Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / b5bb29c71fccc3c36c9175764303ef75c4fc73ee / . / docs / LangRef.html
blob: b20f085176edc5abb4a0d083909b3dba8f80c7d7 [file] [log] [blame]
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencer3921c742004-08-26 20:44:00 +0000[diff] [blame]6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]8 <meta name="description"
Reid Spencer3921c742004-08-26 20:44:00 +0000[diff] [blame]9 content="LLVM Assembly Language Reference Manual.">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]12
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]13<body>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]14
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]15<h1>LLVM Language Reference Manual</h1>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]16<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a>
24 <ol>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]27 <li><a href="#linkage_linker_private_weak">'<tt>linker_private_weak</tt>' Linkage</a></li>
Bill Wendling55ae5152010-08-20 22:05:50 +0000[diff] [blame]28 <li><a href="#linkage_linker_private_weak_def_auto">'<tt>linker_private_weak_def_auto</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]29 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
30 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
31 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
32 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
33 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
35 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner5a2d8752009-10-10 18:26:06 +0000[diff] [blame]36 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]37 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
38 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
39 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
40 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000[diff] [blame]41 </ol>
42 </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]43 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]44 <li><a href="#namedtypes">Named Types</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]45 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]46 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]47 <li><a href="#aliasstructure">Aliases</a></li>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]48 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]49 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]50 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]51 <li><a href="#gc">Garbage Collector Names</a></li>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]52 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]53 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]54 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]55 <li><a href="#volatile">Volatile Memory Accesses</a></li>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]56 <li><a href="#memmodel">Memory Model for Concurrent Operations</a></li>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]57 <li><a href="#ordering">Atomic Memory Ordering Constraints</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]58 </ol>
59 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]60 <li><a href="#typesystem">Type System</a>
61 <ol>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]62 <li><a href="#t_classifications">Type Classifications</a></li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]63 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]64 <ol>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]65 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]66 <li><a href="#t_floating">Floating Point Types</a></li>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]67 <li><a href="#t_x86mmx">X86mmx Type</a></li>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]68 <li><a href="#t_void">Void Type</a></li>
69 <li><a href="#t_label">Label Type</a></li>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]70 <li><a href="#t_metadata">Metadata Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]71 </ol>
72 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]73 <li><a href="#t_derived">Derived Types</a>
74 <ol>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]75 <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 Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]79 <li><a href="#t_opaque">Opaque Structure Types</a></li>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]80 <li><a href="#t_vector">Vector Type</a></li>
81 </ol>
82 </li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]83 <li><a href="#t_function">Function Type</a></li>
84 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]85 </ol>
86 </li>
87 </ol>
88 </li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]89 <li><a href="#constants">Constants</a>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]90 <ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]91 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner70882792009-02-28 18:32:25 +0000[diff] [blame]92 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]93 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
94 <li><a href="#undefvalues">Undefined Values</a></li>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]95 <li><a href="#trapvalues">Trap Values</a></li>
Chris Lattnerf9d078e2009-10-27 21:19:13 +0000[diff] [blame]96 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]97 <li><a href="#constantexprs">Constant Expressions</a></li>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]98 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]99 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]100 <li><a href="#othervalues">Other Values</a>
101 <ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]102 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]104 </ol>
105 </li>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]106 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
107 <ol>
108 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]109 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
110 Global Variable</a></li>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]111 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
112 Global Variable</a></li>
113 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
114 Global Variable</a></li>
115 </ol>
116 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]117 <li><a href="#instref">Instruction Reference</a>
118 <ol>
119 <li><a href="#terminators">Terminator Instructions</a>
120 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]121 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
122 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]123 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]124 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]125 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]126 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]127 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]128 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]129 </ol>
130 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]131 <li><a href="#binaryops">Binary Operations</a>
132 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]133 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]134 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]135 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]136 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]137 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]138 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]139 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
140 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
141 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]142 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
143 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
144 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]145 </ol>
146 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]147 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
148 <ol>
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]149 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
150 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
151 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]152 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]153 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]154 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]155 </ol>
156 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]157 <li><a href="#vectorops">Vector Operations</a>
158 <ol>
159 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
160 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
161 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]162 </ol>
163 </li>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]164 <li><a href="#aggregateops">Aggregate Operations</a>
165 <ol>
166 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
167 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
168 </ol>
169 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]170 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]171 <ol>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]172 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
173 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
174 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
175 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
176 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
177 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]178 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]179 </ol>
180 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]181 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]182 <ol>
183 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
184 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
185 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
186 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
187 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]188 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
189 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
190 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
191 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]192 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
193 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]194 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]195 </ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]196 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]197 <li><a href="#otherops">Other Operations</a>
198 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]199 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
200 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]201 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]202 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]203 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]204 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]205 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]206 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]207 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]208 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]209 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]210 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]211 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]212 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
213 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]214 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
215 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
216 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]217 </ol>
218 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]219 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
220 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]221 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
222 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
223 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]224 </ol>
225 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]226 <li><a href="#int_codegen">Code Generator Intrinsics</a>
227 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]228 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
229 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
230 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
231 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
232 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
233 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohman31f1af12010-05-26 21:56:15 +0000[diff] [blame]234 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000[diff] [blame]235 </ol>
236 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]237 <li><a href="#int_libc">Standard C Library Intrinsics</a>
238 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]239 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
240 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
241 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
242 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
243 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]244 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
245 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
246 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohman08b280b2011-05-27 00:36:31 +0000[diff] [blame]247 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
248 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]249 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]250 </ol>
251 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]252 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]253 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]254 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]255 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
256 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
257 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]258 </ol>
259 </li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]260 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
261 <ol>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]262 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
263 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
264 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
265 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
266 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]267 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]268 </ol>
269 </li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]270 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
271 <ol>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]272 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
273 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]274 </ol>
275 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]276 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]277 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]278 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]279 <ol>
280 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]281 </ol>
282 </li>
Bill Wendling3c44f5b2008-11-18 22:10:53 +0000[diff] [blame]283 <li><a href="#int_atomics">Atomic intrinsics</a>
284 <ol>
285 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
286 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
287 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
288 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
289 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
290 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
291 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
292 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
293 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
294 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
295 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
296 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
297 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
298 </ol>
299 </li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]300 <li><a href="#int_memorymarkers">Memory Use Markers</a>
301 <ol>
302 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
303 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
304 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
305 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
306 </ol>
307 </li>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame]308 <li><a href="#int_general">General intrinsics</a>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]309 <ol>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame]310 <li><a href="#int_var_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]311 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]312 <li><a href="#int_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]313 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]314 <li><a href="#int_trap">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]315 '<tt>llvm.trap</tt>' Intrinsic</a></li>
316 <li><a href="#int_stackprotector">
317 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]318 <li><a href="#int_objectsize">
319 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]320 </ol>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]321 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]322 </ol>
323 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]324</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]325
326<div class="doc_author">
327 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
328 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]329</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]330
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]331<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]332<h2><a name="abstract">Abstract</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]333<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]334
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]335<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]336
337<p>This document is a reference manual for the LLVM assembly language. LLVM is
338 a Static Single Assignment (SSA) based representation that provides type
339 safety, low-level operations, flexibility, and the capability of representing
340 'all' high-level languages cleanly. It is the common code representation
341 used throughout all phases of the LLVM compilation strategy.</p>
342
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]343</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]344
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]345<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]346<h2><a name="introduction">Introduction</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]347<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]348
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]349<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]350
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]351<p>The LLVM code representation is designed to be used in three different forms:
352 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
353 for fast loading by a Just-In-Time compiler), and as a human readable
354 assembly language representation. This allows LLVM to provide a powerful
355 intermediate representation for efficient compiler transformations and
356 analysis, while providing a natural means to debug and visualize the
357 transformations. The three different forms of LLVM are all equivalent. This
358 document describes the human readable representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]359
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]360<p>The LLVM representation aims to be light-weight and low-level while being
361 expressive, typed, and extensible at the same time. It aims to be a
362 "universal IR" of sorts, by being at a low enough level that high-level ideas
363 may be cleanly mapped to it (similar to how microprocessors are "universal
364 IR's", allowing many source languages to be mapped to them). By providing
365 type information, LLVM can be used as the target of optimizations: for
366 example, through pointer analysis, it can be proven that a C automatic
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]367 variable is never accessed outside of the current function, allowing it to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]368 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]369
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]370<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]371<h4>
372 <a name="wellformed">Well-Formedness</a>
373</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]374
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]375<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]376
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]377<p>It is important to note that this document describes 'well formed' LLVM
378 assembly language. There is a difference between what the parser accepts and
379 what is considered 'well formed'. For example, the following instruction is
380 syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]381
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]382<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]383%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]384</pre>
385
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]386<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
387 LLVM infrastructure provides a verification pass that may be used to verify
388 that an LLVM module is well formed. This pass is automatically run by the
389 parser after parsing input assembly and by the optimizer before it outputs
390 bitcode. The violations pointed out by the verifier pass indicate bugs in
391 transformation passes or input to the parser.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]392
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]393</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]394
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]395</div>
396
Chris Lattnercc689392007-10-03 17:34:29 +0000[diff] [blame]397<!-- Describe the typesetting conventions here. -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]398
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]399<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]400<h2><a name="identifiers">Identifiers</a></h2>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]401<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]402
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]403<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]404
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]405<p>LLVM identifiers come in two basic types: global and local. Global
406 identifiers (functions, global variables) begin with the <tt>'@'</tt>
407 character. Local identifiers (register names, types) begin with
408 the <tt>'%'</tt> character. Additionally, there are three different formats
409 for identifiers, for different purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]410
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]411<ol>
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]412 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]413 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
414 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
415 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
416 other characters in their names can be surrounded with quotes. Special
417 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
418 ASCII code for the character in hexadecimal. In this way, any character
419 can be used in a name value, even quotes themselves.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]420
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]421 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]422 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]423
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]424 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]425 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]426</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]427
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]428<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]429 don't need to worry about name clashes with reserved words, and the set of
430 reserved words may be expanded in the future without penalty. Additionally,
431 unnamed identifiers allow a compiler to quickly come up with a temporary
432 variable without having to avoid symbol table conflicts.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]433
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]434<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]435 languages. There are keywords for different opcodes
436 ('<tt><a href="#i_add">add</a></tt>',
437 '<tt><a href="#i_bitcast">bitcast</a></tt>',
438 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
439 ('<tt><a href="#t_void">void</a></tt>',
440 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
441 reserved words cannot conflict with variable names, because none of them
442 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]443
444<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]445 '<tt>%X</tt>' by 8:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]446
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]447<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]448
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]449<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]450%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]451</pre>
452
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]453<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]454
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]455<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]456%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]457</pre>
458
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]459<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]460
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]461<pre class="doc_code">
Gabor Greifec58f752009-10-28 13:05:07 +0000[diff] [blame]462%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
463%1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]464%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]465</pre>
466
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]467<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
468 lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]469
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]470<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]471 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]472 line.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]473
474 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]475 assigned to a named value.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]476
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]477 <li>Unnamed temporaries are numbered sequentially</li>
478</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]479
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]480<p>It also shows a convention that we follow in this document. When
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]481 demonstrating instructions, we will follow an instruction with a comment that
482 defines the type and name of value produced. Comments are shown in italic
483 text.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]484
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]485</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]486
487<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]488<h2><a name="highlevel">High Level Structure</a></h2>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]489<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]490<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]491<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]492<h3>
493 <a name="modulestructure">Module Structure</a>
494</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]495
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]496<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]497
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]498<p>LLVM programs are composed of "Module"s, each of which is a translation unit
499 of the input programs. Each module consists of functions, global variables,
500 and symbol table entries. Modules may be combined together with the LLVM
501 linker, which merges function (and global variable) definitions, resolves
502 forward declarations, and merges symbol table entries. Here is an example of
503 the "hello world" module:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]504
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]505<pre class="doc_code">
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]506<i>; Declare the string constant as a global constant.</i>&nbsp;
Nick Lewyckydb9cd762011-01-29 01:09:53 +0000[diff] [blame]507<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a>&nbsp;<a href="#globalvars">constant</a>&nbsp;<a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]508
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]509<i>; External declaration of the puts function</i>&nbsp;
510<a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]511
512<i>; Definition of main function</i>
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]513define i32 @main() { <i>; i32()* </i>&nbsp;
514 <i>; Convert [13 x i8]* to i8 *...</i>&nbsp;
515 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]516
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]517 <i>; Call puts function to write out the string to stdout.</i>&nbsp;
518 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>&nbsp;
519 <a href="#i_ret">ret</a> i32 0&nbsp;
520}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]521
522<i>; Named metadata</i>
523!1 = metadata !{i32 41}
524!foo = !{!1, null}
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]525</pre>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]526
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]527<p>This example is made up of a <a href="#globalvars">global variable</a> named
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]528 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]529 a <a href="#functionstructure">function definition</a> for
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]530 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
531 "<tt>foo"</tt>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]532
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]533<p>In general, a module is made up of a list of global values, where both
534 functions and global variables are global values. Global values are
535 represented by a pointer to a memory location (in this case, a pointer to an
536 array of char, and a pointer to a function), and have one of the
537 following <a href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]538
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]539</div>
540
541<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]542<h3>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]543 <a name="linkage">Linkage Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]544</h3>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]545
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]546<div>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]547
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]548<p>All Global Variables and Functions have one of the following types of
549 linkage:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]550
551<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]552 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]553 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
554 by objects in the current module. In particular, linking code into a
555 module with an private global value may cause the private to be renamed as
556 necessary to avoid collisions. Because the symbol is private to the
557 module, all references can be updated. This doesn't show up in any symbol
558 table in the object file.</dd>
Rafael Espindolabb46f522009-01-15 20:18:42 +0000[diff] [blame]559
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]560 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]561 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
562 assembler and evaluated by the linker. Unlike normal strong symbols, they
563 are removed by the linker from the final linked image (executable or
564 dynamic library).</dd>
565
566 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
567 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
568 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
569 linker. The symbols are removed by the linker from the final linked image
570 (executable or dynamic library).</dd>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000[diff] [blame]571
Bill Wendling55ae5152010-08-20 22:05:50 +0000[diff] [blame]572 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
573 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
574 of the object is not taken. For instance, functions that had an inline
575 definition, but the compiler decided not to inline it. Note,
576 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
577 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
578 visibility. The symbols are removed by the linker from the final linked
579 image (executable or dynamic library).</dd>
580
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]581 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling07d31772010-06-29 22:34:52 +0000[diff] [blame]582 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]583 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
584 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]585
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]586 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000[diff] [blame]587 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]588 into the object file corresponding to the LLVM module. They exist to
589 allow inlining and other optimizations to take place given knowledge of
590 the definition of the global, which is known to be somewhere outside the
591 module. Globals with <tt>available_externally</tt> linkage are allowed to
592 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
593 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000[diff] [blame]594
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]595 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattner4887bd82007-01-14 06:51:48 +0000[diff] [blame]596 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner873187c2010-01-09 19:15:14 +0000[diff] [blame]597 the same name when linkage occurs. This can be used to implement
598 some forms of inline functions, templates, or other code which must be
599 generated in each translation unit that uses it, but where the body may
600 be overridden with a more definitive definition later. Unreferenced
601 <tt>linkonce</tt> globals are allowed to be discarded. Note that
602 <tt>linkonce</tt> linkage does not actually allow the optimizer to
603 inline the body of this function into callers because it doesn't know if
604 this definition of the function is the definitive definition within the
605 program or whether it will be overridden by a stronger definition.
