6238
Robert Bocchinof72fdfe2006-01-15 20:48:27 +00006239<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006240<h4>
Chris Lattnere23c1392005-05-06 05:47:36 +00006241 <a name="i_call">'<tt>call</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006242</h4>
Chris Lattnere23c1392005-05-06 05:47:36 +00006243
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006244<div>
Chris Lattnere23c1392005-05-06 05:47:36 +00006245
Chris Lattner2f7c9632001-06-06 20:29:01 +00006246<h5>Syntax:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00006247<pre>
Devang Patel02256232008-10-07 17:48:33 +00006248 &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 +00006249</pre>
6250
Chris Lattner2f7c9632001-06-06 20:29:01 +00006251<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00006252<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00006253
Chris Lattner2f7c9632001-06-06 20:29:01 +00006254<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00006255<p>This instruction requires several arguments:</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00006256
Chris Lattnera8292f32002-05-06 22:08:29 +00006257<ol>
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00006258 <li>The optional "tail" marker indicates that the callee function does not
6259 access any allocas or varargs in the caller. Note that calls may be
6260 marked "tail" even if they do not occur before
6261 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
6262 present, the function call is eligible for tail call optimization,
6263 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Cheng59676492010-03-08 21:05:02 +00006264 optimized into a jump</a>. The code generator may optimize calls marked
6265 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
6266 sibling call optimization</a> when the caller and callee have
6267 matching signatures, or 2) forced tail call optimization when the
6268 following extra requirements are met:
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00006269 <ul>
6270 <li>Caller and callee both have the calling
6271 convention <tt>fastcc</tt>.</li>
6272 <li>The call is in tail position (ret immediately follows call and ret
6273 uses value of call or is void).</li>
6274 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohman6232f732010-03-02 01:08:11 +00006275 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00006276 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
6277 constraints are met.</a></li>
6278 </ul>
6279 </li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00006280
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006281 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
6282 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskinb8677462010-01-09 19:44:16 +00006283 defaults to using C calling conventions. The calling convention of the
6284 call must match the calling convention of the target function, or else the
6285 behavior is undefined.</li>
Devang Patel7e9b05e2008-10-06 18:50:38 +00006286
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006287 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
6288 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
6289 '<tt>inreg</tt>' attributes are valid here.</li>
6290
6291 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
6292 type of the return value. Functions that return no value are marked
6293 <tt><a href="#t_void">void</a></tt>.</li>
6294
6295 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
6296 being invoked. The argument types must match the types implied by this
6297 signature. This type can be omitted if the function is not varargs and if
6298 the function type does not return a pointer to a function.</li>
6299
6300 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
6301 be invoked. In most cases, this is a direct function invocation, but
6302 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
6303 to function value.</li>
6304
6305 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner47f2a832010-03-02 06:36:51 +00006306 signature argument types and parameter attributes. All arguments must be
6307 of <a href="#t_firstclass">first class</a> type. If the function
6308 signature indicates the function accepts a variable number of arguments,
6309 the extra arguments can be specified.</li>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006310
6311 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
6312 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
6313 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattnera8292f32002-05-06 22:08:29 +00006314</ol>
Chris Lattnere23c1392005-05-06 05:47:36 +00006315
Chris Lattner2f7c9632001-06-06 20:29:01 +00006316<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006317<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
6318 a specified function, with its incoming arguments bound to the specified
6319 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
6320 function, control flow continues with the instruction after the function
6321 call, and the return value of the function is bound to the result
6322 argument.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +00006323
Chris Lattner2f7c9632001-06-06 20:29:01 +00006324<h5>Example:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +00006325<pre>
Nick Lewyckya9b13d52007-09-08 13:57:50 +00006326 %retval = call i32 @test(i32 %argc)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006327 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattnerfb7c88d2008-03-21 17:24:17 +00006328 %X = tail call i32 @foo() <i>; yields i32</i>
6329 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
6330 call void %foo(i8 97 signext)
Devang Pateld6cff512008-03-10 20:49:15 +00006331
6332 %struct.A = type { i32, i8 }
Devang Patel7e9b05e2008-10-06 18:50:38 +00006333 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmancc3132e2008-10-04 19:00:07 +00006334 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
6335 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner6cbe8e92008-10-08 06:26:11 +00006336 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmaneefa7df2008-10-07 10:03:45 +00006337 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattnere23c1392005-05-06 05:47:36 +00006338</pre>
6339
Dale Johannesen68f971b2009-09-24 18:38:21 +00006340<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen722212d2009-09-25 17:04:42 +00006341standard C99 library as being the C99 library functions, and may perform
6342optimizations or generate code for them under that assumption. This is
6343something we'd like to change in the future to provide better support for
Dan Gohmana269a0a2010-03-01 17:41:39 +00006344freestanding environments and non-C-based languages.</p>
Dale Johannesen68f971b2009-09-24 18:38:21 +00006345
Misha Brukman76307852003-11-08 01:05:38 +00006346</div>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006347
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006348<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006349<h4>
Chris Lattner33337472006-01-13 23:26:01 +00006350 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006351</h4>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006352
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006353<div>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006354
Chris Lattner26ca62e2003-10-18 05:51:36 +00006355<h5>Syntax:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006356<pre>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00006357 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattner6a4a0492004-09-27 21:51:25 +00006358</pre>
6359
Chris Lattner26ca62e2003-10-18 05:51:36 +00006360<h5>Overview:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00006361<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006362 the "variable argument" area of a function call. It is used to implement the
6363 <tt>va_arg</tt> macro in C.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006364
Chris Lattner26ca62e2003-10-18 05:51:36 +00006365<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006366<p>This instruction takes a <tt>va_list*</tt> value and the type of the
6367 argument. It returns a value of the specified argument type and increments
6368 the <tt>va_list</tt> to point to the next argument. The actual type
6369 of <tt>va_list</tt> is target specific.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006370
Chris Lattner26ca62e2003-10-18 05:51:36 +00006371<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006372<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
6373 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
6374 to the next argument. For more information, see the variable argument
6375 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006376
6377<p>It is legal for this instruction to be called in a function which does not
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006378 take a variable number of arguments, for example, the <tt>vfprintf</tt>
6379 function.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006380
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006381<p><tt>va_arg</tt> is an LLVM instruction instead of
6382 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
6383 argument.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006384
Chris Lattner26ca62e2003-10-18 05:51:36 +00006385<h5>Example:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +00006386<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6387
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006388<p>Note that the code generator does not yet fully support va_arg on many
6389 targets. Also, it does not currently support va_arg with aggregate types on
6390 any target.</p>
Dan Gohman3065b612009-01-12 23:12:39 +00006391
Misha Brukman76307852003-11-08 01:05:38 +00006392</div>
Chris Lattner941515c2004-01-06 05:31:32 +00006393
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006394<!-- _______________________________________________________________________ -->
6395<h4>
6396 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6397</h4>
6398
6399<div>
6400
6401<h5>Syntax:</h5>
6402<pre>
Duncan Sandsdf9d7812012-01-13 19:59:16 +00006403 &lt;resultval&gt; = landingpad &lt;resultty&gt; personality &lt;type&gt; &lt;pers_fn&gt; &lt;clause&gt;+
6404 &lt;resultval&gt; = landingpad &lt;resultty&gt; personality &lt;type&gt; &lt;pers_fn&gt; cleanup &lt;clause&gt;*
Bill Wendling49bfb122011-08-08 08:06:05 +00006405
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006406 &lt;clause&gt; := catch &lt;type&gt; &lt;value&gt;
Bill Wendlingfae14752011-08-12 20:24:12 +00006407 &lt;clause&gt; := filter &lt;array constant type&gt; &lt;array constant&gt;
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006408</pre>
6409
6410<h5>Overview:</h5>
6411<p>The '<tt>landingpad</tt>' instruction is used by
6412 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6413 system</a> to specify that a basic block is a landing pad &mdash; one where
6414 the exception lands, and corresponds to the code found in the
6415 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6416 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6417 re-entry to the function. The <tt>resultval</tt> has the
Duncan Sandsdf9d7812012-01-13 19:59:16 +00006418 type <tt>resultty</tt>.</p>
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006419
6420<h5>Arguments:</h5>
6421<p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6422 function associated with the unwinding mechanism. The optional
6423 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6424
6425<p>A <tt>clause</tt> begins with the clause type &mdash; <tt>catch</tt>
Bill Wendlingfae14752011-08-12 20:24:12 +00006426 or <tt>filter</tt> &mdash; and contains the global variable representing the
6427 "type" that may be caught or filtered respectively. Unlike the
6428 <tt>catch</tt> clause, the <tt>filter</tt> clause takes an array constant as
6429 its argument. Use "<tt>[0 x i8**] undef</tt>" for a filter which cannot
6430 throw. The '<tt>landingpad</tt>' instruction must contain <em>at least</em>
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006431 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6432
6433<h5>Semantics:</h5>
6434<p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6435 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6436 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6437 calling conventions, how the personality function results are represented in
6438 LLVM IR is target specific.</p>
6439
Bill Wendling0524b8d2011-08-03 17:17:06 +00006440<p>The clauses are applied in order from top to bottom. If two
6441 <tt>landingpad</tt> instructions are merged together through inlining, the
Duncan Sandsdf9d7812012-01-13 19:59:16 +00006442 clauses from the calling function are appended to the list of clauses.
6443 When the call stack is being unwound due to an exception being thrown, the
6444 exception is compared against each <tt>clause</tt> in turn. If it doesn't
6445 match any of the clauses, and the <tt>cleanup</tt> flag is not set, then
6446 unwinding continues further up the call stack.</p>
Bill Wendling0524b8d2011-08-03 17:17:06 +00006447
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006448<p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6449
6450<ul>
6451 <li>A landing pad block is a basic block which is the unwind destination of an
6452 '<tt>invoke</tt>' instruction.</li>
6453 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6454 first non-PHI instruction.</li>
6455 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6456 pad block.</li>
6457 <li>A basic block that is not a landing pad block may not include a
6458 '<tt>landingpad</tt>' instruction.</li>
6459 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6460 personality function.</li>
6461</ul>
6462
6463<h5>Example:</h5>
6464<pre>
6465 ;; A landing pad which can catch an integer.