606 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
607 linkage.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]608
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]609 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]610 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
611 <tt>linkonce</tt> linkage, except that unreferenced globals with
612 <tt>weak</tt> linkage may not be discarded. This is used for globals that
613 are declared "weak" in C source code.</dd>
614
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]615 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]616 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
617 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
618 global scope.
619 Symbols with "<tt>common</tt>" linkage are merged in the same way as
620 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000[diff] [blame]621 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]622 must have a zero initializer, and may not be marked '<a
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000[diff] [blame]623 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
624 have common linkage.</dd>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]625
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]626
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]627 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]628 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]629 pointer to array type. When two global variables with appending linkage
630 are linked together, the two global arrays are appended together. This is
631 the LLVM, typesafe, equivalent of having the system linker append together
632 "sections" with identical names when .o files are linked.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]633
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]634 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]635 <dd>The semantics of this linkage follow the ELF object file model: the symbol
636 is weak until linked, if not linked, the symbol becomes null instead of
637 being an undefined reference.</dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]638
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]639 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
640 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]641 <dd>Some languages allow differing globals to be merged, such as two functions
642 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]643 that only equivalent globals are ever merged (the "one definition rule"
644 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]645 and <tt>weak_odr</tt> linkage types to indicate that the global will only
646 be merged with equivalent globals. These linkage types are otherwise the
647 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]648
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]649 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]650 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]651 visible, meaning that it participates in linkage and can be used to
652 resolve external symbol references.</dd>
Reid Spencerc8910842007-04-11 23:49:50 +0000[diff] [blame]653</dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]654
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]655<p>The next two types of linkage are targeted for Microsoft Windows platform
656 only. They are designed to support importing (exporting) symbols from (to)
657 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]658
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]659<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]660 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]661 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]662 or variable via a global pointer to a pointer that is set up by the DLL
663 exporting the symbol. On Microsoft Windows targets, the pointer name is
664 formed by combining <code>__imp_</code> and the function or variable
665 name.</dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]666
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]667 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]668 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]669 pointer to a pointer in a DLL, so that it can be referenced with the
670 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
671 name is formed by combining <code>__imp_</code> and the function or
672 variable name.</dd>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]673</dl>
674
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]675<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
676 another module defined a "<tt>.LC0</tt>" variable and was linked with this
677 one, one of the two would be renamed, preventing a collision. Since
678 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
679 declarations), they are accessible outside of the current module.</p>
680
681<p>It is illegal for a function <i>declaration</i> to have any linkage type
682 other than "externally visible", <tt>dllimport</tt>
683 or <tt>extern_weak</tt>.</p>
684
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]685<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]686 or <tt>weak_odr</tt> linkages.</p>
687
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]688</div>
689
690<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]691<h3>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]692 <a name="callingconv">Calling Conventions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]693</h3>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]694
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]695<div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]696
697<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]698 and <a href="#i_invoke">invokes</a> can all have an optional calling
699 convention specified for the call. The calling convention of any pair of
700 dynamic caller/callee must match, or the behavior of the program is
701 undefined. The following calling conventions are supported by LLVM, and more
702 may be added in the future:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]703
704<dl>
705 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]706 <dd>This calling convention (the default if no other calling convention is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]707 specified) matches the target C calling conventions. This calling
708 convention supports varargs function calls and tolerates some mismatch in
709 the declared prototype and implemented declaration of the function (as
710 does normal C).</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]711
712 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]713 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]714 (e.g. by passing things in registers). This calling convention allows the
715 target to use whatever tricks it wants to produce fast code for the
716 target, without having to conform to an externally specified ABI
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]717 (Application Binary Interface).
718 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattner29689432010-03-11 00:22:57 +0000[diff] [blame]719 when this or the GHC convention is used.</a> This calling convention
720 does not support varargs and requires the prototype of all callees to
721 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]722
723 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]724 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]725 as possible under the assumption that the call is not commonly executed.
726 As such, these calls often preserve all registers so that the call does
727 not break any live ranges in the caller side. This calling convention
728 does not support varargs and requires the prototype of all callees to
729 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]730
Chris Lattner29689432010-03-11 00:22:57 +0000[diff] [blame]731 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
732 <dd>This calling convention has been implemented specifically for use by the
733 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
734 It passes everything in registers, going to extremes to achieve this by
735 disabling callee save registers. This calling convention should not be
736 used lightly but only for specific situations such as an alternative to
737 the <em>register pinning</em> performance technique often used when
738 implementing functional programming languages.At the moment only X86
739 supports this convention and it has the following limitations:
740 <ul>
741 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
742 floating point types are supported.</li>
743 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
744 6 floating point parameters.</li>
745 </ul>
746 This calling convention supports
747 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
748 requires both the caller and callee are using it.
749 </dd>
750
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]751 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]752 <dd>Any calling convention may be specified by number, allowing
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]753 target-specific calling conventions to be used. Target specific calling
754 conventions start at 64.</dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]755</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]756
757<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]758 support Pascal conventions or any other well-known target-independent
759 convention.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]760
761</div>
762
763<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]764<h3>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]765 <a name="visibility">Visibility Styles</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]766</h3>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]767
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]768<div>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]769
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]770<p>All Global Variables and Functions have one of the following visibility
771 styles:</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]772
773<dl>
774 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]775 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]776 that the declaration is visible to other modules and, in shared libraries,
777 means that the declared entity may be overridden. On Darwin, default
778 visibility means that the declaration is visible to other modules. Default
779 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]780
781 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]782 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]783 object if they are in the same shared object. Usually, hidden visibility
784 indicates that the symbol will not be placed into the dynamic symbol
785 table, so no other module (executable or shared library) can reference it
786 directly.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]787
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]788 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]789 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]790 the dynamic symbol table, but that references within the defining module
791 will bind to the local symbol. That is, the symbol cannot be overridden by
792 another module.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]793</dl>
794
795</div>
796
797<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]798<h3>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]799 <a name="namedtypes">Named Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]800</h3>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]801
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]802<div>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]803
804<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]805 it easier to read the IR and make the IR more condensed (particularly when
806 recursive types are involved). An example of a name specification is:</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]807
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]808<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]809%mytype = type { %mytype*, i32 }
810</pre>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]811
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]812<p>You may give a name to any <a href="#typesystem">type</a> except
Chris Lattnerdc65f222010-08-17 23:26:04 +0000[diff] [blame]813 "<a href="#t_void">void</a>". Type name aliases may be used anywhere a type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]814 is expected with the syntax "%mytype".</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]815
816<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]817 and that you can therefore specify multiple names for the same type. This
818 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
819 uses structural typing, the name is not part of the type. When printing out
820 LLVM IR, the printer will pick <em>one name</em> to render all types of a
821 particular shape. This means that if you have code where two different
822 source types end up having the same LLVM type, that the dumper will sometimes
823 print the "wrong" or unexpected type. This is an important design point and
824 isn't going to change.</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]825
826</div>
827
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]828<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]829<h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]830 <a name="globalvars">Global Variables</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]831</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]832
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]833<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]834
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]835<p>Global variables define regions of memory allocated at compilation time
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]836 instead of run-time. Global variables may optionally be initialized, may
837 have an explicit section to be placed in, and may have an optional explicit
838 alignment specified. A variable may be defined as "thread_local", which
839 means that it will not be shared by threads (each thread will have a
840 separated copy of the variable). A variable may be defined as a global
841 "constant," which indicates that the contents of the variable
842 will <b>never</b> be modified (enabling better optimization, allowing the
843 global data to be placed in the read-only section of an executable, etc).
844 Note that variables that need runtime initialization cannot be marked
845 "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]846
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]847<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
848 constant, even if the final definition of the global is not. This capability
849 can be used to enable slightly better optimization of the program, but
850 requires the language definition to guarantee that optimizations based on the
851 'constantness' are valid for the translation units that do not include the
852 definition.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]853
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]854<p>As SSA values, global variables define pointer values that are in scope
855 (i.e. they dominate) all basic blocks in the program. Global variables
856 always define a pointer to their "content" type because they describe a
857 region of memory, and all memory objects in LLVM are accessed through
858 pointers.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]859
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]860<p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
861 that the address is not significant, only the content. Constants marked
Rafael Espindolaa5eaa862011-01-15 08:20:57 +0000[diff] [blame]862 like this can be merged with other constants if they have the same
863 initializer. Note that a constant with significant address <em>can</em>
864 be merged with a <tt>unnamed_addr</tt> constant, the result being a
865 constant whose address is significant.</p>
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]866
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]867<p>A global variable may be declared to reside in a target-specific numbered
868 address space. For targets that support them, address spaces may affect how
869 optimizations are performed and/or what target instructions are used to
870 access the variable. The default address space is zero. The address space
871 qualifier must precede any other attributes.</p>
Christopher Lamb284d9922007-12-11 09:31:00 +0000[diff] [blame]872
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]873<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]874 supports it, it will emit globals to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]875
Chris Lattnerce99fa92010-04-28 00:13:42 +0000[diff] [blame]876<p>An explicit alignment may be specified for a global, which must be a power
877 of 2. If not present, or if the alignment is set to zero, the alignment of
878 the global is set by the target to whatever it feels convenient. If an
879 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner2d4b8ee2010-04-28 00:31:12 +0000[diff] [blame]880 alignment. Targets and optimizers are not allowed to over-align the global
881 if the global has an assigned section. In this case, the extra alignment
882 could be observable: for example, code could assume that the globals are
883 densely packed in their section and try to iterate over them as an array,
884 alignment padding would break this iteration.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]885
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]886<p>For example, the following defines a global in a numbered address space with
887 an initializer, section, and alignment:</p>
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]888
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]889<pre class="doc_code">
Dan Gohman398873c2009-01-11 00:40:00 +0000[diff] [blame]890@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]891</pre>
892
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]893</div>
894
895
896<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]897<h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]898 <a name="functionstructure">Functions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]899</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]900
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]901<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]902
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]903<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]904 optional <a href="#linkage">linkage type</a>, an optional
905 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]906 <a href="#callingconv">calling convention</a>,
907 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]908 <a href="#paramattrs">parameter attribute</a> for the return type, a function
909 name, a (possibly empty) argument list (each with optional
910 <a href="#paramattrs">parameter attributes</a>), optional
911 <a href="#fnattrs">function attributes</a>, an optional section, an optional
912 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
913 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]914
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]915<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
916 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]917 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]918 <a href="#callingconv">calling convention</a>,
919 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]920 <a href="#paramattrs">parameter attribute</a> for the return type, a function
921 name, a possibly empty list of arguments, an optional alignment, and an
922 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]923
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]924<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]925 (Control Flow Graph) for the function. Each basic block may optionally start
926 with a label (giving the basic block a symbol table entry), contains a list
927 of instructions, and ends with a <a href="#terminators">terminator</a>
928 instruction (such as a branch or function return).</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]929
Chris Lattner4a3c9012007-06-08 16:52:14 +0000[diff] [blame]930<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]931 executed on entrance to the function, and it is not allowed to have
932 predecessor basic blocks (i.e. there can not be any branches to the entry
933 block of a function). Because the block can have no predecessors, it also
934 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]935
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]936<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]937 supports it, it will emit functions to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]938
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]939<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]940 the alignment is set to zero, the alignment of the function is set by the
941 target to whatever it feels convenient. If an explicit alignment is
942 specified, the function is forced to have at least that much alignment. All
943 alignments must be a power of 2.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]944
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]945<p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
946 be significant and two identical functions can be merged</p>.
947
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]948<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]949<pre class="doc_code">
Chris Lattner50ad45c2008-10-13 16:55:18 +0000[diff] [blame]950define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]951 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
952 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
953 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
954 [<a href="#gc">gc</a>] { ... }
955</pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]956
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]957</div>
958
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]959<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]960<h3>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]961 <a name="aliasstructure">Aliases</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]962</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]963
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]964<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]965
966<p>Aliases act as "second name" for the aliasee value (which can be either
967 function, global variable, another alias or bitcast of global value). Aliases
968 may have an optional <a href="#linkage">linkage type</a>, and an
969 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]970
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]971<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]972<pre class="doc_code">
Duncan Sands0b23ac12008-09-12 20:48:21 +0000[diff] [blame]973@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]974</pre>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]975
976</div>
977
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]978<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]979<h3>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]980 <a name="namedmetadatastructure">Named Metadata</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]981</h3>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]982
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]983<div>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]984
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]985<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]986 nodes</a> (but not metadata strings) are the only valid operands for
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]987 a named metadata.</p>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]988
989<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]990<pre class="doc_code">
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]991; Some unnamed metadata nodes, which are referenced by the named metadata.
992!0 = metadata !{metadata !"zero"}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]993!1 = metadata !{metadata !"one"}
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]994!2 = metadata !{metadata !"two"}
Dan Gohman1005bc52010-07-13 19:48:13 +0000[diff] [blame]995; A named metadata.
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]996!name = !{!0, !1, !2}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]997</pre>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]998
999</div>
1000
1001<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1002<h3>
1003 <a name="paramattrs">Parameter Attributes</a>
1004</h3>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1005
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1006<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1007
1008<p>The return type and each parameter of a function type may have a set of
1009 <i>parameter attributes</i> associated with them. Parameter attributes are
1010 used to communicate additional information about the result or parameters of
1011 a function. Parameter attributes are considered to be part of the function,
1012 not of the function type, so functions with different parameter attributes
1013 can have the same function type.</p>
1014
1015<p>Parameter attributes are simple keywords that follow the type specified. If
1016 multiple parameter attributes are needed, they are space separated. For
1017 example:</p>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1018
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1019<pre class="doc_code">
Nick Lewyckyb6a7d252009-02-15 23:06:14 +0000[diff] [blame]1020declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattner66d922c2008-10-04 18:33:34 +0000[diff] [blame]1021declare i32 @atoi(i8 zeroext)
1022declare signext i8 @returns_signed_char()
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1023</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1024
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1025<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1026 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1027
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1028<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1029
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1030<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1031 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1032 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarichebe81732011-03-16 22:20:18 +0000[diff] [blame]1033 should be zero-extended to the extent required by the target's ABI (which
1034 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1035 parameter) or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1036
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1037 <dt><tt><b>signext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1038 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarich9e69ff92011-03-17 14:21:58 +0000[diff] [blame]1039 should be sign-extended to the extent required by the target's ABI (which
1040 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1041 return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1042
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1043 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1044 <dd>This indicates that this parameter or return value should be treated in a
1045 special target-dependent fashion during while emitting code for a function
1046 call or return (usually, by putting it in a register as opposed to memory,
1047 though some targets use it to distinguish between two different kinds of
1048 registers). Use of this attribute is target-specific.</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1049
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1050 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Chris Lattnera6fd81d2010-11-20 23:49:06 +0000[diff] [blame]1051 <dd><p>This indicates that the pointer parameter should really be passed by
1052 value to the function. The attribute implies that a hidden copy of the
1053 pointee
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1054 is made between the caller and the callee, so the callee is unable to
1055 modify the value in the callee. This attribute is only valid on LLVM
1056 pointer arguments. It is generally used to pass structs and arrays by
1057 value, but is also valid on pointers to scalars. The copy is considered
1058 to belong to the caller not the callee (for example,
1059 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1060 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnera6fd81d2010-11-20 23:49:06 +0000[diff] [blame]1061 values.</p>
1062
1063 <p>The byval attribute also supports specifying an alignment with
1064 the align attribute. It indicates the alignment of the stack slot to
1065 form and the known alignment of the pointer specified to the call site. If
1066 the alignment is not specified, then the code generator makes a
1067 target-specific assumption.</p></dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1068
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1069 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1070 <dd>This indicates that the pointer parameter specifies the address of a
1071 structure that is the return value of the function in the source program.