6466 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6467 catch i8** @_ZTIi
6468 ;; A landing pad that is a cleanup.
6469 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
Bill Wendlingfae14752011-08-12 20:24:12 +00006470 cleanup
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006471 ;; A landing pad which can catch an integer and can only throw a double.
6472 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6473 catch i8** @_ZTIi
Bill Wendlingfae14752011-08-12 20:24:12 +00006474 filter [1 x i8**] [@_ZTId]
Bill Wendlingbbcb7cd2011-08-02 21:52:38 +00006475</pre>
6476
6477</div>
6478
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006479</div>
6480
6481</div>
6482
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006483<!-- *********************************************************************** -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006484<h2><a name="intrinsics">Intrinsic Functions</a></h2>
Chris Lattner48b383b02003-11-25 01:02:51 +00006485<!-- *********************************************************************** -->
Chris Lattner941515c2004-01-06 05:31:32 +00006486
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006487<div>
Chris Lattnerfee11462004-02-12 17:01:32 +00006488
6489<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006490 well known names and semantics and are required to follow certain
6491 restrictions. Overall, these intrinsics represent an extension mechanism for
6492 the LLVM language that does not require changing all of the transformations
6493 in LLVM when adding to the language (or the bitcode reader/writer, the
6494 parser, etc...).</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006495
John Criswell88190562005-05-16 16:17:45 +00006496<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006497 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6498 begin with this prefix. Intrinsic functions must always be external
6499 functions: you cannot define the body of intrinsic functions. Intrinsic
6500 functions may only be used in call or invoke instructions: it is illegal to
6501 take the address of an intrinsic function. Additionally, because intrinsic
6502 functions are part of the LLVM language, it is required if any are added that
6503 they be documented here.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006504
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006505<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6506 family of functions that perform the same operation but on different data
6507 types. Because LLVM can represent over 8 million different integer types,
6508 overloading is used commonly to allow an intrinsic function to operate on any
6509 integer type. One or more of the argument types or the result type can be
6510 overloaded to accept any integer type. Argument types may also be defined as
6511 exactly matching a previous argument's type or the result type. This allows
6512 an intrinsic function which accepts multiple arguments, but needs all of them
6513 to be of the same type, to only be overloaded with respect to a single
6514 argument or the result.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006515
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006516<p>Overloaded intrinsics will have the names of its overloaded argument types
6517 encoded into its function name, each preceded by a period. Only those types
6518 which are overloaded result in a name suffix. Arguments whose type is matched
6519 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6520 can take an integer of any width and returns an integer of exactly the same
6521 integer width. This leads to a family of functions such as
6522 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6523 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6524 suffix is required. Because the argument's type is matched against the return
6525 type, it does not require its own name suffix.</p>
Reid Spencer4eefaab2007-04-01 08:04:23 +00006526
Eric Christopher455c5772009-12-05 02:46:03 +00006527<p>To learn how to add an intrinsic function, please see the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006528 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00006529
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006530<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006531<h3>
Chris Lattner941515c2004-01-06 05:31:32 +00006532 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006533</h3>
Chris Lattner941515c2004-01-06 05:31:32 +00006534
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006535<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006536
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006537<p>Variable argument support is defined in LLVM with
6538 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6539 intrinsic functions. These functions are related to the similarly named
6540 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006541
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006542<p>All of these functions operate on arguments that use a target-specific value
6543 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6544 not define what this type is, so all transformations should be prepared to
6545 handle these functions regardless of the type used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006546
Chris Lattner30b868d2006-05-15 17:26:46 +00006547<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006548 instruction and the variable argument handling intrinsic functions are
6549 used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006550
Benjamin Kramer79698be2010-07-13 12:26:09 +00006551<pre class="doc_code">
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00006552define i32 @test(i32 %X, ...) {
Chris Lattnerfee11462004-02-12 17:01:32 +00006553 ; Initialize variable argument processing
Jeff Cohen222a8a42007-04-29 01:07:00 +00006554 %ap = alloca i8*
Chris Lattnerdb0790c2007-01-08 07:55:15 +00006555 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00006556 call void @llvm.va_start(i8* %ap2)
Chris Lattnerfee11462004-02-12 17:01:32 +00006557
6558 ; Read a single integer argument
Jeff Cohen222a8a42007-04-29 01:07:00 +00006559 %tmp = va_arg i8** %ap, i32
Chris Lattnerfee11462004-02-12 17:01:32 +00006560
6561 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohen222a8a42007-04-29 01:07:00 +00006562 %aq = alloca i8*
Chris Lattnerdb0790c2007-01-08 07:55:15 +00006563 %aq2 = bitcast i8** %aq to i8*
Jeff Cohen222a8a42007-04-29 01:07:00 +00006564 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00006565 call void @llvm.va_end(i8* %aq2)
Chris Lattnerfee11462004-02-12 17:01:32 +00006566
6567 ; Stop processing of arguments.
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00006568 call void @llvm.va_end(i8* %ap2)
Reid Spencerb5ebf3d2006-12-31 07:07:53 +00006569 ret i32 %tmp
Chris Lattnerfee11462004-02-12 17:01:32 +00006570}
Anton Korobeynikov640bbe02007-03-21 23:58:04 +00006571
6572declare void @llvm.va_start(i8*)
6573declare void @llvm.va_copy(i8*, i8*)
6574declare void @llvm.va_end(i8*)
Chris Lattnerfee11462004-02-12 17:01:32 +00006575</pre>
Chris Lattner941515c2004-01-06 05:31:32 +00006576
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006577<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006578<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006579 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006580</h4>
Chris Lattner941515c2004-01-06 05:31:32 +00006581
6582
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006583<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006584
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006585<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006586<pre>
6587 declare void %llvm.va_start(i8* &lt;arglist&gt;)
6588</pre>
6589
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006590<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006591<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
6592 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00006593
6594<h5>Arguments:</h5>
Dan Gohmanef9462f2008-10-14 16:51:45 +00006595<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00006596
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006597<h5>Semantics:</h5>
Dan Gohmanef9462f2008-10-14 16:51:45 +00006598<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006599 macro available in C. In a target-dependent way, it initializes
6600 the <tt>va_list</tt> element to which the argument points, so that the next
6601 call to <tt>va_arg</tt> will produce the first variable argument passed to
6602 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6603 need to know the last argument of the function as the compiler can figure
6604 that out.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00006605
Misha Brukman76307852003-11-08 01:05:38 +00006606</div>
Chris Lattner941515c2004-01-06 05:31:32 +00006607
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006608<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006609<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006610 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006611</h4>
Chris Lattner941515c2004-01-06 05:31:32 +00006612
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006613<div>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00006614
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006615<h5>Syntax:</h5>
6616<pre>
6617 declare void @llvm.va_end(i8* &lt;arglist&gt;)
6618</pre>
6619
6620<h5>Overview:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00006621<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006622 which has been initialized previously
6623 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6624 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00006625
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006626<h5>Arguments:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00006627<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00006628
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006629<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +00006630<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006631 macro available in C. In a target-dependent way, it destroys
6632 the <tt>va_list</tt> element to which the argument points. Calls
6633 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6634 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6635 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +00006636
Misha Brukman76307852003-11-08 01:05:38 +00006637</div>
Chris Lattner941515c2004-01-06 05:31:32 +00006638
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006639<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006640<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006641 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006642</h4>
Chris Lattner941515c2004-01-06 05:31:32 +00006643
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006644<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006645
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006646<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006647<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00006648 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattner757528b0b2004-05-23 21:06:01 +00006649</pre>
6650
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006651<h5>Overview:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00006652<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006653 from the source argument list to the destination argument list.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006654
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006655<h5>Arguments:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +00006656<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006657 The second argument is a pointer to a <tt>va_list</tt> element to copy
6658 from.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006659
Chris Lattnerbd64b4e2003-05-08 04:57:36 +00006660<h5>Semantics:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +00006661<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006662 macro available in C. In a target-dependent way, it copies the
6663 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6664 element. This intrinsic is necessary because
6665 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6666 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006667
Misha Brukman76307852003-11-08 01:05:38 +00006668</div>
Chris Lattner941515c2004-01-06 05:31:32 +00006669
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006670</div>
6671
Chris Lattnerfee11462004-02-12 17:01:32 +00006672<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006673<h3>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006674 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006675</h3>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006676
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006677<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006678
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006679<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner67c37d12008-08-05 18:29:16 +00006680Collection</a> (GC) requires the implementation and generation of these
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006681intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6682roots on the stack</a>, as well as garbage collector implementations that
6683require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6684barriers. Front-ends for type-safe garbage collected languages should generate
6685these intrinsics to make use of the LLVM garbage collectors. For more details,
6686see <a href="GarbageCollection.html">Accurate Garbage Collection with
6687LLVM</a>.</p>
Christopher Lamb55c6d4f2007-12-17 01:00:21 +00006688
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006689<p>The garbage collection intrinsics only operate on objects in the generic
6690 address space (address space zero).</p>
Christopher Lamb55c6d4f2007-12-17 01:00:21 +00006691
Chris Lattner757528b0b2004-05-23 21:06:01 +00006692<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006693<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006694 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006695</h4>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006696
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006697<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006698
6699<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006700<pre>
Chris Lattner12477732007-09-21 17:30:40 +00006701 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattner757528b0b2004-05-23 21:06:01 +00006702</pre>
6703
6704<h5>Overview:</h5>
John Criswelldfe6a862004-12-10 15:51:16 +00006705<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006706 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006707
6708<h5>Arguments:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006709<p>The first argument specifies the address of a stack object that contains the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006710 root pointer. The second pointer (which must be either a constant or a
6711 global value address) contains the meta-data to be associated with the
6712 root.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006713
6714<h5>Semantics:</h5>
Chris Lattner851b7712008-04-24 05:59:56 +00006715<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006716 location. At compile-time, the code generator generates information to allow
6717 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6718 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6719 algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006720
6721</div>
6722
Chris Lattner757528b0b2004-05-23 21:06:01 +00006723<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006724<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006725 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006726</h4>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006727
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006728<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006729
6730<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006731<pre>
Chris Lattner12477732007-09-21 17:30:40 +00006732 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattner757528b0b2004-05-23 21:06:01 +00006733</pre>
6734
6735<h5>Overview:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006736<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006737 locations, allowing garbage collector implementations that require read
6738 barriers.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006739
6740<h5>Arguments:</h5>
Chris Lattnerf9228072006-03-14 20:02:51 +00006741<p>The second argument is the address to read from, which should be an address
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006742 allocated from the garbage collector. The first object is a pointer to the
6743 start of the referenced object, if needed by the language runtime (otherwise
6744 null).</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006745
6746<h5>Semantics:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006747<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006748 instruction, but may be replaced with substantially more complex code by the
6749 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6750 may only be used in a function which <a href="#gc">specifies a GC
6751 algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006752
6753</div>
6754
Chris Lattner757528b0b2004-05-23 21:06:01 +00006755<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006756<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006757 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006758</h4>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006759
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006760<div>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006761
6762<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006763<pre>
Chris Lattner12477732007-09-21 17:30:40 +00006764 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattner757528b0b2004-05-23 21:06:01 +00006765</pre>
6766
6767<h5>Overview:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006768<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006769 locations, allowing garbage collector implementations that require write
6770 barriers (such as generational or reference counting collectors).</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006771
6772<h5>Arguments:</h5>
Chris Lattnerf9228072006-03-14 20:02:51 +00006773<p>The first argument is the reference to store, the second is the start of the
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006774 object to store it to, and the third is the address of the field of Obj to
6775 store to. If the runtime does not require a pointer to the object, Obj may
6776 be null.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006777
6778<h5>Semantics:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006779<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006780 instruction, but may be replaced with substantially more complex code by the
6781 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6782 may only be used in a function which <a href="#gc">specifies a GC
6783 algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +00006784
6785</div>
6786
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006787</div>
6788
Chris Lattner757528b0b2004-05-23 21:06:01 +00006789<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006790<h3>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006791 <a name="int_codegen">Code Generator Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006792</h3>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006793
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006794<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006795
6796<p>These intrinsics are provided by LLVM to expose special features that may
6797 only be implemented with code generator support.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006798
Chris Lattner3649c3a2004-02-14 04:08:35 +00006799<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006800<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006801 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006802</h4>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006803
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006804<div>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006805
6806<h5>Syntax:</h5>
6807<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00006808 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00006809</pre>
6810
6811<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006812<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6813 target-specific value indicating the return address of the current function
6814 or one of its callers.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006815
6816<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006817<p>The argument to this intrinsic indicates which function to return the address
6818 for. Zero indicates the calling function, one indicates its caller, etc.