1072 This pointer must be guaranteed by the caller to be valid: loads and
1073 stores to the structure may be assumed by the callee to not to trap. This
1074 may only be applied to the first parameter. This is not a valid attribute
1075 for return values. </dd>
1076
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1077 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1078 <dd>This indicates that pointer values
1079 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmanefca7f92010-07-02 23:46:54 +0000[diff] [blame]1080 value do not alias pointer values which are not <i>based</i> on it,
1081 ignoring certain "irrelevant" dependencies.
1082 For a call to the parent function, dependencies between memory
1083 references from before or after the call and from those during the call
1084 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1085 return value used in that call.
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1086 The caller shares the responsibility with the callee for ensuring that
1087 these requirements are met.
1088 For further details, please see the discussion of the NoAlias response in
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]1089 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1090<br>
John McCall191d4ee2010-07-06 21:07:14 +0000[diff] [blame]1091 Note that this definition of <tt>noalias</tt> is intentionally
1092 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner211244a2010-07-06 20:51:35 +0000[diff] [blame]1093 arguments, though it is slightly weaker.
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]1094<br>
1095 For function return values, C99's <tt>restrict</tt> is not meaningful,
1096 while LLVM's <tt>noalias</tt> is.
1097 </dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1098
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1099 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1100 <dd>This indicates that the callee does not make any copies of the pointer
1101 that outlive the callee itself. This is not a valid attribute for return
1102 values.</dd>
1103
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1104 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1105 <dd>This indicates that the pointer parameter can be excised using the
1106 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1107 attribute for return values.</dd>
1108</dl>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1109
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1110</div>
1111
1112<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1113<h3>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1114 <a name="gc">Garbage Collector Names</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1115</h3>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1116
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1117<div>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1118
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1119<p>Each function may specify a garbage collector name, which is simply a
1120 string:</p>
1121
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1122<pre class="doc_code">
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1123define void @f() gc "name" { ... }
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1124</pre>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1125
1126<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1127 collector which will cause the compiler to alter its output in order to
1128 support the named garbage collection algorithm.</p>
1129
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1130</div>
1131
1132<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1133<h3>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1134 <a name="fnattrs">Function Attributes</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1135</h3>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1136
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1137<div>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1138
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1139<p>Function attributes are set to communicate additional information about a
1140 function. Function attributes are considered to be part of the function, not
1141 of the function type, so functions with different parameter attributes can
1142 have the same function type.</p>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1143
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1144<p>Function attributes are simple keywords that follow the type specified. If
1145 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1146
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1147<pre class="doc_code">
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1148define void @f() noinline { ... }
1149define void @f() alwaysinline { ... }
1150define void @f() alwaysinline optsize { ... }
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1151define void @f() optsize { ... }
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1152</pre>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1153
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1154<dl>
Charles Davis1e063d12010-02-12 00:31:15 +0000[diff] [blame]1155 <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 Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1160 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1161 <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 Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1164
Charles Davis970bfcc2010-10-25 15:37:09 +0000[diff] [blame]1165 <dt><tt><b>hotpatch</b></tt></dt>
Charles Davis6f12e292010-10-25 16:29:03 +0000[diff] [blame]1166 <dd>This attribute indicates that the function should be 'hotpatchable',
Charles Davis0076d202010-10-25 19:07:39 +0000[diff] [blame]1167 meaning the function can be patched and/or hooked even while it is
1168 loaded into memory. On x86, the function prologue will be preceded
1169 by six bytes of padding and will begin with a two-byte instruction.
1170 Most of the functions in the Windows system DLLs in Windows XP SP2 or
1171 higher were compiled in this fashion.</dd>
Charles Davis970bfcc2010-10-25 15:37:09 +0000[diff] [blame]1172
Dan Gohman129bd562011-06-16 16:03:13 +0000[diff] [blame]1173 <dt><tt><b>nonlazybind</b></tt></dt>
1174 <dd>This attribute suppresses lazy symbol binding for the function. This
1175 may make calls to the function faster, at the cost of extra program
1176 startup time if the function is not called during program startup.</dd>
1177
Jakob Stoklund Olesen570a4a52010-02-06 01:16:28 +0000[diff] [blame]1178 <dt><tt><b>inlinehint</b></tt></dt>
1179 <dd>This attribute indicates that the source code contained a hint that inlining
1180 this function is desirable (such as the "inline" keyword in C/C++). It
1181 is just a hint; it imposes no requirements on the inliner.</dd>
1182
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1183 <dt><tt><b>naked</b></tt></dt>
1184 <dd>This attribute disables prologue / epilogue emission for the function.
1185 This can have very system-specific consequences.</dd>
1186
1187 <dt><tt><b>noimplicitfloat</b></tt></dt>
1188 <dd>This attributes disables implicit floating point instructions.</dd>
1189
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1190 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1191 <dd>This attribute indicates that the inliner should never inline this
1192 function in any situation. This attribute may not be used together with
1193 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1194
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1195 <dt><tt><b>noredzone</b></tt></dt>
1196 <dd>This attribute indicates that the code generator should not use a red
1197 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1198
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1199 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1200 <dd>This function attribute indicates that the function never returns
1201 normally. This produces undefined behavior at runtime if the function
1202 ever does dynamically return.</dd>
Bill Wendling31359ba2008-11-13 01:02:51 +0000[diff] [blame]1203
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1204 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1205 <dd>This function attribute indicates that the function never returns with an
1206 unwind or exceptional control flow. If the function does unwind, its
1207 runtime behavior is undefined.</dd>
Bill Wendlingfbaa7ed2008-11-26 19:07:40 +0000[diff] [blame]1208
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1209 <dt><tt><b>optsize</b></tt></dt>
1210 <dd>This attribute suggests that optimization passes and code generator passes
1211 make choices that keep the code size of this function low, and otherwise
1212 do optimizations specifically to reduce code size.</dd>
1213
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1214 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1215 <dd>This attribute indicates that the function computes its result (or decides
1216 to unwind an exception) based strictly on its arguments, without
1217 dereferencing any pointer arguments or otherwise accessing any mutable
1218 state (e.g. memory, control registers, etc) visible to caller functions.
1219 It does not write through any pointer arguments
1220 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1221 changes any state visible to callers. This means that it cannot unwind
1222 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1223 could use the <tt>unwind</tt> instruction.</dd>
Devang Patel5d96fda2009-06-12 19:45:19 +0000[diff] [blame]1224
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1225 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1226 <dd>This attribute indicates that the function does not write through any
1227 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1228 arguments) or otherwise modify any state (e.g. memory, control registers,
1229 etc) visible to caller functions. It may dereference pointer arguments
1230 and read state that may be set in the caller. A readonly function always
1231 returns the same value (or unwinds an exception identically) when called
1232 with the same set of arguments and global state. It cannot unwind an
1233 exception by calling the <tt>C++</tt> exception throwing methods, but may
1234 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovc5ec8a72009-07-17 18:07:26 +0000[diff] [blame]1235
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1236 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1237 <dd>This attribute indicates that the function should emit a stack smashing
1238 protector. It is in the form of a "canary"&mdash;a random value placed on
1239 the stack before the local variables that's checked upon return from the
1240 function to see if it has been overwritten. A heuristic is used to
1241 determine if a function needs stack protectors or not.<br>
1242<br>
1243 If a function that has an <tt>ssp</tt> attribute is inlined into a
1244 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1245 function will have an <tt>ssp</tt> attribute.</dd>
1246
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1247 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1248 <dd>This attribute indicates that the function should <em>always</em> emit a
1249 stack smashing protector. This overrides
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1250 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1251<br>
1252 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1253 function that doesn't have an <tt>sspreq</tt> attribute or which has
1254 an <tt>ssp</tt> attribute, then the resulting function will have
1255 an <tt>sspreq</tt> attribute.</dd>
Rafael Espindolafbff0ec2011-07-25 15:27:59 +0000[diff] [blame]1256
1257 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1258 <dd>This attribute indicates that the ABI being targeted requires that
1259 an unwind table entry be produce for this function even if we can
1260 show that no exceptions passes by it. This is normally the case for
1261 the ELF x86-64 abi, but it can be disabled for some compilation
1262 units.</dd>
1263
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1264</dl>
1265
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1266</div>
1267
1268<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1269<h3>
Chris Lattner1eeeb0c2006-04-08 04:40:53 +0000[diff] [blame]1270 <a name="moduleasm">Module-Level Inline Assembly</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1271</h3>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1272
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1273<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1274
1275<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1276 the GCC "file scope inline asm" blocks. These blocks are internally
1277 concatenated by LLVM and treated as a single unit, but may be separated in
1278 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1279
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1280<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1281module asm "inline asm code goes here"
1282module asm "more can go here"
1283</pre>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1284
1285<p>The strings can contain any character by escaping non-printable characters.
1286 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1287 for the number.</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1288
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1289<p>The inline asm code is simply printed to the machine code .s file when
1290 assembly code is generated.</p>
1291
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1292</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1293
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1294<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1295<h3>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1296 <a name="datalayout">Data Layout</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1297</h3>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1298
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1299<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1300
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1301<p>A module may specify a target specific data layout string that specifies how
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1302 data is to be laid out in memory. The syntax for the data layout is
1303 simply:</p>
1304
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1305<pre class="doc_code">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1306target datalayout = "<i>layout specification</i>"
1307</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1308
1309<p>The <i>layout specification</i> consists of a list of specifications
1310 separated by the minus sign character ('-'). Each specification starts with
1311 a letter and may include other information after the letter to define some
1312 aspect of the data layout. The specifications accepted are as follows:</p>
1313
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1314<dl>
1315 <dt><tt>E</tt></dt>
1316 <dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1317 bits with the most significance have the lowest address location.</dd>
1318
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1319 <dt><tt>e</tt></dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]1320 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1321 the bits with the least significance have the lowest address
1322 location.</dd>
1323
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1324 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1325 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1326 <i>preferred</i> alignments. All sizes are in bits. Specifying
1327 the <i>pref</i> alignment is optional. If omitted, the
1328 preceding <tt>:</tt> should be omitted too.</dd>
1329
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1330 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1331 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1332 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1333
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1334 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1335 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1336 <i>size</i>.</dd>
1337
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1338 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1339 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesen9d8d2212010-05-28 18:54:47 +0000[diff] [blame]1340 <i>size</i>. Only values of <i>size</i> that are supported by the target
1341 will work. 32 (float) and 64 (double) are supported on all targets;
1342 80 or 128 (different flavors of long double) are also supported on some
1343 targets.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1344
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1345 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1346 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1347 <i>size</i>.</dd>
1348
Daniel Dunbar87bde0b2009-06-08 22:17:53 +0000[diff] [blame]1349 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1350 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1351 <i>size</i>.</dd>
Chris Lattnere82bdc42009-11-07 09:35:34 +0000[diff] [blame]1352
1353 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1354 <dd>This specifies a set of native integer widths for the target CPU
1355 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1356 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1357 this set are considered to support most general arithmetic
Chris Lattnere82bdc42009-11-07 09:35:34 +0000[diff] [blame]1358 operations efficiently.</dd>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1359</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1360
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1361<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman1c70c002010-04-28 00:36:01 +0000[diff] [blame]1362 default set of specifications which are then (possibly) overridden by the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1363 specifications in the <tt>datalayout</tt> keyword. The default specifications
1364 are given in this list:</p>
1365
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1366<ul>
1367 <li><tt>E</tt> - big endian</li>
Dan Gohmanfdf2e8c2010-02-23 02:44:03 +0000[diff] [blame]1368 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1369 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1370 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1371 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1372 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]1373 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1374 alignment of 64-bits</li>
1375 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1376 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1377 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1378 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1379 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar87bde0b2009-06-08 22:17:53 +0000[diff] [blame]1380 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1381</ul>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1382
1383<p>When LLVM is determining the alignment for a given type, it uses the
1384 following rules:</p>
1385
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1386<ol>
1387 <li>If the type sought is an exact match for one of the specifications, that
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1388 specification is used.</li>
1389
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1390 <li>If no match is found, and the type sought is an integer type, then the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1391 smallest integer type that is larger than the bitwidth of the sought type
1392 is used. If none of the specifications are larger than the bitwidth then
1393 the the largest integer type is used. For example, given the default
1394 specifications above, the i7 type will use the alignment of i8 (next
1395 largest) while both i65 and i256 will use the alignment of i64 (largest
1396 specified).</li>
1397
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1398 <li>If no match is found, and the type sought is a vector type, then the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1399 largest vector type that is smaller than the sought vector type will be
1400 used as a fall back. This happens because &lt;128 x double&gt; can be
1401 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1402</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1403
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1404</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1405
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1406<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1407<h3>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1408 <a name="pointeraliasing">Pointer Aliasing Rules</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1409</h3>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1410
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1411<div>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1412
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1413<p>Any memory access must be done through a pointer value associated
Andreas Bolka99a82052009-07-27 20:37:10 +0000[diff] [blame]1414with an address range of the memory access, otherwise the behavior
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1415is undefined. Pointer values are associated with address ranges
1416according to the following rules:</p>
1417
1418<ul>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1419 <li>A pointer value is associated with the addresses associated with
1420 any value it is <i>based</i> on.
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1421 <li>An address of a global variable is associated with the address
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1422 range of the variable's storage.</li>
1423 <li>The result value of an allocation instruction is associated with
1424 the address range of the allocated storage.</li>
1425 <li>A null pointer in the default address-space is associated with
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1426 no address.</li>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1427 <li>An integer constant other than zero or a pointer value returned
1428 from a function not defined within LLVM may be associated with address
1429 ranges allocated through mechanisms other than those provided by
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1430 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1431 allocated by mechanisms provided by LLVM.</li>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1432</ul>
1433
1434<p>A pointer value is <i>based</i> on another pointer value according
1435 to the following rules:</p>
1436
1437<ul>
1438 <li>A pointer value formed from a
1439 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1440 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1441 <li>The result value of a
1442 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1443 of the <tt>bitcast</tt>.</li>
1444 <li>A pointer value formed by an
1445 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1446 pointer values that contribute (directly or indirectly) to the
1447 computation of the pointer's value.</li>
1448 <li>The "<i>based</i> on" relationship is transitive.</li>
1449</ul>
1450
1451<p>Note that this definition of <i>"based"</i> is intentionally
1452 similar to the definition of <i>"based"</i> in C99, though it is
1453 slightly weaker.</p>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1454
1455<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1456<tt><a href="#i_load">load</a></tt> merely indicates the size and
1457alignment of the memory from which to load, as well as the
Dan Gohmanc22c0f32010-06-17 19:23:50 +0000[diff] [blame]1458interpretation of the value. The first operand type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1459<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1460and alignment of the store.</p>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1461
1462<p>Consequently, type-based alias analysis, aka TBAA, aka
1463<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1464LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1465additional information which specialized optimization passes may use
1466to implement type-based alias analysis.</p>
1467
1468</div>
1469
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1470<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1471<h3>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1472 <a name="volatile">Volatile Memory Accesses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1473</h3>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1474
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1475<div>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1476
1477<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1478href="#i_store"><tt>store</tt></a>s, and <a
1479href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1480The optimizers must not change the number of volatile operations or change their
1481order of execution relative to other volatile operations. The optimizers
1482<i>may</i> change the order of volatile operations relative to non-volatile
1483operations. This is not Java's "volatile" and has no cross-thread
1484synchronization behavior.</p>
1485
1486</div>
1487
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1488<!-- ======================================================================= -->
1489<h3>
1490 <a name="memmodel">Memory Model for Concurrent Operations</a>
1491</h3>
1492
1493<div>
1494
1495<p>The LLVM IR does not define any way to start parallel threads of execution
1496or to register signal handlers. Nonetheless, there are platform-specific
1497ways to create them, and we define LLVM IR's behavior in their presence. This
1498model is inspired by the C++0x memory model.</p>
1499
1500<p>We define a <i>happens-before</i> partial order as the least partial order
1501that</p>
1502<ul>
1503 <li>Is a superset of single-thread program order, and</li>
1504 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1505 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1506 by platform-specific techniques, like pthread locks, thread
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1507 creation, thread joining, etc., and by atomic instructions.
1508 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1509 </li>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1510</ul>
1511
1512<p>Note that program order does not introduce <i>happens-before</i> edges
1513between a thread and signals executing inside that thread.</p>
1514
1515<p>Every (defined) read operation (load instructions, memcpy, atomic
1516loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1517(defined) write operations (store instructions, atomic
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1518stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1519initialized globals are considered to have a write of the initializer which is
1520atomic and happens before any other read or write of the memory in question.
1521For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1522any write to the same byte, except:</p>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1523
1524<ul>
1525 <li>If <var>write<sub>1</sub></var> happens before
1526 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1527 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1528 does not see <var>write<sub>1</sub></var>.
Bill Wendling0246bb72011-07-31 06:45:03 +0000[diff] [blame]1529 <li>If <var>R<sub>byte</sub></var> happens before
1530 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1531 see <var>write<sub>3</sub></var>.
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1532</ul>
1533
1534<p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1535<ul>
1536 <li>If there is no write to the same byte that happens before
1537 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1538 <tt>undef</tt> for that byte.
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1539 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1540 <var>R<sub>byte</sub></var> returns the value written by that
1541 write.</li>
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1542 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1543 <var>R<sub>byte</sub></var> may see are atomic, it chooses one of the
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1544 values written. See the <a href="#ordering">Atomic Memory Ordering
1545 Constraints</a> section for additional constraints on how the choice
1546 is made.
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1547 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1548</ul>
1549
1550<p><var>R</var> returns the value composed of the series of bytes it read.
1551This implies that some bytes within the value may be <tt>undef</tt>
1552<b>without</b> the entire value being <tt>undef</tt>. Note that this only
1553defines the semantics of the operation; it doesn't mean that targets will
1554emit more than one instruction to read the series of bytes.</p>
1555
1556<p>Note that in cases where none of the atomic intrinsics are used, this model
1557places only one restriction on IR transformations on top of what is required
1558for single-threaded execution: introducing a store to a byte which might not
Eli Friedman101c81d2011-08-02 01:15:34 +0000[diff] [blame]1559otherwise be stored is not allowed in general. (Specifically, in the case
1560where another thread might write to and read from an address, introducing a
1561store can change a load that may see exactly one write into a load that may
1562see multiple writes.)</p>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1563
1564<!-- FIXME: This model assumes all targets where concurrency is relevant have
1565a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1566none of the backends currently in the tree fall into this category; however,
1567there might be targets which care. If there are, we want a paragraph
1568like the following:
1569
1570Targets may specify that stores narrower than a certain width are not
1571available; on such a target, for the purposes of this model, treat any
1572non-atomic write with an alignment or width less than the minimum width
1573as if it writes to the relevant surrounding bytes.
1574-->
1575
1576</div>
1577
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1578<!-- ======================================================================= -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]1579<h3>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1580 <a name="ordering">Atomic Memory Ordering Constraints</a>
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]1581</h3>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1582
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]1583<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1584
1585<p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]1586<a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1587<a href="#i_fence"><code>fence</code></a>,
1588<a href="#i_load"><code>atomic load</code></a>, and
Eli Friedman8fa281a2011-08-09 23:26:12 +0000[diff] [blame]1589<a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1590that determines which other atomic instructions on the same address they
1591<i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1592but are somewhat more colloquial. If these descriptions aren't precise enough,
1593check those specs. <a href="#i_fence"><code>fence</code></a> instructions
1594treat these orderings somewhat differently since they don't take an address.
1595See that instruction's documentation for details.</p>
1596
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1597<dl>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1598<dt><code>unordered</code></dt>
1599<dd>The set of values that can be read is governed by the happens-before
1600partial order. A value cannot be read unless some operation wrote it.
1601This is intended to provide a guarantee strong enough to model Java's
1602non-volatile shared variables. This ordering cannot be specified for
1603read-modify-write operations; it is not strong enough to make them atomic
1604in any interesting way.</dd>
1605<dt><code>monotonic</code></dt>
1606<dd>In addition to the guarantees of <code>unordered</code>, there is a single
1607total order for modifications by <code>monotonic</code> operations on each
1608address. All modification orders must be compatible with the happens-before
1609order. There is no guarantee that the modification orders can be combined to
1610a global total order for the whole program (and this often will not be
1611possible). The read in an atomic read-modify-write operation
1612(<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1613<a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1614reads the value in the modification order immediately before the value it
1615writes. If one atomic read happens before another atomic read of the same
1616address, the later read must see the same value or a later value in the
1617address's modification order. This disallows reordering of
1618<code>monotonic</code> (or stronger) operations on the same address. If an
1619address is written <code>monotonic</code>ally by one thread, and other threads
1620<code>monotonic</code>ally read that address repeatedly, the other threads must
1621eventually see the write. This is intended to model C++'s relaxed atomic
1622variables.</dd>
1623<dt><code>acquire</code></dt>
1624<dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1625reads a value written by a <code>release</code> atomic operation, it
1626<i>synchronizes-with</i> that operation.</dd>
1627<dt><code>release</code></dt>
1628<dd>In addition to the guarantees of <code>monotonic</code>,
1629a <i>synchronizes-with</i> edge may be formed by an <code>acquire</code>
1630operation.</dd>
1631<dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
1632<code>acquire</code> and <code>release</code> operation on its address.</dd>
1633<dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1634<dd>In addition to the guarantees of <code>acq_rel</code>
1635(<code>acquire</code> for an operation which only reads, <code>release</code>
1636for an operation which only writes), there is a global total order on all
1637sequentially-consistent operations on all addresses, which is consistent with
1638the <i>happens-before</i> partial order and with the modification orders of
1639all the affected addresses. Each sequentially-consistent read sees the last
1640preceding write to the same address in this global order. This is intended
1641to model C++'s sequentially-consistent atomic variables and Java's volatile
1642shared variables.</dd>
1643</dl>
1644
1645<p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1646it only <i>synchronizes with</i> or participates in modification and seq_cst
1647total orderings with other operations running in the same thread (for example,
1648in signal handlers).</p>
1649
1650</div>
1651
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1652</div>
1653
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1654<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1655<h2><a name="typesystem">Type System</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1656<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1657
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1658<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1659
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1660<p>The LLVM type system is one of the most important features of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1661 intermediate representation. Being typed enables a number of optimizations
1662 to be performed on the intermediate representation directly, without having
1663 to do extra analyses on the side before the transformation. A strong type
1664 system makes it easier to read the generated code and enables novel analyses
1665 and transformations that are not feasible to perform on normal three address
1666 code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1667
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1668<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1669<h3>
1670 <a name="t_classifications">Type Classifications</a>
1671</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1672
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1673<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1674
1675<p>The types fall into a few useful classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1676
1677<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1678 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1679 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1680 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1681 <td><a href="#t_integer">integer</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1682 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1683 </tr>
1684 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1685 <td><a href="#t_floating">floating point</a></td>
1686 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1687 </tr>
1688 <tr>
1689 <td><a name="t_firstclass">first class</a></td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1690 <td><a href="#t_integer">integer</a>,
1691 <a href="#t_floating">floating point</a>,
1692 <a href="#t_pointer">pointer</a>,
Dan Gohman0066db62008-06-18 18:42:13 +0000[diff] [blame]1693 <a href="#t_vector">vector</a>,
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]1694 <a href="#t_struct">structure</a>,
1695 <a href="#t_array">array</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1696 <a href="#t_label">label</a>,
1697 <a href="#t_metadata">metadata</a>.
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1698 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1699 </tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1700 <tr>
1701 <td><a href="#t_primitive">primitive</a></td>
1702 <td><a href="#t_label">label</a>,
1703 <a href="#t_void">void</a>,
Tobias Grosser05387292010-12-28 20:29:31 +0000[diff] [blame]1704 <a href="#t_integer">integer</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1705 <a href="#t_floating">floating point</a>,
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1706 <a href="#t_x86mmx">x86mmx</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1707 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1708 </tr>
1709 <tr>
1710 <td><a href="#t_derived">derived</a></td>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1711 <td><a href="#t_array">array</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1712 <a href="#t_function">function</a>,
1713 <a href="#t_pointer">pointer</a>,
1714 <a href="#t_struct">structure</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1715 <a href="#t_vector">vector</a>,
1716 <a href="#t_opaque">opaque</a>.
Dan Gohman01ac1012008-10-14 16:32:04 +0000[diff] [blame]1717 </td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1718 </tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1719 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1720</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1721
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1722<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1723 important. Values of these types are the only ones which can be produced by
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1724 instructions.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1725
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1726</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1727
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1728<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1729<h3>
1730 <a name="t_primitive">Primitive Types</a>
1731</h3>
Chris Lattner8f8c7b72008-01-04 04:34:14 +0000[diff] [blame]1732
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1733<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1734
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1735<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1736 system.</p>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1737
1738<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1739<h4>
1740 <a name="t_integer">Integer Type</a>
1741</h4>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1742
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1743<div>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1744
1745<h5>Overview:</h5>
1746<p>The integer type is a very simple type that simply specifies an arbitrary
1747 bit width for the integer type desired. Any bit width from 1 bit to
1748 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1749
1750<h5>Syntax:</h5>
1751<pre>
1752 iN
1753</pre>
1754
1755<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1756 value.</p>
1757
1758<h5>Examples:</h5>
1759<table class="layout">
1760 <tr class="layout">
1761 <td class="left"><tt>i1</tt></td>
1762 <td class="left">a single-bit integer.</td>
1763 </tr>
1764 <tr class="layout">
1765 <td class="left"><tt>i32</tt></td>
1766 <td class="left">a 32-bit integer.</td>
1767 </tr>
1768 <tr class="layout">
1769 <td class="left"><tt>i1942652</tt></td>
1770 <td class="left">a really big integer of over 1 million bits.</td>
1771 </tr>
1772</table>
1773
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1774</div>
1775
1776<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1777<h4>
1778 <a name="t_floating">Floating Point Types</a>
1779</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1780
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1781<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1782
1783<table>
1784 <tbody>
1785 <tr><th>Type</th><th>Description</th></tr>
1786 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1787 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1788 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1789 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1790 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1791 </tbody>
1792</table>
1793
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1794</div>
1795
1796<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1797<h4>
1798 <a name="t_x86mmx">X86mmx Type</a>
1799</h4>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1800
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1801<div>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1802
1803<h5>Overview:</h5>
1804<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>
1805
1806<h5>Syntax:</h5>
1807<pre>
Dale Johannesen473a8c82010-10-01 01:07:02 +0000[diff] [blame]1808 x86mmx
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1809</pre>
1810
1811</div>
1812
1813<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1814<h4>
1815 <a name="t_void">Void Type</a>
1816</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1817
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1818<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1819
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1820<h5>Overview:</h5>
1821<p>The void type does not represent any value and has no size.</p>
1822
1823<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1824<pre>
1825 void
1826</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1827
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1828</div>
1829
1830<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1831<h4>
1832 <a name="t_label">Label Type</a>
1833</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1834
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1835<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1836
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1837<h5>Overview:</h5>
1838<p>The label type represents code labels.</p>
1839
1840<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1841<pre>
1842 label
1843</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1844
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1845</div>
1846
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1847<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1848<h4>
1849 <a name="t_metadata">Metadata Type</a>
1850</h4>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1851
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1852<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1853
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1854<h5>Overview:</h5>
Nick Lewyckyc261df92009-09-27 23:27:42 +0000[diff] [blame]1855<p>The metadata type represents embedded metadata. No derived types may be
1856 created from metadata except for <a href="#t_function">function</a>
1857 arguments.
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1858
1859<h5>Syntax:</h5>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1860<pre>
1861 metadata
1862</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1863
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1864</div>
1865
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1866</div>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1867
1868<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1869<h3>
1870 <a name="t_derived">Derived Types</a>
1871</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1872
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1873<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1874
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1875<p>The real power in LLVM comes from the derived types in the system. This is
1876 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1877 useful types. Each of these types contain one or more element types which
1878 may be a primitive type, or another derived type. For example, it is
1879 possible to have a two dimensional array, using an array as the element type
1880 of another array.</p>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]1881
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]1882</div>
1883
1884
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1885<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1886<h4>
1887 <a name="t_aggregate">Aggregate Types</a>
1888</h4>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1889
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1890<div>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1891
1892<p>Aggregate Types are a subset of derived types that can contain multiple
1893 member types. <a href="#t_array">Arrays</a>,
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]1894 <a href="#t_struct">structs</a>, and <a href="#t_vector">vectors</a> are
1895 aggregate types.</p>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1896
1897</div>
1898
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1899<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1900<h4>
1901 <a name="t_array">Array Type</a>
1902</h4>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1903
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1904<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1905
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1906<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1907<p>The array type is a very simple derived type that arranges elements
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1908 sequentially in memory. The array type requires a size (number of elements)
1909 and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1910
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1911<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1912<pre>
1913 [&lt;# elements&gt; x &lt;elementtype&gt;]
1914</pre>
1915
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1916<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1917 be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1918
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1919<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1920<table class="layout">
1921 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]1922 <td class="left"><tt>[40 x i32]</tt></td>
1923 <td class="left">Array of 40 32-bit integer values.</td>
1924 </tr>
1925 <tr class="layout">
1926 <td class="left"><tt>[41 x i32]</tt></td>
1927 <td class="left">Array of 41 32-bit integer values.</td>
1928 </tr>
1929 <tr class="layout">
1930 <td class="left"><tt>[4 x i8]</tt></td>
1931 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1932 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1933</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1934<p>Here are some examples of multidimensional arrays:</p>
1935<table class="layout">
1936 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]1937 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1938 <td class="left">3x4 array of 32-bit integer values.</td>
1939 </tr>
1940 <tr class="layout">
1941 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1942 <td class="left">12x10 array of single precision floating point values.</td>
1943 </tr>
1944 <tr class="layout">
1945 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1946 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1947 </tr>
1948</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1949
Dan Gohman7657f6b2009-11-09 19:01:53 +0000[diff] [blame]1950<p>There is no restriction on indexing beyond the end of the array implied by
1951 a static type (though there are restrictions on indexing beyond the bounds
1952 of an allocated object in some cases). This means that single-dimension
1953 'variable sized array' addressing can be implemented in LLVM with a zero
1954 length array type. An implementation of 'pascal style arrays' in LLVM could
1955 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1956
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1957</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1958
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1959<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1960<h4>
1961 <a name="t_function">Function Type</a>
1962</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1963
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1964<div>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1965
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1966<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1967<p>The function type can be thought of as a function signature. It consists of
1968 a return type and a list of formal parameter types. The return type of a
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]1969 function type is a first class type or a void type.</p>
Devang Patelc3fc6df2008-03-10 20:49:15 +0000[diff] [blame]1970
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1971<h5>Syntax:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1972<pre>
Nick Lewycky51386942009-09-27 07:55:32 +0000[diff] [blame]1973 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1974</pre>
1975
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1976<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1977 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1978 which indicates that the function takes a variable number of arguments.