6819 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006820
6821<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006822<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6823 indicating the return address of the specified call frame, or zero if it
6824 cannot be identified. The value returned by this intrinsic is likely to be
6825 incorrect or 0 for arguments other than zero, so it should only be used for
6826 debugging purposes.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006827
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006828<p>Note that calling this intrinsic does not prevent function inlining or other
6829 aggressive transformations, so the value returned may not be that of the
6830 obvious source-language caller.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006831
Chris Lattner3649c3a2004-02-14 04:08:35 +00006832</div>
6833
Chris Lattner3649c3a2004-02-14 04:08:35 +00006834<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006835<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006836 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006837</h4>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006838
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006839<div>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006840
6841<h5>Syntax:</h5>
6842<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006843 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00006844</pre>
6845
6846<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006847<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6848 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006849
6850<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006851<p>The argument to this intrinsic indicates which function to return the frame
6852 pointer for. Zero indicates the calling function, one indicates its caller,
6853 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006854
6855<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006856<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6857 indicating the frame address of the specified call frame, or zero if it
6858 cannot be identified. The value returned by this intrinsic is likely to be
6859 incorrect or 0 for arguments other than zero, so it should only be used for
6860 debugging purposes.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006861
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006862<p>Note that calling this intrinsic does not prevent function inlining or other
6863 aggressive transformations, so the value returned may not be that of the
6864 obvious source-language caller.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00006865
Chris Lattner3649c3a2004-02-14 04:08:35 +00006866</div>
6867
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006868<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006869<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006870 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006871</h4>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006872
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006873<div>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006874
6875<h5>Syntax:</h5>
6876<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006877 declare i8* @llvm.stacksave()
Chris Lattner2f0f0012006-01-13 02:03:13 +00006878</pre>
6879
6880<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006881<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6882 of the function stack, for use
6883 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6884 useful for implementing language features like scoped automatic variable
6885 sized arrays in C99.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006886
6887<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006888<p>This intrinsic returns a opaque pointer value that can be passed
6889 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6890 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6891 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6892 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6893 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6894 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006895
6896</div>
6897
6898<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006899<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006900 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006901</h4>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006902
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006903<div>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006904
6905<h5>Syntax:</h5>
6906<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00006907 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner2f0f0012006-01-13 02:03:13 +00006908</pre>
6909
6910<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006911<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6912 the function stack to the state it was in when the
6913 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6914 executed. This is useful for implementing language features like scoped
6915 automatic variable sized arrays in C99.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006916
6917<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006918<p>See the description
6919 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner2f0f0012006-01-13 02:03:13 +00006920
6921</div>
6922
Chris Lattner2f0f0012006-01-13 02:03:13 +00006923<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006924<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006925 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006926</h4>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006927
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006928<div>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006929
6930<h5>Syntax:</h5>
6931<pre>
Bruno Cardoso Lopesdc9ff3a2011-06-14 04:58:37 +00006932 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;, i32 &lt;cache type&gt;)
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006933</pre>
6934
6935<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006936<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6937 insert a prefetch instruction if supported; otherwise, it is a noop.
6938 Prefetches have no effect on the behavior of the program but can change its
6939 performance characteristics.</p>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006940
6941<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006942<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6943 specifier determining if the fetch should be for a read (0) or write (1),
6944 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Bruno Cardoso Lopesdc9ff3a2011-06-14 04:58:37 +00006945 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
6946 specifies whether the prefetch is performed on the data (1) or instruction (0)
6947 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
6948 must be constant integers.</p>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006949
6950<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006951<p>This intrinsic does not modify the behavior of the program. In particular,
6952 prefetches cannot trap and do not produce a value. On targets that support
6953 this intrinsic, the prefetch can provide hints to the processor cache for
6954 better performance.</p>
Chris Lattnerc8a2c222005-02-28 19:24:19 +00006955
6956</div>
6957
Andrew Lenharthb4427912005-03-28 20:05:49 +00006958<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006959<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006960 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006961</h4>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006962
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006963<div>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006964
6965<h5>Syntax:</h5>
6966<pre>
Chris Lattner12477732007-09-21 17:30:40 +00006967 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharthb4427912005-03-28 20:05:49 +00006968</pre>
6969
6970<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006971<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6972 Counter (PC) in a region of code to simulators and other tools. The method
6973 is target specific, but it is expected that the marker will use exported
6974 symbols to transmit the PC of the marker. The marker makes no guarantees
6975 that it will remain with any specific instruction after optimizations. It is
6976 possible that the presence of a marker will inhibit optimizations. The
6977 intended use is to be inserted after optimizations to allow correlations of
6978 simulation runs.</p>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006979
6980<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006981<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006982
6983<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00006984<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmana269a0a2010-03-01 17:41:39 +00006985 not support this intrinsic may ignore it.</p>
Andrew Lenharthb4427912005-03-28 20:05:49 +00006986
6987</div>
6988
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006989<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006990<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00006991 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00006992</h4>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006993
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00006994<div>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006995
6996<h5>Syntax:</h5>
6997<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00006998 declare i64 @llvm.readcyclecounter()
Andrew Lenharth01aa5632005-11-11 16:47:30 +00006999</pre>
7000
7001<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007002<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
7003 counter register (or similar low latency, high accuracy clocks) on those
7004 targets that support it. On X86, it should map to RDTSC. On Alpha, it
7005 should map to RPCC. As the backing counters overflow quickly (on the order
7006 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00007007
7008<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007009<p>When directly supported, reading the cycle counter should not modify any
7010 memory. Implementations are allowed to either return a application specific
7011 value or a system wide value. On backends without support, this is lowered
7012 to a constant 0.</p>
Andrew Lenharth01aa5632005-11-11 16:47:30 +00007013
7014</div>
7015
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007016</div>
7017
Chris Lattner3649c3a2004-02-14 04:08:35 +00007018<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007019<h3>
Chris Lattnerfee11462004-02-12 17:01:32 +00007020 <a name="int_libc">Standard C Library Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007021</h3>
Chris Lattnerfee11462004-02-12 17:01:32 +00007022
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007023<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007024
7025<p>LLVM provides intrinsics for a few important standard C library functions.