1979 Variable argument functions can access their arguments with
1980 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]1981 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewyckyc261df92009-09-27 23:27:42 +0000[diff] [blame]1982 <a href="#t_label">label</a>.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1983
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1984<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1985<table class="layout">
1986 <tr class="layout">
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1987 <td class="left"><tt>i32 (i32)</tt></td>
1988 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1989 </td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1990 </tr><tr class="layout">
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]1991 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencerf17a0b72006-12-31 07:20:23 +0000[diff] [blame]1992 </tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1993 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]1994 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1995 returning <tt>float</tt>.
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]1996 </td>
1997 </tr><tr class="layout">
1998 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1999 <td class="left">A vararg function that takes at least one
2000 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2001 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2002 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2003 </td>
Devang Patela582f402008-03-24 05:35:41 +0000[diff] [blame]2004 </tr><tr class="layout">
2005 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky51386942009-09-27 07:55:32 +0000[diff] [blame]2006 <td class="left">A function taking an <tt>i32</tt>, returning a
2007 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patela582f402008-03-24 05:35:41 +0000[diff] [blame]2008 </td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2009 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2010</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2011
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2012</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2013
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2014<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2015<h4>
2016 <a name="t_struct">Structure Type</a>
2017</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2018
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2019<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2020
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2021<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2022<p>The structure type is used to represent a collection of data members together
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2023 in memory. The elements of a structure may be any type that has a size.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2024
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +0000[diff] [blame]2025<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2026 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2027 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2028 Structures in registers are accessed using the
2029 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2030 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2031
2032<p>Structures may optionally be "packed" structures, which indicate that the
2033 alignment of the struct is one byte, and that there is no padding between
2034 the elements. In non-packed structs, padding between field types is defined
2035 by the target data string to match the underlying processor.</p>
2036
2037<p>Structures can either be "anonymous" or "named". An anonymous structure is
2038 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) and a named types
2039 are always defined at the top level with a name. Anonmyous types are uniqued
2040 by their contents and can never be recursive since there is no way to write
2041 one. Named types can be recursive.
2042</p>
2043
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2044<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2045<pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2046 %T1 = type { &lt;type list&gt; } <i>; Named normal struct type</i>
2047 %T2 = type &lt;{ &lt;type list&gt; }&gt; <i>; Named packed struct type</i>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2048</pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2049
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2050<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2051<table class="layout">
2052 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2053 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2054 <td class="left">A triple of three <tt>i32</tt> values</td>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2055 </tr>
2056 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2057 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
2058 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2059 second element is a <a href="#t_pointer">pointer</a> to a
2060 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2061 an <tt>i32</tt>.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2062 </tr>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2063 <tr class="layout">
2064 <td class="left"><tt>&lt;{ i8, i32 }&gt;</tt></td>
2065 <td class="left">A packed struct known to be 5 bytes in size.</td>
2066 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2067</table>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]2068
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2069</div>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2070
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2071<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2072<h4>
Chris Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]2073 <a name="t_opaque">Opaque Structure Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2074</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2075
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2076<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2077
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2078<h5>Overview:</h5>
Chris Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]2079<p>Opaque structure types are used to represent named structure types that do
2080 not have a body specified. This corresponds (for example) to the C notion of
2081 a forward declared structure.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2082
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2083<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2084<pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2085 %X = type opaque
2086 %52 = type opaque
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2087</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2088
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2089<h5>Examples:</h5>
2090<table class="layout">
2091 <tr class="layout">
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2092 <td class="left"><tt>opaque</tt></td>
2093 <td class="left">An opaque type.</td>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2094 </tr>
2095</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2096
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2097</div>
2098
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2099
2100
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2101<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2102<h4>
2103 <a name="t_pointer">Pointer Type</a>
2104</h4>
Chris Lattner0fd4a272009-02-08 19:53:29 +0000[diff] [blame]2105
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2106<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2107
2108<h5>Overview:</h5>
Dan Gohmanff3ef322010-02-25 16:50:07 +0000[diff] [blame]2109<p>The pointer type is used to specify memory locations.
2110 Pointers are commonly used to reference objects in memory.</p>
2111
2112<p>Pointer types may have an optional address space attribute defining the
2113 numbered address space where the pointed-to object resides. The default
2114 address space is number zero. The semantics of non-zero address
2115 spaces are target-specific.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2116
2117<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2118 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner0fd4a272009-02-08 19:53:29 +0000[diff] [blame]2119
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2120<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2121<pre>
2122 &lt;type&gt; *
2123</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2124
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2125<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2126<table class="layout">
2127 <tr class="layout">
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]2128 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2129 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2130 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2131 </tr>
2132 <tr class="layout">
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]2133 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2134 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2135 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2136 <tt>i32</tt>.</td>
2137 </tr>
2138 <tr class="layout">
2139 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2140 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2141 that resides in address space #5.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2142 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2143</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2144
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2145</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2146
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2147<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2148<h4>
2149 <a name="t_vector">Vector Type</a>
2150</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2151
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2152<div>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2153
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2154<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2155<p>A vector type is a simple derived type that represents a vector of elements.
2156 Vector types are used when multiple primitive data are operated in parallel
2157 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sandsd40d14e2009-11-27 13:38:03 +0000[diff] [blame]2158 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2159 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2160
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2161<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2162<pre>
2163 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
2164</pre>
2165
Chris Lattner7d2e7be2010-10-10 18:20:35 +0000[diff] [blame]2166<p>The number of elements is a constant integer value larger than 0; elementtype
2167 may be any integer or floating point type. Vectors of size zero are not
2168 allowed, and pointers are not allowed as the element type.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2169
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2170<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2171<table class="layout">
2172 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2173 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
2174 <td class="left">Vector of 4 32-bit integer values.</td>
2175 </tr>
2176 <tr class="layout">
2177 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
2178 <td class="left">Vector of 8 32-bit floating-point values.</td>
2179 </tr>
2180 <tr class="layout">
2181 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
2182 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2183 </tr>
2184</table>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]2185
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2186</div>
2187
Bill Wendlingaf75f0c2011-07-31 06:47:33 +0000[diff] [blame]2188</div>
2189
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2190<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2191<h2><a name="constants">Constants</a></h2>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2192<!-- *********************************************************************** -->
2193
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2194<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2195
2196<p>LLVM has several different basic types of constants. This section describes
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2197 them all and their syntax.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2198
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2199<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2200<h3>
2201 <a name="simpleconstants">Simple Constants</a>
2202</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2203
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2204<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2205
2206<dl>
2207 <dt><b>Boolean constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2208 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]2209 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2210
2211 <dt><b>Integer constants</b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2212 <dd>Standard integers (such as '4') are constants of
2213 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2214 with integer types.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2215
2216 <dt><b>Floating point constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2217 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2218 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2219 notation (see below). The assembler requires the exact decimal value of a
2220 floating-point constant. For example, the assembler accepts 1.25 but
2221 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2222 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2223
2224 <dt><b>Null pointer constants</b></dt>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2225 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2226 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2227</dl>
2228
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2229<p>The one non-intuitive notation for constants is the hexadecimal form of
2230 floating point constants. For example, the form '<tt>double
2231 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2232 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2233 constants are required (and the only time that they are generated by the
2234 disassembler) is when a floating point constant must be emitted but it cannot
2235 be represented as a decimal floating point number in a reasonable number of
2236 digits. For example, NaN's, infinities, and other special values are
2237 represented in their IEEE hexadecimal format so that assembly and disassembly
2238 do not cause any bits to change in the constants.</p>
2239
Dale Johannesenbd5e5a82009-02-11 22:14:51 +0000[diff] [blame]2240<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2241 represented using the 16-digit form shown above (which matches the IEEE754
2242 representation for double); float values must, however, be exactly
2243 representable as IEE754 single precision. Hexadecimal format is always used
2244 for long double, and there are three forms of long double. The 80-bit format
2245 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2246 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2247 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2248 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2249 currently supported target uses this format. Long doubles will only work if
2250 they match the long double format on your target. All hexadecimal formats
2251 are big-endian (sign bit at the left).</p>
2252
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]2253<p>There are no constants of type x86mmx.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2254</div>
2255
2256<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2257<h3>
Bill Wendlingd9fe2982009-07-20 02:32:41 +0000[diff] [blame]2258<a name="aggregateconstants"></a> <!-- old anchor -->
2259<a name="complexconstants">Complex Constants</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2260</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2261
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2262<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2263
Chris Lattner70882792009-02-28 18:32:25 +0000[diff] [blame]2264<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2265 constants and smaller complex constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2266
2267<dl>
2268 <dt><b>Structure constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2269 <dd>Structure constants are represented with notation similar to structure
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2270 type definitions (a comma separated list of elements, surrounded by braces
2271 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2272 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2273 Structure constants must have <a href="#t_struct">structure type</a>, and
2274 the number and types of elements must match those specified by the
2275 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2276
2277 <dt><b>Array constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2278 <dd>Array constants are represented with notation similar to array type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2279 definitions (a comma separated list of elements, surrounded by square
2280 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2281 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2282 the number and types of elements must match those specified by the
2283 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2284
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2285 <dt><b>Vector constants</b></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2286 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2287 definitions (a comma separated list of elements, surrounded by
2288 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2289 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2290 have <a href="#t_vector">vector type</a>, and the number and types of
2291 elements must match those specified by the type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2292
2293 <dt><b>Zero initialization</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2294 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]2295 value to zero of <em>any</em> type, including scalar and
2296 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2297 This is often used to avoid having to print large zero initializers
2298 (e.g. for large arrays) and is always exactly equivalent to using explicit
2299 zero initializers.</dd>
Nick Lewycky21cc4462009-04-04 07:22:01 +0000[diff] [blame]2300
2301 <dt><b>Metadata node</b></dt>
Nick Lewycky1e8c7a62009-05-30 16:08:30 +0000[diff] [blame]2302 <dd>A metadata node is a structure-like constant with
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2303 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2304 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2305 be interpreted as part of the instruction stream, metadata is a place to
2306 attach additional information such as debug info.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2307</dl>
2308
2309</div>
2310
2311<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2312<h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2313 <a name="globalconstants">Global Variable and Function Addresses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2314</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2315
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2316<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2317
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2318<p>The addresses of <a href="#globalvars">global variables</a>
2319 and <a href="#functionstructure">functions</a> are always implicitly valid
2320 (link-time) constants. These constants are explicitly referenced when
2321 the <a href="#identifiers">identifier for the global</a> is used and always
2322 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2323 legal LLVM file:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2324
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2325<pre class="doc_code">
Chris Lattnera18a4242007-06-06 18:28:13 +0000[diff] [blame]2326@X = global i32 17
2327@Y = global i32 42
2328@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2329</pre>
2330
2331</div>
2332
2333<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2334<h3>
2335 <a name="undefvalues">Undefined Values</a>
2336</h3>
2337
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2338<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2339
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2340<p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]2341 indicates that the user of the value may receive an unspecified bit-pattern.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2342 Undefined values may be of any type (other than '<tt>label</tt>'
2343 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2344
Chris Lattnerc608cb12009-09-11 01:49:31 +0000[diff] [blame]2345<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2346 program is well defined no matter what value is used. This gives the
2347 compiler more freedom to optimize. Here are some examples of (potentially
2348 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2349
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2350
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2351<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2352 %A = add %X, undef
2353 %B = sub %X, undef
2354 %C = xor %X, undef
2355Safe:
2356 %A = undef
2357 %B = undef
2358 %C = undef
2359</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2360
2361<p>This is safe because all of the output bits are affected by the undef bits.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2362 Any output bit can have a zero or one depending on the input bits.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2363
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2364<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2365 %A = or %X, undef
2366 %B = and %X, undef
2367Safe:
2368 %A = -1
2369 %B = 0
2370Unsafe:
2371 %A = undef
2372 %B = undef
2373</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2374
2375<p>These logical operations have bits that are not always affected by the input.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2376 For example, if <tt>%X</tt> has a zero bit, then the output of the
2377 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2378 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2379 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2380 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2381 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2382 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2383 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2384
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2385<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2386 %A = select undef, %X, %Y
2387 %B = select undef, 42, %Y
2388 %C = select %X, %Y, undef
2389Safe:
2390 %A = %X (or %Y)
2391 %B = 42 (or %Y)
2392 %C = %Y
2393Unsafe:
2394 %A = undef
2395 %B = undef
2396 %C = undef
2397</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2398
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2399<p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2400 branch) conditions can go <em>either way</em>, but they have to come from one
2401 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2402 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2403 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2404 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2405 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2406 eliminated.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2407
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2408<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2409 %A = xor undef, undef
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2410
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2411 %B = undef
2412 %C = xor %B, %B
2413
2414 %D = undef
2415 %E = icmp lt %D, 4
2416 %F = icmp gte %D, 4
2417
2418Safe:
2419 %A = undef
2420 %B = undef
2421 %C = undef
2422 %D = undef
2423 %E = undef
2424 %F = undef
2425</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2426
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2427<p>This example points out that two '<tt>undef</tt>' operands are not
2428 necessarily the same. This can be surprising to people (and also matches C
2429 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2430 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2431 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2432 its value over its "live range". This is true because the variable doesn't
2433 actually <em>have a live range</em>. Instead, the value is logically read
2434 from arbitrary registers that happen to be around when needed, so the value
2435 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2436 need to have the same semantics or the core LLVM "replace all uses with"
2437 concept would not hold.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2438
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2439<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2440 %A = fdiv undef, %X
2441 %B = fdiv %X, undef
2442Safe:
2443 %A = undef
2444b: unreachable
2445</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2446
2447<p>These examples show the crucial difference between an <em>undefined
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2448 value</em> and <em>undefined behavior</em>. An undefined value (like
2449 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2450 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2451 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2452 defined on SNaN's. However, in the second example, we can make a more
2453 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2454 arbitrary value, we are allowed to assume that it could be zero. Since a
2455 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2456 the operation does not execute at all. This allows us to delete the divide and
2457 all code after it. Because the undefined operation "can't happen", the
2458 optimizer can assume that it occurs in dead code.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2459
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2460<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2461a: store undef -> %X
2462b: store %X -> undef
2463Safe:
2464a: &lt;deleted&gt;
2465b: unreachable
2466</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2467
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2468<p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2469 undefined value can be assumed to not have any effect; we can assume that the
2470 value is overwritten with bits that happen to match what was already there.
2471 However, a store <em>to</em> an undefined location could clobber arbitrary
2472 memory, therefore, it has undefined behavior.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2473
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2474</div>
2475
2476<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2477<h3>
2478 <a name="trapvalues">Trap Values</a>
2479</h3>
2480
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2481<div>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2482
Dan Gohmanc68ce062010-04-26 20:21:21 +0000[diff] [blame]2483<p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2484 instead of representing an unspecified bit pattern, they represent the
2485 fact that an instruction or constant expression which cannot evoke side
2486 effects has nevertheless detected a condition which results in undefined
Dan Gohmanc68ce062010-04-26 20:21:21 +0000[diff] [blame]2487 behavior.</p>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2488
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2489<p>There is currently no way of representing a trap value in the IR; they
Dan Gohman855abed2010-05-03 14:51:43 +0000[diff] [blame]2490 only exist when produced by operations such as
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2491 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2492
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2493<p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2494
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2495<ul>
2496<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2497 their operands.</li>
2498
2499<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2500 to their dynamic predecessor basic block.</li>
2501
2502<li>Function arguments depend on the corresponding actual argument values in
2503 the dynamic callers of their functions.</li>
2504
2505<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2506 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2507 control back to them.</li>
2508
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2509<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2510 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2511 or exception-throwing call instructions that dynamically transfer control
2512 back to them.</li>
2513
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2514<li>Non-volatile loads and stores depend on the most recent stores to all of the
2515 referenced memory addresses, following the order in the IR
2516 (including loads and stores implied by intrinsics such as
2517 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2518
Dan Gohman7c24ff12010-05-03 14:59:34 +0000[diff] [blame]2519<!-- TODO: In the case of multiple threads, this only applies if the store
2520 "happens-before" the load or store. -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2521
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2522<!-- TODO: floating-point exception state -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2523
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2524<li>An instruction with externally visible side effects depends on the most
2525 recent preceding instruction with externally visible side effects, following
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]2526 the order in the IR. (This includes
2527 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2528
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2529<li>An instruction <i>control-depends</i> on a
2530 <a href="#terminators">terminator instruction</a>
2531 if the terminator instruction has multiple successors and the instruction
2532 is always executed when control transfers to one of the successors, and
Chris Lattner7a2bdde2011-04-15 05:18:47 +0000[diff] [blame]2533 may not be executed when control is transferred to another.</li>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2534
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2535<li>Additionally, an instruction also <i>control-depends</i> on a terminator
2536 instruction if the set of instructions it otherwise depends on would be
Chris Lattner7a2bdde2011-04-15 05:18:47 +0000[diff] [blame]2537 different if the terminator had transferred control to a different
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2538 successor.</li>
2539
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2540<li>Dependence is transitive.</li>
2541
2542</ul>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2543
2544<p>Whenever a trap value is generated, all values which depend on it evaluate
2545 to trap. If they have side effects, the evoke their side effects as if each
2546 operand with a trap value were undef. If they have externally-visible side
2547 effects, the behavior is undefined.</p>
2548
2549<p>Here are some examples:</p>
Dan Gohmanc30f6e12010-04-26 20:54:53 +0000[diff] [blame]2550
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2551<pre class="doc_code">
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2552entry:
2553 %trap = sub nuw i32 0, 1 ; Results in a trap value.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2554 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2555 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2556 store i32 0, i32* %trap_yet_again ; undefined behavior
2557
2558 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2559 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2560
2561 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2562
2563 %narrowaddr = bitcast i32* @g to i16*
2564 %wideaddr = bitcast i32* @g to i64*
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2565 %trap3 = load i16* %narrowaddr ; Returns a trap value.
2566 %trap4 = load i64* %wideaddr ; Returns a trap value.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2567
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2568 %cmp = icmp slt i32 %trap, 0 ; Returns a trap value.
2569 br i1 %cmp, label %true, label %end ; Branch to either destination.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2570
2571true:
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2572 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2573 ; it has undefined behavior.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2574 br label %end
2575
2576end:
2577 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2578 ; Both edges into this PHI are
2579 ; control-dependent on %cmp, so this
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2580 ; always results in a trap value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2581
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2582 volatile store i32 0, i32* @g ; This would depend on the store in %true
2583 ; if %cmp is true, or the store in %entry
2584 ; otherwise, so this is undefined behavior.
2585
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2586 br i1 %cmp, label %second_true, label %second_end
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2587 ; The same branch again, but this time the
2588 ; true block doesn't have side effects.
2589
2590second_true:
2591 ; No side effects!
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2592 ret void
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2593
2594second_end:
2595 volatile store i32 0, i32* @g ; This time, the instruction always depends
2596 ; on the store in %end. Also, it is
2597 ; control-equivalent to %end, so this is
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2598 ; well-defined (again, ignoring earlier
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2599 ; undefined behavior in this example).
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2600</pre>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2601
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2602</div>
2603
2604<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2605<h3>
2606 <a name="blockaddress">Addresses of Basic Blocks</a>
2607</h3>
2608
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2609<div>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2610
Chris Lattnercdfc9402009-11-01 01:27:45 +0000[diff] [blame]2611<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2612
2613<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner2dfdf2a2009-10-27 21:49:40 +0000[diff] [blame]2614 basic block in the specified function, and always has an i8* type. Taking
Chris Lattnercdfc9402009-11-01 01:27:45 +0000[diff] [blame]2615 the address of the entry block is illegal.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2616
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2617<p>This value only has defined behavior when used as an operand to the
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2618 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2619 comparisons against null. Pointer equality tests between labels addresses
2620 results in undefined behavior &mdash; though, again, comparison against null
2621 is ok, and no label is equal to the null pointer. This may be passed around
2622 as an opaque pointer sized value as long as the bits are not inspected. This
2623 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2624 long as the original value is reconstituted before the <tt>indirectbr</tt>
2625 instruction.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2626
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2627<p>Finally, some targets may provide defined semantics when using the value as
2628 the operand to an inline assembly, but that is target specific.</p>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2629
2630</div>
2631
2632
2633<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2634<h3>
2635 <a name="constantexprs">Constant Expressions</a>
2636</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2637
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2638<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2639
2640<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2641 to be used as constants. Constant expressions may be of
2642 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2643 operation that does not have side effects (e.g. load and call are not
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2644 supported). The following is the syntax for constant expressions:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2645
2646<dl>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2647 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2648 <dd>Truncate a constant to another type. The bit size of CST must be larger
2649 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2650
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2651 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2652 <dd>Zero extend a constant to another type. The bit size of CST must be
Duncan Sands28afd432010-07-13 12:06:14 +0000[diff] [blame]2653 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2654
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2655 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2656 <dd>Sign extend a constant to another type. The bit size of CST must be
Duncan Sands28afd432010-07-13 12:06:14 +0000[diff] [blame]2657 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2658
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2659 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2660 <dd>Truncate a floating point constant to another floating point type. The
2661 size of CST must be larger than the size of TYPE. Both types must be
2662 floating point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2663
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2664 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2665 <dd>Floating point extend a constant to another type. The size of CST must be
2666 smaller or equal to the size of TYPE. Both types must be floating
2667 point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2668
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2669 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2670 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2671 constant. TYPE must be a scalar or vector integer type. CST must be of
2672 scalar or vector floating point type. Both CST and TYPE must be scalars,
2673 or vectors of the same number of elements. If the value won't fit in the
2674 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2675
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2676 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2677 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2678 constant. TYPE must be a scalar or vector integer type. CST must be of
2679 scalar or vector floating point type. Both CST and TYPE must be scalars,
2680 or vectors of the same number of elements. If the value won't fit in the
2681 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2682
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2683 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2684 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2685 constant. TYPE must be a scalar or vector floating point type. CST must be
2686 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2687 vectors of the same number of elements. If the value won't fit in the
2688 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2689
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2690 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2691 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2692 constant. TYPE must be a scalar or vector floating point type. CST must be
2693 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2694 vectors of the same number of elements. If the value won't fit in the
2695 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2696
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2697 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2698 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2699 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2700 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2701 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2702
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2703 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2704 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2705 type. CST must be of integer type. The CST value is zero extended,
2706 truncated, or unchanged to make it fit in a pointer size. This one is
2707 <i>really</i> dangerous!</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2708
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2709 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner03bbad62009-02-28 18:27:03 +0000[diff] [blame]2710 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2711 are the same as those for the <a href="#i_bitcast">bitcast
2712 instruction</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2713
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2714 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2715 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2716 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2717 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2718 instruction, the index list may have zero or more indexes, which are
2719 required to make sense for the type of "CSTPTR".</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2720
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2721 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2722 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2723
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2724 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2725 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2726
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2727 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2728 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]2729
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2730 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2731 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2732 constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]2733
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2734 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2735 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2736 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]2737
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2738 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2739 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2740 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]2741
Nick Lewycky9e130ce2010-05-29 06:44:15 +0000[diff] [blame]2742 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2743 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2744 constants. The index list is interpreted in a similar manner as indices in
2745 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2746 index value must be specified.</dd>
2747
2748 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2749 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2750 constants. The index list is interpreted in a similar manner as indices in
2751 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2752 index value must be specified.</dd>
2753
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2754 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2755 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2756 be any of the <a href="#binaryops">binary</a>
2757 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2758 on operands are the same as those for the corresponding instruction
2759 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2760</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2761
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2762</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000[diff] [blame]2763
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2764</div>
2765
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2766<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2767<h2><a name="othervalues">Other Values</a></h2>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2768<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2769<div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2770<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2771<h3>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2772<a name="inlineasm">Inline Assembler Expressions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2773</h3>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2774
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2775<div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2776
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2777<p>LLVM supports inline assembler expressions (as opposed
2778 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2779 a special value. This value represents the inline assembler as a string
2780 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2781 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2782 expression has side effects, and a flag indicating whether the function
2783 containing the asm needs to align its stack conservatively. An example
2784 inline assembler expression is:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2785
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2786<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2787i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2788</pre>
2789
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2790<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2791 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2792 have:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2793
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2794<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2795%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2796</pre>
2797
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2798<p>Inline asms with side effects not visible in the constraint list must be
2799 marked as having side effects. This is done through the use of the
2800 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2801
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2802<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2803call void asm sideeffect "eieio", ""()
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2804</pre>
2805
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2806<p>In some cases inline asms will contain code that will not work unless the
2807 stack is aligned in some way, such as calls or SSE instructions on x86,
2808 yet will not contain code that does that alignment within the asm.
2809 The compiler should make conservative assumptions about what the asm might
2810 contain and should generate its usual stack alignment code in the prologue
2811 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2812
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2813<pre class="doc_code">
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2814call void asm alignstack "eieio", ""()
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2815</pre>
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2816
2817<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2818 first.</p>
2819
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2820<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2821 documented here. Constraints on what can be done (e.g. duplication, moving,
2822 etc need to be documented). This is probably best done by reference to
2823 another document that covers inline asm from a holistic perspective.</p>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2824
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2825<h4>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2826<a name="inlineasm_md">Inline Asm Metadata</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2827</h4>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2828
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2829<div>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2830
2831<p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
Chris Lattnerce1b9ad2010-11-17 08:20:42 +0000[diff] [blame]2832 attached to it that contains a list of constant integers. If present, the
2833 code generator will use the integer as the location cookie value when report
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2834 errors through the LLVMContext error reporting mechanisms. This allows a
Dan Gohman1c70c002010-04-28 00:36:01 +0000[diff] [blame]2835 front-end to correlate backend errors that occur with inline asm back to the
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2836 source code that produced it. For example:</p>
2837
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2838<pre class="doc_code">
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2839call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2840...
2841!42 = !{ i32 1234567 }
2842</pre>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2843
2844<p>It is up to the front-end to make sense of the magic numbers it places in the
Chris Lattnerce1b9ad2010-11-17 08:20:42 +0000[diff] [blame]2845 IR. If the MDNode contains multiple constants, the code generator will use
2846 the one that corresponds to the line of the asm that the error occurs on.</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2847
2848</div>
2849
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2850</div>
2851
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2852<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2853<h3>
2854 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2855</h3>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2856
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2857<div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2858
2859<p>LLVM IR allows metadata to be attached to instructions in the program that
2860 can convey extra information about the code to the optimizers and code
2861 generator. One example application of metadata is source-level debug
2862 information. There are two metadata primitives: strings and nodes. All
2863 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2864 preceding exclamation point ('<tt>!</tt>').</p>
2865
2866<p>A metadata string is a string surrounded by double quotes. It can contain
2867 any character by escaping non-printable characters with "\xx" where "xx" is
2868 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2869
2870<p>Metadata nodes are represented with notation similar to structure constants
2871 (a comma separated list of elements, surrounded by braces and preceded by an
2872 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2873 10}</tt>". Metadata nodes can have any values as their operand.</p>
2874
2875<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2876 metadata nodes, which can be looked up in the module symbol table. For
2877 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2878
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2879<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2880 function is using two metadata arguments.</p>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2881
Bill Wendling9ff5de92011-03-02 02:17:11 +0000[diff] [blame]2882<div class="doc_code">
2883<pre>
2884call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2885</pre>
2886</div>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2887
2888<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2889 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.</p>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2890
Bill Wendling9ff5de92011-03-02 02:17:11 +0000[diff] [blame]2891<div class="doc_code">
2892<pre>
2893%indvar.next = add i64 %indvar, 1, !dbg !21
2894</pre>
2895</div>
2896
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2897</div>
2898
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2899</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2900
2901<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2902<h2>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2903 <a name="intrinsic_globals">Intrinsic Global Variables</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2904</h2>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2905<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2906<div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2907<p>LLVM has a number of "magic" global variables that contain data that affect
2908code generation or other IR semantics. These are documented here. All globals
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]2909of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2910section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2911by LLVM.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2912
2913<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2914<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2915<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2916</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2917
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2918<div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2919
2920<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2921href="#linkage_appending">appending linkage</a>. This array contains a list of
2922pointers to global variables and functions which may optionally have a pointer
2923cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2924
2925<pre>
2926 @X = global i8 4
2927 @Y = global i32 123
2928
2929 @llvm.used = appending global [2 x i8*] [
2930 i8* @X,
2931 i8* bitcast (i32* @Y to i8*)
2932 ], section "llvm.metadata"
2933</pre>
2934
2935<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2936compiler, assembler, and linker are required to treat the symbol as if there is
2937a reference to the global that it cannot see. For example, if a variable has
2938internal linkage and no references other than that from the <tt>@llvm.used</tt>
2939list, it cannot be deleted. This is commonly used to represent references from
2940inline asms and other things the compiler cannot "see", and corresponds to
2941"attribute((used))" in GNU C.</p>
2942
2943<p>On some targets, the code generator must emit a directive to the assembler or
2944object file to prevent the assembler and linker from molesting the symbol.</p>
2945
2946</div>
2947
2948<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2949<h3>
2950 <a name="intg_compiler_used">
2951 The '<tt>llvm.compiler.used</tt>' Global Variable
2952 </a>
2953</h3>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]2954
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2955<div>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]2956
2957<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2958<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2959touching the symbol. On targets that support it, this allows an intelligent
2960linker to optimize references to the symbol without being impeded as it would be
2961by <tt>@llvm.used</tt>.</p>
2962
2963<p>This is a rare construct that should only be used in rare circumstances, and
2964should not be exposed to source languages.</p>
2965
2966</div>
2967
2968<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2969<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2970<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2971</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2972
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2973<div>
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]2974<pre>
2975%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +0000[diff] [blame]2976@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]2977</pre>
2978<p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor functions and associated priorities. The functions referenced by this array will be called in ascending order of priority (i.e. lowest first) when the module is loaded. The order of functions with the same priority is not defined.