7026 These intrinsics allow source-language front-ends to pass information about
7027 the alignment of the pointer arguments to the code generator, providing
7028 opportunity for more efficient code generation.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00007029
Chris Lattnerfee11462004-02-12 17:01:32 +00007030<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007031<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00007032 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007033</h4>
Chris Lattnerfee11462004-02-12 17:01:32 +00007034
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007035<div>
Chris Lattnerfee11462004-02-12 17:01:32 +00007036
7037<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007038<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
Mon P Wang508127b2010-04-07 06:35:53 +00007039 integer bit width and for different address spaces. Not all targets support
7040 all bit widths however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007041
Chris Lattnerfee11462004-02-12 17:01:32 +00007042<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00007043 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007044 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00007045 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007046 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattnerfee11462004-02-12 17:01:32 +00007047</pre>
7048
7049<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007050<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
7051 source location to the destination location.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00007052
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007053<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007054 intrinsics do not return a value, takes extra alignment/isvolatile arguments
7055 and the pointers can be in specified address spaces.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00007056
7057<h5>Arguments:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007058
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007059<p>The first argument is a pointer to the destination, the second is a pointer
7060 to the source. The third argument is an integer argument specifying the
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007061 number of bytes to copy, the fourth argument is the alignment of the
7062 source and destination locations, and the fifth is a boolean indicating a
7063 volatile access.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00007064
Dan Gohmana269a0a2010-03-01 17:41:39 +00007065<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007066 then the caller guarantees that both the source and destination pointers are
7067 aligned to that boundary.</p>
Chris Lattner4c67c482004-02-12 21:18:15 +00007068
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00007069<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7070 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
7071 The detailed access behavior is not very cleanly specified and it is unwise
7072 to depend on it.</p>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007073
Chris Lattnerfee11462004-02-12 17:01:32 +00007074<h5>Semantics:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007075
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007076<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
7077 source location to the destination location, which are not allowed to
7078 overlap. It copies "len" bytes of memory over. If the argument is known to
7079 be aligned to some boundary, this can be specified as the fourth argument,
7080 otherwise it should be set to 0 or 1.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +00007081
Chris Lattnerfee11462004-02-12 17:01:32 +00007082</div>
7083
Chris Lattnerf30152e2004-02-12 18:10:10 +00007084<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007085<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00007086 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007087</h4>
Chris Lattnerf30152e2004-02-12 18:10:10 +00007088
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007089<div>
Chris Lattnerf30152e2004-02-12 18:10:10 +00007090
7091<h5>Syntax:</h5>
Chris Lattnerdd708342008-11-21 16:42:48 +00007092<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Mon P Wang508127b2010-04-07 06:35:53 +00007093 width and for different address space. Not all targets support all bit
7094 widths however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007095
Chris Lattnerf30152e2004-02-12 18:10:10 +00007096<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00007097 declare void @llvm.memmove.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007098 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00007099 declare void @llvm.memmove.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007100 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattnerf30152e2004-02-12 18:10:10 +00007101</pre>
7102
7103<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007104<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
7105 source location to the destination location. It is similar to the
7106 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
7107 overlap.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00007108
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007109<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007110 intrinsics do not return a value, takes extra alignment/isvolatile arguments
7111 and the pointers can be in specified address spaces.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00007112
7113<h5>Arguments:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007114
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007115<p>The first argument is a pointer to the destination, the second is a pointer
7116 to the source. The third argument is an integer argument specifying the
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007117 number of bytes to copy, the fourth argument is the alignment of the
7118 source and destination locations, and the fifth is a boolean indicating a
7119 volatile access.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00007120
Dan Gohmana269a0a2010-03-01 17:41:39 +00007121<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007122 then the caller guarantees that the source and destination pointers are
7123 aligned to that boundary.</p>
Chris Lattner4c67c482004-02-12 21:18:15 +00007124
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00007125<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7126 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
7127 The detailed access behavior is not very cleanly specified and it is unwise
7128 to depend on it.</p>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007129
Chris Lattnerf30152e2004-02-12 18:10:10 +00007130<h5>Semantics:</h5>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007131
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007132<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
7133 source location to the destination location, which may overlap. It copies
7134 "len" bytes of memory over. If the argument is known to be aligned to some
7135 boundary, this can be specified as the fourth argument, otherwise it should
7136 be set to 0 or 1.</p>
Chris Lattnerf30152e2004-02-12 18:10:10 +00007137
Chris Lattnerf30152e2004-02-12 18:10:10 +00007138</div>
7139
Chris Lattner3649c3a2004-02-14 04:08:35 +00007140<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007141<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00007142 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007143</h4>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007144
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007145<div>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007146
7147<h5>Syntax:</h5>
Chris Lattnerdd708342008-11-21 16:42:48 +00007148<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
John Criswellad05ae42010-07-30 16:30:28 +00007149 width and for different address spaces. However, not all targets support all
7150 bit widths.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007151
Chris Lattner3649c3a2004-02-14 04:08:35 +00007152<pre>
Dan Gohmanaabfdb32010-05-28 17:13:49 +00007153 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattner685db9d2010-04-08 00:54:34 +00007154 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanaabfdb32010-05-28 17:13:49 +00007155 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattner685db9d2010-04-08 00:54:34 +00007156 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +00007157</pre>
7158
7159<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007160<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
7161 particular byte value.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007162
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007163<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
John Criswellad05ae42010-07-30 16:30:28 +00007164 intrinsic does not return a value and takes extra alignment/volatile
7165 arguments. Also, the destination can be in an arbitrary address space.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007166
7167<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007168<p>The first argument is a pointer to the destination to fill, the second is the
John Criswellad05ae42010-07-30 16:30:28 +00007169 byte value with which to fill it, the third argument is an integer argument
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007170 specifying the number of bytes to fill, and the fourth argument is the known
John Criswellad05ae42010-07-30 16:30:28 +00007171 alignment of the destination location.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007172
Dan Gohmana269a0a2010-03-01 17:41:39 +00007173<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007174 then the caller guarantees that the destination pointer is aligned to that
7175 boundary.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007176
Jeffrey Yasskin5d284ae2010-04-26 21:21:24 +00007177<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7178 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
7179 The detailed access behavior is not very cleanly specified and it is unwise
7180 to depend on it.</p>
Chris Lattnerbd4ca622010-04-08 00:53:57 +00007181
Chris Lattner3649c3a2004-02-14 04:08:35 +00007182<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007183<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
7184 at the destination location. If the argument is known to be aligned to some
7185 boundary, this can be specified as the fourth argument, otherwise it should
7186 be set to 0 or 1.</p>
Chris Lattner3649c3a2004-02-14 04:08:35 +00007187
Chris Lattner3649c3a2004-02-14 04:08:35 +00007188</div>
7189
Chris Lattner3b4f4372004-06-11 02:28:03 +00007190<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007191<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00007192 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007193</h4>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007194
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007195<div>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007196
7197<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007198<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
7199 floating point or vector of floating point type. Not all targets support all
7200 types however.</p>
7201
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007202<pre>
Dale Johannesendd89d272007-10-02 17:47:38 +00007203 declare float @llvm.sqrt.f32(float %Val)
7204 declare double @llvm.sqrt.f64(double %Val)
7205 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
7206 declare fp128 @llvm.sqrt.f128(fp128 %Val)
7207 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007208</pre>
7209
7210<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007211<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
7212 returning the same value as the libm '<tt>sqrt</tt>' functions would.
7213 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
7214 behavior for negative numbers other than -0.0 (which allows for better
7215 optimization, because there is no need to worry about errno being
7216 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007217
7218<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007219<p>The argument and return value are floating point numbers of the same
7220 type.</p>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007221
7222<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007223<p>This function returns the sqrt of the specified operand if it is a
7224 nonnegative floating point number.</p>
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007225
Chris Lattner8a8f2e52005-07-21 01:29:16 +00007226</div>
7227
Chris Lattner33b73f92006-09-08 06:34:02 +00007228<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007229<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00007230 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007231</h4>
Chris Lattner33b73f92006-09-08 06:34:02 +00007232
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007233<div>
Chris Lattner33b73f92006-09-08 06:34:02 +00007234
7235<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007236<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
7237 floating point or vector of floating point type. Not all targets support all
7238 types however.</p>
7239
Chris Lattner33b73f92006-09-08 06:34:02 +00007240<pre>
Dale Johannesendd89d272007-10-02 17:47:38 +00007241 declare float @llvm.powi.f32(float %Val, i32 %power)
7242 declare double @llvm.powi.f64(double %Val, i32 %power)
7243 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
7244 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
7245 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattner33b73f92006-09-08 06:34:02 +00007246</pre>
7247
7248<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007249<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
7250 specified (positive or negative) power. The order of evaluation of
7251 multiplications is not defined. When a vector of floating point type is
7252 used, the second argument remains a scalar integer value.</p>
Chris Lattner33b73f92006-09-08 06:34:02 +00007253
7254<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007255<p>The second argument is an integer power, and the first is a value to raise to
7256 that power.</p>
Chris Lattner33b73f92006-09-08 06:34:02 +00007257
7258<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007259<p>This function returns the first value raised to the second power with an
7260 unspecified sequence of rounding operations.</p>
Chris Lattner33b73f92006-09-08 06:34:02 +00007261
Chris Lattner33b73f92006-09-08 06:34:02 +00007262</div>
7263
Dan Gohmanb6324c12007-10-15 20:30:11 +00007264<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007265<h4>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007266 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007267</h4>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007268
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007269<div>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007270
7271<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007272<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
7273 floating point or vector of floating point type. Not all targets support all
7274 types however.</p>
7275
Dan Gohmanb6324c12007-10-15 20:30:11 +00007276<pre>
7277 declare float @llvm.sin.f32(float %Val)
7278 declare double @llvm.sin.f64(double %Val)
7279 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7280 declare fp128 @llvm.sin.f128(fp128 %Val)
7281 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7282</pre>
7283
7284<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007285<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007286
7287<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007288<p>The argument and return value are floating point numbers of the same
7289 type.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007290
7291<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007292<p>This function returns the sine of the specified operand, returning the same
7293 values as the libm <tt>sin</tt> functions would, and handles error conditions
7294 in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007295
Dan Gohmanb6324c12007-10-15 20:30:11 +00007296</div>
7297
7298<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007299<h4>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007300 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007301</h4>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007302
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007303<div>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007304
7305<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007306<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
7307 floating point or vector of floating point type. Not all targets support all
7308 types however.</p>
7309
Dan Gohmanb6324c12007-10-15 20:30:11 +00007310<pre>
7311 declare float @llvm.cos.f32(float %Val)
7312 declare double @llvm.cos.f64(double %Val)
7313 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7314 declare fp128 @llvm.cos.f128(fp128 %Val)
7315 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7316</pre>
7317
7318<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007319<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007320
7321<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007322<p>The argument and return value are floating point numbers of the same
7323 type.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007324
7325<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007326<p>This function returns the cosine of the specified operand, returning the same
7327 values as the libm <tt>cos</tt> functions would, and handles error conditions
7328 in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007329
Dan Gohmanb6324c12007-10-15 20:30:11 +00007330</div>
7331
7332<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007333<h4>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007334 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007335</h4>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007336
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007337<div>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007338
7339<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007340<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
7341 floating point or vector of floating point type. Not all targets support all
7342 types however.</p>
7343
Dan Gohmanb6324c12007-10-15 20:30:11 +00007344<pre>
7345 declare float @llvm.pow.f32(float %Val, float %Power)
7346 declare double @llvm.pow.f64(double %Val, double %Power)
7347 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7348 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7349 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7350</pre>
7351
7352<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007353<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
7354 specified (positive or negative) power.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007355
7356<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007357<p>The second argument is a floating point power, and the first is a value to
7358 raise to that power.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007359
7360<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007361<p>This function returns the first value raised to the second power, returning
7362 the same values as the libm <tt>pow</tt> functions would, and handles error
7363 conditions in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +00007364
Dan Gohmanb6324c12007-10-15 20:30:11 +00007365</div>
7366
Dan Gohman911fa902011-05-23 21:13:03 +00007367<!-- _______________________________________________________________________ -->
7368<h4>
7369 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
7370</h4>
7371
7372<div>
7373
7374<h5>Syntax:</h5>
7375<p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
7376 floating point or vector of floating point type. Not all targets support all
7377 types however.</p>
7378
7379<pre>
7380 declare float @llvm.exp.f32(float %Val)
7381 declare double @llvm.exp.f64(double %Val)
7382 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7383 declare fp128 @llvm.exp.f128(fp128 %Val)
7384 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7385</pre>
7386
7387<h5>Overview:</h5>
7388<p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7389
7390<h5>Arguments:</h5>
7391<p>The argument and return value are floating point numbers of the same
7392 type.</p>
7393
7394<h5>Semantics:</h5>
7395<p>This function returns the same values as the libm <tt>exp</tt> functions
7396 would, and handles error conditions in the same way.</p>
7397
7398</div>
7399
7400<!-- _______________________________________________________________________ -->
7401<h4>
7402 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7403</h4>
7404
7405<div>
7406
7407<h5>Syntax:</h5>
7408<p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7409 floating point or vector of floating point type. Not all targets support all
7410 types however.</p>
7411
7412<pre>
7413 declare float @llvm.log.f32(float %Val)
7414 declare double @llvm.log.f64(double %Val)
7415 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7416 declare fp128 @llvm.log.f128(fp128 %Val)
7417 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7418</pre>
7419
7420<h5>Overview:</h5>
7421<p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7422
7423<h5>Arguments:</h5>
7424<p>The argument and return value are floating point numbers of the same
7425 type.</p>
7426
7427<h5>Semantics:</h5>
7428<p>This function returns the same values as the libm <tt>log</tt> functions
7429 would, and handles error conditions in the same way.</p>
7430
Nick Lewyckycd196f62011-10-31 01:32:21 +00007431</div>
7432
7433<!-- _______________________________________________________________________ -->
Cameron Zwarichf03fa182011-07-08 21:39:21 +00007434<h4>
7435 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7436</h4>
7437
7438<div>
7439
7440<h5>Syntax:</h5>
7441<p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7442 floating point or vector of floating point type. Not all targets support all
7443 types however.</p>
7444
7445<pre>
7446 declare float @llvm.fma.f32(float %a, float %b, float %c)
7447 declare double @llvm.fma.f64(double %a, double %b, double %c)
7448 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7449 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7450 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7451</pre>
7452
7453<h5>Overview:</h5>
Cameron Zwaricha32fd212011-07-08 22:13:55 +00007454<p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
Cameron Zwarichf03fa182011-07-08 21:39:21 +00007455 operation.</p>
7456
7457<h5>Arguments:</h5>
7458<p>The argument and return value are floating point numbers of the same
7459 type.</p>
7460
7461<h5>Semantics:</h5>
7462<p>This function returns the same values as the libm <tt>fma</tt> functions
7463 would.</p>
7464
Dan Gohman911fa902011-05-23 21:13:03 +00007465</div>
7466
NAKAMURA Takumia35cdd62011-10-31 13:04:26 +00007467</div>
7468
Andrew Lenharth1d463522005-05-03 18:01:48 +00007469<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007470<h3>
Nate Begeman0f223bb2006-01-13 23:26:38 +00007471 <a name="int_manip">Bit Manipulation Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007472</h3>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007473
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007474<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007475
7476<p>LLVM provides intrinsics for a few important bit manipulation operations.
7477 These allow efficient code generation for some algorithms.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007478
Andrew Lenharth1d463522005-05-03 18:01:48 +00007479<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007480<h4>
Reid Spencer96a5f022007-04-04 02:42:35 +00007481 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007482</h4>
Nate Begeman0f223bb2006-01-13 23:26:38 +00007483
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007484<div>
Nate Begeman0f223bb2006-01-13 23:26:38 +00007485
7486<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +00007487<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007488 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7489
Nate Begeman0f223bb2006-01-13 23:26:38 +00007490<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +00007491 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
7492 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
7493 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman0f223bb2006-01-13 23:26:38 +00007494</pre>
7495
7496<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007497<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7498 values with an even number of bytes (positive multiple of 16 bits). These
7499 are useful for performing operations on data that is not in the target's
7500 native byte order.</p>
Nate Begeman0f223bb2006-01-13 23:26:38 +00007501
7502<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007503<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7504 and low byte of the input i16 swapped. Similarly,
7505 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7506 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7507 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7508 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7509 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7510 more, respectively).</p>
Nate Begeman0f223bb2006-01-13 23:26:38 +00007511
7512</div>
7513
7514<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007515<h4>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +00007516 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007517</h4>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007518
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007519<div>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007520
7521<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +00007522<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007523 width, or on any vector with integer elements. Not all targets support all
7524 bit widths or vector types, however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007525
Andrew Lenharth1d463522005-05-03 18:01:48 +00007526<pre>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007527 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +00007528 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +00007529 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +00007530 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
7531 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007532 declare &lt;2 x i32&gt; @llvm.ctpop.v2i32(&lt;2 x i32&gt; &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +00007533</pre>
7534
7535<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007536<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7537 in a value.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007538
7539<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007540<p>The only argument is the value to be counted. The argument may be of any
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007541 integer type, or a vector with integer elements.
7542 The return type must match the argument type.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007543
7544<h5>Semantics:</h5>
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007545<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7546 element of a vector.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007547
Andrew Lenharth1d463522005-05-03 18:01:48 +00007548</div>
7549
7550<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007551<h4>
Chris Lattnerb748c672006-01-16 22:34:14 +00007552 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007553</h4>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007554
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007555<div>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007556
7557<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007558<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007559 integer bit width, or any vector whose elements are integers. Not all
7560 targets support all bit widths or vector types, however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007561
Andrew Lenharth1d463522005-05-03 18:01:48 +00007562<pre>
Chandler Carruthf6bb2782011-12-12 04:36:04 +00007563 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7564 declare i16 @llvm.ctlz.i16 (i16 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7565 declare i32 @llvm.ctlz.i32 (i32 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7566 declare i64 @llvm.ctlz.i64 (i64 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7567 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7568 declase &lt;2 x i32&gt; @llvm.ctlz.v2i32(&lt;2 x i32&gt; &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +00007569</pre>
7570
7571<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007572<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7573 leading zeros in a variable.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007574
7575<h5>Arguments:</h5>
Chandler Carruthf6bb2782011-12-12 04:36:04 +00007576<p>The first argument is the value to be counted. This argument may be of any
7577 integer type, or a vectory with integer element type. The return type
7578 must match the first argument type.</p>
7579
7580<p>The second argument must be a constant and is a flag to indicate whether the
7581 intrinsic should ensure that a zero as the first argument produces a defined
7582 result. Historically some architectures did not provide a defined result for
7583 zero values as efficiently, and many algorithms are now predicated on
7584 avoiding zero-value inputs.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007585
7586<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007587<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
Chandler Carruthf6bb2782011-12-12 04:36:04 +00007588 zeros in a variable, or within each element of the vector.
7589 If <tt>src == 0</tt> then the result is the size in bits of the type of
7590 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7591 For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharth1d463522005-05-03 18:01:48 +00007592
Andrew Lenharth1d463522005-05-03 18:01:48 +00007593</div>
Chris Lattner3b4f4372004-06-11 02:28:03 +00007594
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007595<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007596<h4>
Chris Lattnerb748c672006-01-16 22:34:14 +00007597 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007598</h4>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007599
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007600<div>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007601
7602<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007603<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007604 integer bit width, or any vector of integer elements. Not all targets
7605 support all bit widths or vector types, however.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007606
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007607<pre>
Chandler Carruthf6bb2782011-12-12 04:36:04 +00007608 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7609 declare i16 @llvm.cttz.i16 (i16 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7610 declare i32 @llvm.cttz.i32 (i32 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7611 declare i64 @llvm.cttz.i64 (i64 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7612 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7613 declase &lt;2 x i32&gt; @llvm.cttz.v2i32(&lt;2 x i32&gt; &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007614</pre>
7615
7616<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007617<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7618 trailing zeros.</p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007619
7620<h5>Arguments:</h5>
Chandler Carruthf6bb2782011-12-12 04:36:04 +00007621<p>The first argument is the value to be counted. This argument may be of any
7622 integer type, or a vectory with integer element type. The return type
7623 must match the first argument type.</p>
7624
7625<p>The second argument must be a constant and is a flag to indicate whether the
7626 intrinsic should ensure that a zero as the first argument produces a defined
7627 result. Historically some architectures did not provide a defined result for
7628 zero values as efficiently, and many algorithms are now predicated on
7629 avoiding zero-value inputs.</p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007630
7631<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007632<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
Owen Anderson2f37bdc2011-07-01 21:52:38 +00007633 zeros in a variable, or within each element of a vector.