2979</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2980
2981</div>
2982
2983<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2984<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2985<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2986</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2987
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2988<div>
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]2989<pre>
2990%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +0000[diff] [blame]2991@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]2992</pre>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2993
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]2994<p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions and associated priorities. The functions referenced by this array will be called in descending order of priority (i.e. highest first) when the module is loaded. The order of functions with the same priority is not defined.
2995</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]2996
2997</div>
2998
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2999</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3000
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]3001<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3002<h2><a name="instref">Instruction Reference</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3003<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3004
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3005<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3006
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3007<p>The LLVM instruction set consists of several different classifications of
3008 instructions: <a href="#terminators">terminator
3009 instructions</a>, <a href="#binaryops">binary instructions</a>,
3010 <a href="#bitwiseops">bitwise binary instructions</a>,
3011 <a href="#memoryops">memory instructions</a>, and
3012 <a href="#otherops">other instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3013
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3014<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3015<h3>
3016 <a name="terminators">Terminator Instructions</a>
3017</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3018
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3019<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3020
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3021<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3022 in a program ends with a "Terminator" instruction, which indicates which
3023 block should be executed after the current block is finished. These
3024 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3025 control flow, not values (the one exception being the
3026 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3027
Chris Lattner6445ecb2011-08-02 20:29:13 +0000[diff] [blame]3028<p>The terminator instructions are:
3029 '<a href="#i_ret"><tt>ret</tt></a>',
3030 '<a href="#i_br"><tt>br</tt></a>',
3031 '<a href="#i_switch"><tt>switch</tt></a>',
3032 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3033 '<a href="#i_invoke"><tt>invoke</tt></a>',
3034 '<a href="#i_unwind"><tt>unwind</tt></a>',
3035 '<a href="#i_resume"><tt>resume</tt></a>', and
3036 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3037
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3038<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3039<h4>
3040 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3041</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3042
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3043<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3044
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3045<h5>Syntax:</h5>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]3046<pre>
3047 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]3048 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3049</pre>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3050
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3051<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3052<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3053 a value) from a function back to the caller.</p>
3054
3055<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3056 value and then causes control flow, and one that just causes control flow to
3057 occur.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3058
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3059<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3060<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3061 return value. The type of the return value must be a
3062 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]3063
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3064<p>A function is not <a href="#wellformed">well formed</a> if it it has a
3065 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3066 value or a return value with a type that does not match its type, or if it
3067 has a void return type and contains a '<tt>ret</tt>' instruction with a
3068 return value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3069
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3070<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3071<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3072 the calling function's context. If the caller is a
3073 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3074 instruction after the call. If the caller was an
3075 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3076 the beginning of the "normal" destination block. If the instruction returns
3077 a value, that value shall set the call or invoke instruction's return
3078 value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3079
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3080<h5>Example:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3081<pre>
3082 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]3083 ret void <i>; Return from a void function</i>
Bill Wendling0a4bbbf2009-02-28 22:12:54 +0000[diff] [blame]3084 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3085</pre>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +0000[diff] [blame]3086
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3087</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3088<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3089<h4>
3090 <a name="i_br">'<tt>br</tt>' Instruction</a>
3091</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3092
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3093<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3094
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3095<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3096<pre>
Bill Wendlingb3aa4712011-07-26 10:41:15 +0000[diff] [blame]3097 br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;
3098 br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3099</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3100
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3101<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3102<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3103 different basic block in the current function. There are two forms of this
3104 instruction, corresponding to a conditional branch and an unconditional
3105 branch.</p>
3106
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3107<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3108<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3109 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3110 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3111 target.</p>
3112
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3113<h5>Semantics:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3114<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3115 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3116 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3117 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3118
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3119<h5>Example:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]3120<pre>
3121Test:
3122 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3123 br i1 %cond, label %IfEqual, label %IfUnequal
3124IfEqual:
3125 <a href="#i_ret">ret</a> i32 1
3126IfUnequal:
3127 <a href="#i_ret">ret</a> i32 0
3128</pre>
3129
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3130</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3131
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3132<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3133<h4>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3134 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3135</h4>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3136
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3137<div>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3138
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3139<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3140<pre>
3141 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
3142</pre>
3143
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3144<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3145<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3146 several different places. It is a generalization of the '<tt>br</tt>'
3147 instruction, allowing a branch to occur to one of many possible
3148 destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3149
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3150<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3151<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3152 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3153 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3154 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3155
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3156<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3157<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3158 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3159 is searched for the given value. If the value is found, control flow is
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]3160 transferred to the corresponding destination; otherwise, control flow is
3161 transferred to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3162
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3163<h5>Implementation:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3164<p>Depending on properties of the target machine and the particular
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3165 <tt>switch</tt> instruction, this instruction may be code generated in
3166 different ways. For example, it could be generated as a series of chained
3167 conditional branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3168
3169<h5>Example:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3170<pre>
3171 <i>; Emulate a conditional br instruction</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3172 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]3173 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3174
3175 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3176 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3177
3178 <i>; Implement a jump table:</i>
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]3179 switch i32 %val, label %otherwise [ i32 0, label %onzero
3180 i32 1, label %onone
3181 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3182</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3183
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3184</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3185
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3186
3187<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3188<h4>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3189 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3190</h4>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3191
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3192<div>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3193
3194<h5>Syntax:</h5>
3195<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3196 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3197</pre>
3198
3199<h5>Overview:</h5>
3200
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3201<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3202 within the current function, whose address is specified by
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]3203 "<tt>address</tt>". Address must be derived from a <a
3204 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3205
3206<h5>Arguments:</h5>
3207
3208<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3209 rest of the arguments indicate the full set of possible destinations that the
3210 address may point to. Blocks are allowed to occur multiple times in the
3211 destination list, though this isn't particularly useful.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3212
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3213<p>This destination list is required so that dataflow analysis has an accurate
3214 understanding of the CFG.</p>
3215
3216<h5>Semantics:</h5>
3217
3218<p>Control transfers to the block specified in the address argument. All
3219 possible destination blocks must be listed in the label list, otherwise this
3220 instruction has undefined behavior. This implies that jumps to labels
3221 defined in other functions have undefined behavior as well.</p>
3222
3223<h5>Implementation:</h5>
3224
3225<p>This is typically implemented with a jump through a register.</p>
3226
3227<h5>Example:</h5>
3228<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3229 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3230</pre>
3231
3232</div>
3233
3234
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3235<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3236<h4>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3237 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3238</h4>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3239
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3240<div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3241
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3242<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3243<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]3244 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Chris Lattner76b8a332006-05-14 18:23:06 +0000[diff] [blame]3245 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3246</pre>
3247
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3248<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3249<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3250 function, with the possibility of control flow transfer to either the
3251 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3252 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3253 control flow will return to the "normal" label. If the callee (or any
3254 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3255 instruction, control is interrupted and continued at the dynamically nearest
3256 "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3257
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3258<p>The '<tt>exception</tt>' label is a
3259 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3260 exception. As such, '<tt>exception</tt>' label is required to have the
3261 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
3262 the information about about the behavior of the program after unwinding
3263 happens, as its first non-PHI instruction. The restrictions on the
3264 "<tt>landingpad</tt>" instruction's tightly couples it to the
3265 "<tt>invoke</tt>" instruction, so that the important information contained
3266 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3267 code motion.</p>
3268
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3269<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3270<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3271
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3272<ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3273 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3274 convention</a> the call should use. If none is specified, the call
3275 defaults to using C calling conventions.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]3276
3277 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3278 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3279 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]3280
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3281 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3282 function value being invoked. In most cases, this is a direct function
3283 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3284 off an arbitrary pointer to function value.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3285
3286 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3287 function to be invoked. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3288
3289 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]3290 signature argument types and parameter attributes. All arguments must be
3291 of <a href="#t_firstclass">first class</a> type. If the function
3292 signature indicates the function accepts a variable number of arguments,
3293 the extra arguments can be specified.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3294
3295 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3296 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3297
3298 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3299 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3300
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]3301 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3302 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3303 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3304</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3305
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3306<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3307<p>This instruction is designed to operate as a standard
3308 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3309 primary difference is that it establishes an association with a label, which
3310 is used by the runtime library to unwind the stack.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3311
3312<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3313 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3314 exception. Additionally, this is important for implementation of
3315 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3316
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3317<p>For the purposes of the SSA form, the definition of the value returned by the
3318 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3319 block to the "normal" label. If the callee unwinds then no return value is
3320 available.</p>
Dan Gohmanf96a4992009-05-22 21:47:08 +0000[diff] [blame]3321
Chris Lattnerdf7a6802010-01-15 18:08:37 +0000[diff] [blame]3322<p>Note that the code generator does not yet completely support unwind, and
3323that the invoke/unwind semantics are likely to change in future versions.</p>
3324
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3325<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3326<pre>
Nick Lewyckyd703f652008-03-16 07:18:12 +0000[diff] [blame]3327 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3328 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckyd703f652008-03-16 07:18:12 +0000[diff] [blame]3329 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3330 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3331</pre>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3332
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3333</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3334
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3335<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3336
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3337<h4>
3338 <a name="i_unwind">'<tt>unwind</tt>' Instruction</a>
3339</h4>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3340
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3341<div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3342
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3343<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3344<pre>
3345 unwind
3346</pre>
3347
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3348<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3349<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3350 at the first callee in the dynamic call stack which used
3351 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3352 This is primarily used to implement exception handling.</p>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3353
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3354<h5>Semantics:</h5>
Chris Lattner72ed2002008-04-19 21:01:16 +0000[diff] [blame]3355<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3356 immediately halt. The dynamic call stack is then searched for the
3357 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3358 Once found, execution continues at the "exceptional" destination block
3359 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3360 instruction in the dynamic call chain, undefined behavior results.</p>
3361
Chris Lattnerdf7a6802010-01-15 18:08:37 +0000[diff] [blame]3362<p>Note that the code generator does not yet completely support unwind, and
3363that the invoke/unwind semantics are likely to change in future versions.</p>
3364
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3365</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3366
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3367 <!-- _______________________________________________________________________ -->
3368
3369<h4>
3370 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3371</h4>
3372
3373<div>
3374
3375<h5>Syntax:</h5>
3376<pre>
3377 resume &lt;type&gt; &lt;value&gt;
3378</pre>
3379
3380<h5>Overview:</h5>
3381<p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3382 successors.</p>
3383
3384<h5>Arguments:</h5>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3385<p>The '<tt>resume</tt>' instruction requires one argument, which must have the
Bill Wendlinge4ad50b2011-08-03 18:37:32 +0000[diff] [blame]3386 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3387 function.</p>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3388
3389<h5>Semantics:</h5>
3390<p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3391 (in-flight) exception whose unwinding was interrupted with
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3392 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3393
3394<h5>Example:</h5>
3395<pre>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3396 resume { i8*, i32 } %exn
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3397</pre>
3398
3399</div>
3400
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3401<!-- _______________________________________________________________________ -->
3402
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3403<h4>
3404 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3405</h4>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3406
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3407<div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3408
3409<h5>Syntax:</h5>
3410<pre>
3411 unreachable
3412</pre>
3413
3414<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3415<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3416 instruction is used to inform the optimizer that a particular portion of the
3417 code is not reachable. This can be used to indicate that the code after a
3418 no-return function cannot be reached, and other facts.</p>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3419
3420<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3421<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3422
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3423</div>
3424
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3425</div>
3426
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3427<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3428<h3>
3429 <a name="binaryops">Binary Operations</a>
3430</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3431
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3432<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3433
3434<p>Binary operators are used to do most of the computation in a program. They
3435 require two operands of the same type, execute an operation on them, and
3436 produce a single value. The operands might represent multiple data, as is
3437 the case with the <a href="#t_vector">vector</a> data type. The result value
3438 has the same type as its operands.</p>
3439
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3440<p>There are several different binary operators:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3441
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3442<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3443<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3444 <a name="i_add">'<tt>add</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3445</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3446
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3447<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3448
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3449<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3450<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3451 &lt;result&gt; = add &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +0000[diff] [blame]3452 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3453 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3454 &lt;result&gt; = add nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3455</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3456
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3457<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3458<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3459
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3460<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3461<p>The two arguments to the '<tt>add</tt>' instruction must
3462 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3463 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3464
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3465<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3466<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3467
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3468<p>If the sum has unsigned overflow, the result returned is the mathematical
3469 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3470
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3471<p>Because LLVM integers use a two's complement representation, this instruction
3472 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3473
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3474<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3475 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3476 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3477 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3478 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3479
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3480<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3481<pre>
3482 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3483</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3484
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3485</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3486
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3487<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3488<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3489 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3490</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3491
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3492<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3493
3494<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3495<pre>
3496 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3497</pre>
3498
3499<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3500<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3501
3502<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3503<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3504 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3505 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3506
3507<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3508<p>The value produced is the floating point sum of the two operands.</p>
3509
3510<h5>Example:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3511<pre>
3512 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3513</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3514
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3515</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3516
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3517<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3518<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3519 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3520</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3521
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3522<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3523
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3524<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3525<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3526 &lt;result&gt; = sub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +0000[diff] [blame]3527 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3528 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3529 &lt;result&gt; = sub nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3530</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3531
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3532<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3533<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3534 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3535
3536<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3537 '<tt>neg</tt>' instruction present in most other intermediate
3538 representations.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3539
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3540<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3541<p>The two arguments to the '<tt>sub</tt>' instruction must
3542 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3543 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3544
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3545<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3546<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3547
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3548<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3549 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3550 result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3551
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3552<p>Because LLVM integers use a two's complement representation, this instruction
3553 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3554
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3555<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3556 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3557 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3558 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3559 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3560
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3561<h5>Example:</h5>
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]3562<pre>
3563 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3564 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3565</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3566
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3567</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3568
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3569<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3570<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3571 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3572</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3573
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3574<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3575
3576<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3577<pre>
3578 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3579</pre>
3580
3581<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3582<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3583 operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3584
3585<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3586 '<tt>fneg</tt>' instruction present in most other intermediate
3587 representations.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3588
3589<h5>Arguments:</h5>
Bill Wendlingd9fe2982009-07-20 02:32:41 +0000[diff] [blame]3590<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3591 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3592 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3593
3594<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3595<p>The value produced is the floating point difference of the two operands.</p>
3596
3597<h5>Example:</h5>
3598<pre>
3599 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3600 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3601</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3602
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3603</div>
3604
3605<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3606<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3607 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3608</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3609
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3610<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3611
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3612<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3613<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3614 &lt;result&gt; = mul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +0000[diff] [blame]3615 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3616 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3617 &lt;result&gt; = mul nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3618</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3619
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3620<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3621<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3622
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3623<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3624<p>The two arguments to the '<tt>mul</tt>' instruction must
3625 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3626 integer values. Both arguments must have identical types.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3627
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3628<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3629<p>The value produced is the integer product of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3630
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3631<p>If the result of the multiplication has unsigned overflow, the result
3632 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3633 width of the result.</p>
3634
3635<p>Because LLVM integers use a two's complement representation, and the result
3636 is the same width as the operands, this instruction returns the correct
3637 result for both signed and unsigned integers. If a full product
3638 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3639 be sign-extended or zero-extended as appropriate to the width of the full
3640 product.</p>
3641
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3642<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3643 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3644 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3645 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3646 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3647
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3648<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3649<pre>
3650 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3651</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3652
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3653</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3654
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3655<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3656<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3657 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3658</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3659
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3660<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3661
3662<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3663<pre>
3664 &lt;result&gt; = fmul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3665</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3666
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3667<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3668<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3669
3670<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3671<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3672 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3673 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3674
3675<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3676<p>The value produced is the floating point product of the two operands.</p>
3677
3678<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3679<pre>
3680 &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3681</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3682
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3683</div>
3684
3685<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3686<h4>
3687 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
3688</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3689
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3690<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3691
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3692<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3693<pre>
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]3694 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3695 &lt;result&gt; = udiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3696</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3697
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3698<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3699<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3700
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3701<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3702<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3703 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3704 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3705
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3706<h5>Semantics:</h5>
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3707<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3708
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3709<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3710 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3711
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3712<p>Division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3713
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]3714<p>If the <tt>exact</tt> keyword is present, the result value of the
3715 <tt>udiv</tt> is a <a href="#trapvalues">trap value</a> if %op1 is not a
3716 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
3717
3718
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3719<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3720<pre>
3721 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3722</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3723
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3724</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3725
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3726<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3727<h4>
3728 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
3729</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3730
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3731<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3732
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3733<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3734<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3735 &lt;result&gt; = sdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanfdfca792009-09-02 17:31:42 +0000[diff] [blame]3736 &lt;result&gt; = sdiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3737</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3738
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3739<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3740<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3741
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3742<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3743<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3744 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3745 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3746
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3747<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3748<p>The value produced is the signed integer quotient of the two operands rounded
3749 towards zero.</p>
3750
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3751<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3752 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3753
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3754<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3755 undefined behavior; this is a rare case, but can occur, for example, by doing
3756 a 32-bit division of -2147483648 by -1.</p>
3757
Dan Gohman9c5beed2009-07-22 00:04:19 +0000[diff] [blame]3758<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohman27ef9972010-04-23 15:23:32 +0000[diff] [blame]3759 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
Dan Gohman38da9272010-07-11 00:08:34 +0000[diff] [blame]3760 be rounded.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3761
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3762<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3763<pre>
3764 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3765</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3766
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3767</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3768
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3769<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3770<h4>
3771 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
3772</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3773
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3774<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3775
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3776<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3777<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]3778 &lt;result&gt; = fdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3779</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3780
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3781<h5>Overview:</h5>
3782<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3783
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3784<h5>Arguments:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3785<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3786 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3787 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3788
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3789<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3790<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3791
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3792<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3793<pre>
3794 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3795</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3796
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3797</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3798
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3799<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3800<h4>
3801 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3802</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3803
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3804<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3805
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3806<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3807<pre>
3808 &lt;result&gt; = urem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3809</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3810
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3811<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3812<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3813 division of its two arguments.</p>
3814
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3815<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3816<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3817 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3818 values. Both arguments must have identical types.</p>
3819
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3820<h5>Semantics:</h5>
3821<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3822 This instruction always performs an unsigned division to get the
3823 remainder.</p>
3824
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3825<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3826 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3827
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3828<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3829
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3830<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3831<pre>
3832 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3833</pre>
3834
3835</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3836
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3837<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3838<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3839 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3840</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3841
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3842<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3843
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3844<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3845<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]3846 &lt;result&gt; = srem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3847</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3848
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3849<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3850<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3851 division of its two operands. This instruction can also take
3852 <a href="#t_vector">vector</a> versions of the values in which case the
3853 elements must be integers.</p>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +0000[diff] [blame]3854
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3855<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3856<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3857 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3858 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3859
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3860<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3861<p>This instruction returns the <i>remainder</i> of a division (where the result
Duncan Sandsdea3a5e2011-03-07 09:12:24 +0000[diff] [blame]3862 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
3863 <i>modulo</i> operator (where the result is either zero or has the same sign
3864 as the divisor, <tt>op2</tt>) of a value.