Chandler Carruthf6bb2782011-12-12 04:36:04 +00007634 If <tt>src == 0</tt> then the result is the size in bits of the type of
7635 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7636 For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007637
Chris Lattnerefa20fa2005-05-15 19:39:26 +00007638</div>
7639
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007640</div>
7641
Bill Wendlingfd2bd722009-02-08 04:04:40 +00007642<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007643<h3>
Bill Wendlingfd2bd722009-02-08 04:04:40 +00007644 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007645</h3>
Bill Wendlingfd2bd722009-02-08 04:04:40 +00007646
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007647<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007648
7649<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingfd2bd722009-02-08 04:04:40 +00007650
Bill Wendlingf4d70622009-02-08 01:40:31 +00007651<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007652<h4>
7653 <a name="int_sadd_overflow">
7654 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7655 </a>
7656</h4>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007657
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007658<div>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007659
7660<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007661<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007662 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007663
7664<pre>
7665 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7666 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7667 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7668</pre>
7669
7670<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007671<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007672 a signed addition of the two arguments, and indicate whether an overflow
7673 occurred during the signed summation.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007674
7675<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007676<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007677 be of integer types of any bit width, but they must have the same bit
7678 width. The second element of the result structure must be of
7679 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7680 undergo signed addition.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007681
7682<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007683<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007684 a signed addition of the two variables. They return a structure &mdash; the
7685 first element of which is the signed summation, and the second element of
7686 which is a bit specifying if the signed summation resulted in an
7687 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007688
7689<h5>Examples:</h5>
7690<pre>
7691 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7692 %sum = extractvalue {i32, i1} %res, 0
7693 %obit = extractvalue {i32, i1} %res, 1
7694 br i1 %obit, label %overflow, label %normal
7695</pre>
7696
7697</div>
7698
7699<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007700<h4>
7701 <a name="int_uadd_overflow">
7702 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7703 </a>
7704</h4>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007705
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007706<div>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007707
7708<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007709<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007710 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007711
7712<pre>
7713 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7714 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7715 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7716</pre>
7717
7718<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007719<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007720 an unsigned addition of the two arguments, and indicate whether a carry
7721 occurred during the unsigned summation.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007722
7723<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007724<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007725 be of integer types of any bit width, but they must have the same bit
7726 width. The second element of the result structure must be of
7727 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7728 undergo unsigned addition.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007729
7730<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007731<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007732 an unsigned addition of the two arguments. They return a structure &mdash;
7733 the first element of which is the sum, and the second element of which is a
7734 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007735
7736<h5>Examples:</h5>
7737<pre>
7738 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7739 %sum = extractvalue {i32, i1} %res, 0
7740 %obit = extractvalue {i32, i1} %res, 1
7741 br i1 %obit, label %carry, label %normal
7742</pre>
7743
7744</div>
7745
7746<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007747<h4>
7748 <a name="int_ssub_overflow">
7749 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7750 </a>
7751</h4>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007752
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007753<div>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007754
7755<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007756<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007757 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007758
7759<pre>
7760 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7761 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7762 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7763</pre>
7764
7765<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007766<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007767 a signed subtraction of the two arguments, and indicate whether an overflow
7768 occurred during the signed subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007769
7770<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007771<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007772 be of integer types of any bit width, but they must have the same bit
7773 width. The second element of the result structure must be of
7774 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7775 undergo signed subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007776
7777<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007778<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007779 a signed subtraction of the two arguments. They return a structure &mdash;
7780 the first element of which is the subtraction, and the second element of
7781 which is a bit specifying if the signed subtraction resulted in an
7782 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007783
7784<h5>Examples:</h5>
7785<pre>
7786 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7787 %sum = extractvalue {i32, i1} %res, 0
7788 %obit = extractvalue {i32, i1} %res, 1
7789 br i1 %obit, label %overflow, label %normal
7790</pre>
7791
7792</div>
7793
7794<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007795<h4>
7796 <a name="int_usub_overflow">
7797 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7798 </a>
7799</h4>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007800
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007801<div>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007802
7803<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007804<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007805 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007806
7807<pre>
7808 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7809 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7810 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7811</pre>
7812
7813<h5>Overview:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007814<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007815 an unsigned subtraction of the two arguments, and indicate whether an
7816 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007817
7818<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007819<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007820 be of integer types of any bit width, but they must have the same bit
7821 width. The second element of the result structure must be of
7822 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7823 undergo unsigned subtraction.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007824
7825<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007826<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007827 an unsigned subtraction of the two arguments. They return a structure &mdash;
7828 the first element of which is the subtraction, and the second element of
7829 which is a bit specifying if the unsigned subtraction resulted in an
7830 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007831
7832<h5>Examples:</h5>
7833<pre>
7834 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7835 %sum = extractvalue {i32, i1} %res, 0
7836 %obit = extractvalue {i32, i1} %res, 1
7837 br i1 %obit, label %overflow, label %normal
7838</pre>
7839
7840</div>
7841
7842<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007843<h4>
7844 <a name="int_smul_overflow">
7845 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7846 </a>
7847</h4>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007848
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007849<div>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007850
7851<h5>Syntax:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007852<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007853 on any integer bit width.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007854
7855<pre>
7856 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7857 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7858 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7859</pre>
7860
7861<h5>Overview:</h5>
7862
7863<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007864 a signed multiplication of the two arguments, and indicate whether an
7865 overflow occurred during the signed multiplication.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007866
7867<h5>Arguments:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007868<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007869 be of integer types of any bit width, but they must have the same bit
7870 width. The second element of the result structure must be of
7871 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7872 undergo signed multiplication.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007873
7874<h5>Semantics:</h5>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007875<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007876 a signed multiplication of the two arguments. They return a structure &mdash;
7877 the first element of which is the multiplication, and the second element of
7878 which is a bit specifying if the signed multiplication resulted in an
7879 overflow.</p>
Bill Wendlingf4d70622009-02-08 01:40:31 +00007880
7881<h5>Examples:</h5>
7882<pre>
7883 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7884 %sum = extractvalue {i32, i1} %res, 0
7885 %obit = extractvalue {i32, i1} %res, 1
7886 br i1 %obit, label %overflow, label %normal
7887</pre>
7888
Reid Spencer5bf54c82007-04-11 23:23:49 +00007889</div>
7890
Bill Wendlingb9a73272009-02-08 23:00:09 +00007891<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007892<h4>
7893 <a name="int_umul_overflow">
7894 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
7895 </a>
7896</h4>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007897
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007898<div>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007899
7900<h5>Syntax:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007901<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007902 on any integer bit width.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007903
7904<pre>
7905 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
7906 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7907 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
7908</pre>
7909
7910<h5>Overview:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007911<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007912 a unsigned multiplication of the two arguments, and indicate whether an
7913 overflow occurred during the unsigned multiplication.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007914
7915<h5>Arguments:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007916<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007917 be of integer types of any bit width, but they must have the same bit
7918 width. The second element of the result structure must be of
7919 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7920 undergo unsigned multiplication.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007921
7922<h5>Semantics:</h5>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007923<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingd9a66f72009-07-20 02:29:24 +00007924 an unsigned multiplication of the two arguments. They return a structure
7925 &mdash; the first element of which is the multiplication, and the second
7926 element of which is a bit specifying if the unsigned multiplication resulted
7927 in an overflow.</p>
Bill Wendlingb9a73272009-02-08 23:00:09 +00007928
7929<h5>Examples:</h5>
7930<pre>
7931 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7932 %sum = extractvalue {i32, i1} %res, 0
7933 %obit = extractvalue {i32, i1} %res, 1
7934 br i1 %obit, label %overflow, label %normal
7935</pre>
7936
7937</div>
7938
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007939</div>
7940
Chris Lattner941515c2004-01-06 05:31:32 +00007941<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007942<h3>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007943 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007944</h3>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007945
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007946<div>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007947
Chris Lattner022a9fb2010-03-15 04:12:21 +00007948<p>Half precision floating point is a storage-only format. This means that it is
7949 a dense encoding (in memory) but does not support computation in the
7950 format.</p>
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00007951
Chris Lattner022a9fb2010-03-15 04:12:21 +00007952<p>This means that code must first load the half-precision floating point
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00007953 value as an i16, then convert it to float with <a
7954 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
7955 Computation can then be performed on the float value (including extending to
Chris Lattner022a9fb2010-03-15 04:12:21 +00007956 double etc). To store the value back to memory, it is first converted to
7957 float if needed, then converted to i16 with
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00007958 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
7959 storing as an i16 value.</p>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007960
7961<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007962<h4>
7963 <a name="int_convert_to_fp16">
7964 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
7965 </a>
7966</h4>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007967
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00007968<div>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007969
7970<h5>Syntax:</h5>
7971<pre>
7972 declare i16 @llvm.convert.to.fp16(f32 %a)
7973</pre>
7974
7975<h5>Overview:</h5>
7976<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7977 a conversion from single precision floating point format to half precision
7978 floating point format.</p>
7979
7980<h5>Arguments:</h5>
7981<p>The intrinsic function contains single argument - the value to be
7982 converted.</p>
7983
7984<h5>Semantics:</h5>
7985<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7986 a conversion from single precision floating point format to half precision
Chris Lattner022a9fb2010-03-15 04:12:21 +00007987 floating point format. The return value is an <tt>i16</tt> which
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00007988 contains the converted number.</p>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00007989
7990<h5>Examples:</h5>
7991<pre>
7992 %res = call i16 @llvm.convert.to.fp16(f32 %a)
7993 store i16 %res, i16* @x, align 2
7994</pre>
7995
7996</div>
7997
7998<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00007999<h4>
8000 <a name="int_convert_from_fp16">
8001 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
8002 </a>
8003</h4>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00008004
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008005<div>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00008006
8007<h5>Syntax:</h5>
8008<pre>
8009 declare f32 @llvm.convert.from.fp16(i16 %a)
8010</pre>
8011
8012<h5>Overview:</h5>
8013<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
8014 a conversion from half precision floating point format to single precision
8015 floating point format.</p>
8016
8017<h5>Arguments:</h5>
8018<p>The intrinsic function contains single argument - the value to be
8019 converted.</p>
8020
8021<h5>Semantics:</h5>
8022<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner022a9fb2010-03-15 04:12:21 +00008023 conversion from half single precision floating point format to single
Chris Lattnerbbd8bd32010-03-14 23:03:31 +00008024 precision floating point format. The input half-float value is represented by
8025 an <tt>i16</tt> value.</p>
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00008026
8027<h5>Examples:</h5>
8028<pre>
8029 %a = load i16* @x, align 2
8030 %res = call f32 @llvm.convert.from.fp16(i16 %a)
8031</pre>
8032
8033</div>
8034
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008035</div>
8036
Anton Korobeynikovcd4dd9c2010-03-14 18:42:47 +00008037<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008038<h3>
Chris Lattner941515c2004-01-06 05:31:32 +00008039 <a name="int_debugger">Debugger Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008040</h3>
Chris Lattner941515c2004-01-06 05:31:32 +00008041
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008042<div>
Chris Lattner941515c2004-01-06 05:31:32 +00008043
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008044<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
8045 prefix), are described in
8046 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
8047 Level Debugging</a> document.</p>
8048
8049</div>
Chris Lattner941515c2004-01-06 05:31:32 +00008050
Jim Laskey2211f492007-03-14 19:31:19 +00008051<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008052<h3>
Jim Laskey2211f492007-03-14 19:31:19 +00008053 <a name="int_eh">Exception Handling Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008054</h3>
Jim Laskey2211f492007-03-14 19:31:19 +00008055
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008056<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008057
8058<p>The LLVM exception handling intrinsics (which all start with
8059 <tt>llvm.eh.</tt> prefix), are described in
8060 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
8061 Handling</a> document.</p>
8062
Jim Laskey2211f492007-03-14 19:31:19 +00008063</div>
8064
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008065<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008066<h3>
Duncan Sandsa0984362011-09-06 13:37:06 +00008067 <a name="int_trampoline">Trampoline Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008068</h3>
Duncan Sands644f9172007-07-27 12:58:54 +00008069
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008070<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008071
Duncan Sandsa0984362011-09-06 13:37:06 +00008072<p>These intrinsics make it possible to excise one parameter, marked with
Dan Gohman3770af52010-07-02 23:18:08 +00008073 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
8074 The result is a callable
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008075 function pointer lacking the nest parameter - the caller does not need to
8076 provide a value for it. Instead, the value to use is stored in advance in a
8077 "trampoline", a block of memory usually allocated on the stack, which also
8078 contains code to splice the nest value into the argument list. This is used
8079 to implement the GCC nested function address extension.</p>
8080
8081<p>For example, if the function is
8082 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
8083 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
8084 follows:</p>
8085
Benjamin Kramer79698be2010-07-13 12:26:09 +00008086<pre class="doc_code">
Duncan Sands86e01192007-09-11 14:10:23 +00008087 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
8088 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Duncan Sandsa0984362011-09-06 13:37:06 +00008089 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
8090 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
Duncan Sands86e01192007-09-11 14:10:23 +00008091 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands644f9172007-07-27 12:58:54 +00008092</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008093
Dan Gohmand6a6f612010-05-28 17:07:41 +00008094<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
8095 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008096
Duncan Sands644f9172007-07-27 12:58:54 +00008097<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008098<h4>
8099 <a name="int_it">
8100 '<tt>llvm.init.trampoline</tt>' Intrinsic
8101 </a>
8102</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008103
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008104<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008105
Duncan Sands644f9172007-07-27 12:58:54 +00008106<h5>Syntax:</h5>
8107<pre>
Duncan Sandsa0984362011-09-06 13:37:06 +00008108 declare void @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands644f9172007-07-27 12:58:54 +00008109</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008110
Duncan Sands644f9172007-07-27 12:58:54 +00008111<h5>Overview:</h5>
Duncan Sandsa0984362011-09-06 13:37:06 +00008112<p>This fills the memory pointed to by <tt>tramp</tt> with executable code,
8113 turning it into a trampoline.</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008114
Duncan Sands644f9172007-07-27 12:58:54 +00008115<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008116<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
8117 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
8118 sufficiently aligned block of memory; this memory is written to by the
8119 intrinsic. Note that the size and the alignment are target-specific - LLVM
8120 currently provides no portable way of determining them, so a front-end that
8121 generates this intrinsic needs to have some target-specific knowledge.
8122 The <tt>func</tt> argument must hold a function bitcast to
8123 an <tt>i8*</tt>.</p>
8124
Duncan Sands644f9172007-07-27 12:58:54 +00008125<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008126<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sandsa0984362011-09-06 13:37:06 +00008127 dependent code, turning it into a function. Then <tt>tramp</tt> needs to be
8128 passed to <a href="#int_at">llvm.adjust.trampoline</a> to get a pointer
8129 which can be <a href="#int_trampoline">bitcast (to a new function) and
8130 called</a>. The new function's signature is the same as that of
8131 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
8132 removed. At most one such <tt>nest</tt> argument is allowed, and it must be of
8133 pointer type. Calling the new function is equivalent to calling <tt>func</tt>
8134 with the same argument list, but with <tt>nval</tt> used for the missing
8135 <tt>nest</tt> argument. If, after calling <tt>llvm.init.trampoline</tt>, the
8136 memory pointed to by <tt>tramp</tt> is modified, then the effect of any later call
8137 to the returned function pointer is undefined.</p>
8138</div>
8139
8140<!-- _______________________________________________________________________ -->
8141<h4>
8142 <a name="int_at">
8143 '<tt>llvm.adjust.trampoline</tt>' Intrinsic
8144 </a>
8145</h4>
8146
8147<div>
8148
8149<h5>Syntax:</h5>
8150<pre>
8151 declare i8* @llvm.adjust.trampoline(i8* &lt;tramp&gt;)
8152</pre>
8153
8154<h5>Overview:</h5>
8155<p>This performs any required machine-specific adjustment to the address of a
8156 trampoline (passed as <tt>tramp</tt>).</p>
8157
8158<h5>Arguments:</h5>
8159<p><tt>tramp</tt> must point to a block of memory which already has trampoline code
8160 filled in by a previous call to <a href="#int_it"><tt>llvm.init.trampoline</tt>
8161 </a>.</p>
8162
8163<h5>Semantics:</h5>
8164<p>On some architectures the address of the code to be executed needs to be
8165 different to the address where the trampoline is actually stored. This
8166 intrinsic returns the executable address corresponding to <tt>tramp</tt>
8167 after performing the required machine specific adjustments.
8168 The pointer returned can then be <a href="#int_trampoline"> bitcast and
8169 executed</a>.
8170</p>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008171
Duncan Sands644f9172007-07-27 12:58:54 +00008172</div>
8173
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008174</div>
8175
Duncan Sands644f9172007-07-27 12:58:54 +00008176<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008177<h3>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008178 <a name="int_memorymarkers">Memory Use Markers</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008179</h3>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008180
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008181<div>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008182
8183<p>This class of intrinsics exists to information about the lifetime of memory
8184 objects and ranges where variables are immutable.</p>
8185
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008186<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008187<h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008188 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008189</h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008190
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008191<div>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008192
8193<h5>Syntax:</h5>
8194<pre>
8195 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8196</pre>
8197
8198<h5>Overview:</h5>
8199<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
8200 object's lifetime.</p>
8201
8202<h5>Arguments:</h5>
Nick Lewycky9bc89042009-10-13 07:57:33 +00008203<p>The first argument is a constant integer representing the size of the
8204 object, or -1 if it is variable sized. The second argument is a pointer to
8205 the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008206
8207<h5>Semantics:</h5>
8208<p>This intrinsic indicates that before this point in the code, the value of the
8209 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
Nick Lewyckyd20fd592009-10-27 16:56:58 +00008210 never be used and has an undefined value. A load from the pointer that
8211 precedes this intrinsic can be replaced with
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008212 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
8213
8214</div>
8215
8216<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008217<h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008218 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008219</h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008220
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008221<div>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008222
8223<h5>Syntax:</h5>
8224<pre>
8225 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8226</pre>
8227
8228<h5>Overview:</h5>
8229<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
8230 object's lifetime.</p>
8231
8232<h5>Arguments:</h5>
Nick Lewycky9bc89042009-10-13 07:57:33 +00008233<p>The first argument is a constant integer representing the size of the
8234 object, or -1 if it is variable sized. The second argument is a pointer to
8235 the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008236
8237<h5>Semantics:</h5>
8238<p>This intrinsic indicates that after this point in the code, the value of the
8239 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8240 never be used and has an undefined value. Any stores into the memory object
8241 following this intrinsic may be removed as dead.
8242
8243</div>
8244
8245<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008246<h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008247 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008248</h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008249
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008250<div>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008251
8252<h5>Syntax:</h5>
8253<pre>
Nick Lewycky2965d3e2010-11-30 04:13:41 +00008254 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008255</pre>
8256
8257<h5>Overview:</h5>
8258<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
8259 a memory object will not change.</p>
8260
8261<h5>Arguments:</h5>
Nick Lewycky9bc89042009-10-13 07:57:33 +00008262<p>The first argument is a constant integer representing the size of the
8263 object, or -1 if it is variable sized. The second argument is a pointer to
8264 the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008265
8266<h5>Semantics:</h5>
8267<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
8268 the return value, the referenced memory location is constant and
8269 unchanging.</p>
8270
8271</div>
8272
8273<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008274<h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008275 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008276</h4>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008277
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008278<div>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008279
8280<h5>Syntax:</h5>
8281<pre>
8282 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8283</pre>
8284
8285<h5>Overview:</h5>
8286<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8287 a memory object are mutable.</p>
8288
8289<h5>Arguments:</h5>
8290<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky9bc89042009-10-13 07:57:33 +00008291 The second argument is a constant integer representing the size of the
8292 object, or -1 if it is variable sized and the third argument is a pointer
8293 to the object.</p>
Nick Lewycky6f7d8342009-10-13 07:03:23 +00008294
8295<h5>Semantics:</h5>
8296<p>This intrinsic indicates that the memory is mutable again.</p>
8297
8298</div>
8299
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008300</div>
8301
Andrew Lenharth9b254ee2008-02-16 01:24:58 +00008302<!-- ======================================================================= -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008303<h3>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008304 <a name="int_general">General Intrinsics</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008305</h3>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008306
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008307<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008308
8309<p>This class of intrinsics is designed to be generic and has no specific
8310 purpose.</p>
8311
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008312<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008313<h4>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008314 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008315</h4>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008316
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008317<div>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008318
8319<h5>Syntax:</h5>
8320<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00008321 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 +00008322</pre>
8323
8324<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008325<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008326
8327<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008328<p>The first argument is a pointer to a value, the second is a pointer to a
8329 global string, the third is a pointer to a global string which is the source
8330 file name, and the last argument is the line number.</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008331
8332<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008333<p>This intrinsic allows annotation of local variables with arbitrary strings.
8334 This can be useful for special purpose optimizations that want to look for
John Criswellf0d536a2011-08-19 16:57:55 +00008335 these annotations. These have no other defined use; they are ignored by code
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008336 generation and optimization.</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008337
Tanya Lattnercb1b9602007-06-15 20:50:54 +00008338</div>
8339
Tanya Lattner293c0372007-09-21 22:59:12 +00008340<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008341<h4>
Tanya Lattner0186a652007-09-21 23:57:59 +00008342 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008343</h4>
Tanya Lattner293c0372007-09-21 22:59:12 +00008344
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008345<div>
Tanya Lattner293c0372007-09-21 22:59:12 +00008346
8347<h5>Syntax:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008348<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8349 any integer bit width.</p>
8350
Tanya Lattner293c0372007-09-21 22:59:12 +00008351<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00008352 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8353 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8354 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8355 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8356 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 +00008357</pre>
8358
8359<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008360<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattner293c0372007-09-21 22:59:12 +00008361
8362<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008363<p>The first argument is an integer value (result of some expression), the
8364 second is a pointer to a global string, the third is a pointer to a global
8365 string which is the source file name, and the last argument is the line
8366 number. It returns the value of the first argument.</p>
Tanya Lattner293c0372007-09-21 22:59:12 +00008367
8368<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008369<p>This intrinsic allows annotations to be put on arbitrary expressions with
8370 arbitrary strings. This can be useful for special purpose optimizations that
John Criswellf0d536a2011-08-19 16:57:55 +00008371 want to look for these annotations. These have no other defined use; they
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008372 are ignored by code generation and optimization.</p>
Tanya Lattner293c0372007-09-21 22:59:12 +00008373
Tanya Lattner293c0372007-09-21 22:59:12 +00008374</div>
Jim Laskey2211f492007-03-14 19:31:19 +00008375
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008376<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008377<h4>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008378 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008379</h4>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008380
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008381<div>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008382
8383<h5>Syntax:</h5>
8384<pre>
8385 declare void @llvm.trap()
8386</pre>
8387
8388<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008389<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008390
8391<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008392<p>None.</p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008393
8394<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008395<p>This intrinsics is lowered to the target dependent trap instruction. If the
8396 target does not have a trap instruction, this intrinsic will be lowered to
8397 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008398
Anton Korobeynikov06cbb652008-01-15 22:31:34 +00008399</div>
8400
Bill Wendling14313312008-11-19 05:56:17 +00008401<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008402<h4>
Misha Brukman50de2b22008-11-22 23:55:29 +00008403 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008404</h4>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008405
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008406<div>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008407
Bill Wendling14313312008-11-19 05:56:17 +00008408<h5>Syntax:</h5>
8409<pre>
Dan Gohmand6a6f612010-05-28 17:07:41 +00008410 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling14313312008-11-19 05:56:17 +00008411</pre>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008412
Bill Wendling14313312008-11-19 05:56:17 +00008413<h5>Overview:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008414<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8415 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8416 ensure that it is placed on the stack before local variables.</p>
8417
Bill Wendling14313312008-11-19 05:56:17 +00008418<h5>Arguments:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008419<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8420 arguments. The first argument is the value loaded from the stack
8421 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8422 that has enough space to hold the value of the guard.</p>
8423
Bill Wendling14313312008-11-19 05:56:17 +00008424<h5>Semantics:</h5>
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008425<p>This intrinsic causes the prologue/epilogue inserter to force the position of
8426 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8427 stack. This is to ensure that if a local variable on the stack is
8428 overwritten, it will destroy the value of the guard. When the function exits,
Bill Wendling6bbe0912010-10-27 01:07:41 +00008429 the guard on the stack is checked against the original guard. If they are
Bill Wendlingd9a66f72009-07-20 02:29:24 +00008430 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8431 function.</p>
8432
Bill Wendling14313312008-11-19 05:56:17 +00008433</div>
8434
Eric Christopher73484322009-11-30 08:03:53 +00008435<!-- _______________________________________________________________________ -->
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008436<h4>
Eric Christopher73484322009-11-30 08:03:53 +00008437 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
NAKAMURA Takumifc8d9302011-04-18 23:59:50 +00008438</h4>
Eric Christopher73484322009-11-30 08:03:53 +00008439
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008440<div>
Eric Christopher73484322009-11-30 08:03:53 +00008441
8442<h5>Syntax:</h5>
8443<pre>
Nuno Lopes01547b32012-05-09 15:52:43 +00008444 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;min&gt;, i32 &lt;runtime&gt;)
8445 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;min&gt;, i32 &lt;runtime&gt;)
Eric Christopher73484322009-11-30 08:03:53 +00008446</pre>
8447
8448<h5>Overview:</h5>
Bill Wendling6bbe0912010-10-27 01:07:41 +00008449<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8450 the optimizers to determine at compile time whether a) an operation (like
8451 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8452 runtime check for overflow isn't necessary. An object in this context means
8453 an allocation of a specific class, structure, array, or other object.</p>
Eric Christopher73484322009-11-30 08:03:53 +00008454
8455<h5>Arguments:</h5>
Nuno Lopes01547b32012-05-09 15:52:43 +00008456<p>The <tt>llvm.objectsize</tt> intrinsic takes three arguments. The first
Eric Christopher31e39bd2009-12-23 00:29:49 +00008457 argument is a pointer to or into the <tt>object</tt>. The second argument
Nuno Lopes01547b32012-05-09 15:52:43 +00008458 is a boolean and determines whether <tt>llvm.objectsize</tt> returns 0 (if true)
8459 or -1 (if false) when the object size is unknown.
8460 The third argument, <tt>runtime</tt>, indicates whether the compiler is allowed
8461 to return a non-constant value. The higher the value, the higher the potential
8462 run-time performance impact.
8463 The second and third arguments only accepts constants.</p>
Eric Christopher31e39bd2009-12-23 00:29:49 +00008464
Eric Christopher73484322009-11-30 08:03:53 +00008465<h5>Semantics:</h5>
Nuno Lopes01547b32012-05-09 15:52:43 +00008466<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to a constant representing
8467 the size of the object concerned. If the size cannot be determined at compile
8468 time, <tt>llvm.objectsize</tt> either returns <tt>i32/i64 -1 or 0</tt>
8469 (depending on the <tt>min</tt> argument) if <tt>runtime</tt> is 0, or a run-time
8470 value (if <tt>runtime</tt> &gt; 0 and an expression could be generated).</p>
Eric Christopher73484322009-11-30 08:03:53 +00008471
8472</div>
Jakub Staszak5fef7922011-12-04 18:29:26 +00008473<!-- _______________________________________________________________________ -->
8474<h4>
8475 <a name="int_expect">'<tt>llvm.expect</tt>' Intrinsic</a>
8476</h4>
Eric Christopher73484322009-11-30 08:03:53 +00008477
Jakub Staszak5fef7922011-12-04 18:29:26 +00008478<div>
8479
8480<h5>Syntax:</h5>
8481<pre>
8482 declare i32 @llvm.expect.i32(i32 &lt;val&gt;, i32 &lt;expected_val&gt;)
8483 declare i64 @llvm.expect.i64(i64 &lt;val&gt;, i64 &lt;expected_val&gt;)
8484</pre>
8485
8486<h5>Overview:</h5>
8487<p>The <tt>llvm.expect</tt> intrinsic provides information about expected (the
8488 most probable) value of <tt>val</tt>, which can be used by optimizers.</p>
8489
8490<h5>Arguments:</h5>
8491<p>The <tt>llvm.expect</tt> intrinsic takes two arguments. The first
8492 argument is a value. The second argument is an expected value, this needs to
8493 be a constant value, variables are not allowed.</p>
8494
8495<h5>Semantics:</h5>
8496<p>This intrinsic is lowered to the <tt>val</tt>.</p>
NAKAMURA Takumiaa3d6242011-04-23 00:30:22 +00008497</div>
8498
8499</div>
8500
Jakub Staszak5fef7922011-12-04 18:29:26 +00008501</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +00008502<!-- *********************************************************************** -->
Chris Lattner2f7c9632001-06-06 20:29:01 +00008503<hr>
Misha Brukmanc501f552004-03-01 17:47:27 +00008504<address>
8505 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +00008506 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmanc501f552004-03-01 17:47:27 +00008507 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +00008508 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukmanc501f552004-03-01 17:47:27 +00008509
8510 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
NAKAMURA Takumica46f5a2011-04-09 02:13:37 +00008511 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Misha Brukmanc501f552004-03-01 17:47:27 +00008512 Last modified: $Date$
8513</address>
Chris Lattnerb8f816e2008-01-04 04:33:49 +00008514
Misha Brukman76307852003-11-08 01:05:38 +00008515</body>
8516</html>