3865 For more information about the difference,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3866 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3867 Math Forum</a>. For a table of how this is implemented in various languages,
3868 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3869 Wikipedia: modulo operation</a>.</p>
3870
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3871<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3872 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3873
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3874<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3875 Overflow also leads to undefined behavior; this is a rare case, but can
3876 occur, for example, by taking the remainder of a 32-bit division of
3877 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3878 lets srem be implemented using instructions that return both the result of
3879 the division and the remainder.)</p>
3880
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3881<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3882<pre>
3883 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3884</pre>
3885
3886</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3887
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3888<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3889<h4>
3890 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
3891</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3892
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3893<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3894
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3895<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3896<pre>
3897 &lt;result&gt; = frem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3898</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3899
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3900<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3901<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3902 its two operands.</p>
3903
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3904<h5>Arguments:</h5>
3905<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3906 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3907 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3908
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3909<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3910<p>This instruction returns the <i>remainder</i> of a division. The remainder
3911 has the same sign as the dividend.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3912
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3913<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3914<pre>
3915 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3916</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3917
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3918</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]3919
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3920</div>
3921
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]3922<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3923<h3>
3924 <a name="bitwiseops">Bitwise Binary Operations</a>
3925</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3926
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3927<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3928
3929<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3930 program. They are generally very efficient instructions and can commonly be
3931 strength reduced from other instructions. They require two operands of the
3932 same type, execute an operation on them, and produce a single value. The
3933 resulting value is the same type as its operands.</p>
3934
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3935<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3936<h4>
3937 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
3938</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3939
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3940<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3941
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3942<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3943<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]3944 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3945 &lt;result&gt; = shl nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3946 &lt;result&gt; = shl nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3947 &lt;result&gt; = shl nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3948</pre>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]3949
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3950<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3951<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3952 a specified number of bits.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]3953
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3954<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3955<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3956 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3957 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3958
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3959<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3960<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3961 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3962 is (statically or dynamically) negative or equal to or larger than the number
3963 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3964 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3965 shift amount in <tt>op2</tt>.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]3966
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]3967<p>If the <tt>nuw</tt> keyword is present, then the shift produces a
3968 <a href="#trapvalues">trap value</a> if it shifts out any non-zero bits. If
Chris Lattner66298c12011-02-09 16:44:44 +0000[diff] [blame]3969 the <tt>nsw</tt> keyword is present, then the shift produces a
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]3970 <a href="#trapvalues">trap value</a> if it shifts out any bits that disagree
3971 with the resultant sign bit. As such, NUW/NSW have the same semantics as
3972 they would if the shift were expressed as a mul instruction with the same
3973 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
3974
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3975<h5>Example:</h5>
3976<pre>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3977 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3978 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3979 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]3980 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]3981 &lt;result&gt; = shl &lt;2 x i32&gt; &lt; i32 1, i32 1&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 2, i32 4&gt;</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3982</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3983
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3984</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3985
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3986<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3987<h4>
3988 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
3989</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3990
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3991<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3992
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3993<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3994<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]3995 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3996 &lt;result&gt; = lshr exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]3997</pre>
3998
3999<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4000<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4001 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4002
4003<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4004<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4005 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4006 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4007
4008<h5>Semantics:</h5>
4009<p>This instruction always performs a logical shift right operation. The most
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4010 significant bits of the result will be filled with zero bits after the shift.
4011 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4012 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4013 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4014 shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4015
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4016<p>If the <tt>exact</tt> keyword is present, the result value of the
4017 <tt>lshr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
4018 shifted out are non-zero.</p>
4019
4020
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4021<h5>Example:</h5>
4022<pre>
4023 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4024 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4025 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4026 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4027 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4028 &lt;result&gt; = lshr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0x7FFFFFFF, i32 1&gt;</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4029</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4030
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4031</div>
4032
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]4033<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4034<h4>
4035 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4036</h4>
4037
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4038<div>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4039
4040<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4041<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4042 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4043 &lt;result&gt; = ashr exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4044</pre>
4045
4046<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4047<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4048 operand shifted to the right a specified number of bits with sign
4049 extension.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4050
4051<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4052<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4053 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4054 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4055
4056<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4057<p>This instruction always performs an arithmetic shift right operation, The
4058 most significant bits of the result will be filled with the sign bit
4059 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4060 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4061 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4062 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4063
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4064<p>If the <tt>exact</tt> keyword is present, the result value of the
4065 <tt>ashr</tt> is a <a href="#trapvalues">trap value</a> if any of the bits
4066 shifted out are non-zero.</p>
4067
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4068<h5>Example:</h5>
4069<pre>
4070 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4071 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4072 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4073 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4074 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4075 &lt;result&gt; = ashr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 3&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 -1, i32 0&gt;</i>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4076</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4077
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4078</div>
4079
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4080<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4081<h4>
4082 <a name="i_and">'<tt>and</tt>' Instruction</a>
4083</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4084
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4085<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4086
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4087<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4088<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]4089 &lt;result&gt; = and &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4090</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4091
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4092<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4093<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4094 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4095
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4096<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4097<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4098 <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 Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4100
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4101<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4102<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4103
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4104<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4105 <tbody>
4106 <tr>
4107 <td>In0</td>
4108 <td>In1</td>
4109 <td>Out</td>
4110 </tr>
4111 <tr>
4112 <td>0</td>
4113 <td>0</td>
4114 <td>0</td>
4115 </tr>
4116 <tr>
4117 <td>0</td>
4118 <td>1</td>
4119 <td>0</td>
4120 </tr>
4121 <tr>
4122 <td>1</td>
4123 <td>0</td>
4124 <td>0</td>
4125 </tr>
4126 <tr>
4127 <td>1</td>
4128 <td>1</td>
4129 <td>1</td>
4130 </tr>
4131 </tbody>
4132</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4133
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4134<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4135<pre>
4136 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4137 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4138 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4139</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4140</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4141<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4142<h4>
4143 <a name="i_or">'<tt>or</tt>' Instruction</a>
4144</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4145
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4146<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4147
4148<h5>Syntax:</h5>
4149<pre>
4150 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4151</pre>
4152
4153<h5>Overview:</h5>
4154<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4155 two operands.</p>
4156
4157<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4158<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4159 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4160 values. Both arguments must have identical types.</p>
4161
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4162<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4163<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4164
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4165<table border="1" cellspacing="0" cellpadding="4">
4166 <tbody>
4167 <tr>
4168 <td>In0</td>
4169 <td>In1</td>
4170 <td>Out</td>
4171 </tr>
4172 <tr>
4173 <td>0</td>
4174 <td>0</td>
4175 <td>0</td>
4176 </tr>
4177 <tr>
4178 <td>0</td>
4179 <td>1</td>
4180 <td>1</td>
4181 </tr>
4182 <tr>
4183 <td>1</td>
4184 <td>0</td>
4185 <td>1</td>
4186 </tr>
4187 <tr>
4188 <td>1</td>
4189 <td>1</td>
4190 <td>1</td>
4191 </tr>
4192 </tbody>
4193</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4194
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4195<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4196<pre>
4197 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4198 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4199 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4200</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4201
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4202</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4203
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4204<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4205<h4>
4206 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4207</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4208
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4209<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4210
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4211<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4212<pre>
4213 &lt;result&gt; = xor &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4214</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4215
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4216<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4217<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4218 its two operands. The <tt>xor</tt> is used to implement the "one's
4219 complement" operation, which is the "~" operator in C.</p>
4220
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4221<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4222<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4223 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4224 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4225
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4226<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4227<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4228
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4229<table border="1" cellspacing="0" cellpadding="4">
4230 <tbody>
4231 <tr>
4232 <td>In0</td>
4233 <td>In1</td>
4234 <td>Out</td>
4235 </tr>
4236 <tr>
4237 <td>0</td>
4238 <td>0</td>
4239 <td>0</td>
4240 </tr>
4241 <tr>
4242 <td>0</td>
4243 <td>1</td>
4244 <td>1</td>
4245 </tr>
4246 <tr>
4247 <td>1</td>
4248 <td>0</td>
4249 <td>1</td>
4250 </tr>
4251 <tr>
4252 <td>1</td>
4253 <td>1</td>
4254 <td>0</td>
4255 </tr>
4256 </tbody>
4257</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4258
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4259<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4260<pre>
4261 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4262 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4263 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4264 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4265</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4266
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4267</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4268
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4269</div>
4270
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4271<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4272<h3>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4273 <a name="vectorops">Vector Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4274</h3>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4275
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4276<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4277
4278<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4279 target-independent manner. These instructions cover the element-access and
4280 vector-specific operations needed to process vectors effectively. While LLVM
4281 does directly support these vector operations, many sophisticated algorithms
4282 will want to use target-specific intrinsics to take full advantage of a
4283 specific target.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4284
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4285<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4286<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4287 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4288</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4289
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4290<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4291
4292<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4293<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4294 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4295</pre>
4296
4297<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4298<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4299 from a vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4300
4301
4302<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4303<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4304 of <a href="#t_vector">vector</a> type. The second operand is an index
4305 indicating the position from which to extract the element. The index may be
4306 a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4307
4308<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4309<p>The result is a scalar of the same type as the element type of
4310 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4311 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4312 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4313
4314<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4315<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4316 &lt;result&gt; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4317</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4318
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4319</div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4320
4321<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4322<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4323 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4324</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4325
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4326<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4327
4328<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4329<pre>
Dan Gohmanf3480b92008-05-12 23:38:42 +0000[diff] [blame]4330 &lt;result&gt; = insertelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, i32 &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4331</pre>
4332
4333<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4334<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4335 vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4336
4337<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4338<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4339 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4340 whose type must equal the element type of the first operand. The third
4341 operand is an index indicating the position at which to insert the value.
4342 The index may be a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4343
4344<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4345<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4346 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4347 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4348 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4349
4350<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4351<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4352 &lt;result&gt; = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4353</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4354
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4355</div>
4356
4357<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4358<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4359 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4360</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4361
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4362<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4363
4364<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4365<pre>
Mon P Wangaeb06d22008-11-10 04:46:22 +0000[diff] [blame]4366 &lt;result&gt; = shufflevector &lt;n x &lt;ty&gt;&gt; &lt;v1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;v2&gt;, &lt;m x i32&gt; &lt;mask&gt; <i>; yields &lt;m x &lt;ty&gt;&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4367</pre>
4368
4369<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4370<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4371 from two input vectors, returning a vector with the same element type as the
4372 input and length that is the same as the shuffle mask.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4373
4374<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4375<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4376 with types that match each other. The third argument is a shuffle mask whose
4377 element type is always 'i32'. The result of the instruction is a vector
4378 whose length is the same as the shuffle mask and whose element type is the
4379 same as the element type of the first two operands.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4380
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4381<p>The shuffle mask operand is required to be a constant vector with either
4382 constant integer or undef values.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4383
4384<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4385<p>The elements of the two input vectors are numbered from left to right across
4386 both of the vectors. The shuffle mask operand specifies, for each element of
4387 the result vector, which element of the two input vectors the result element
4388 gets. The element selector may be undef (meaning "don't care") and the
4389 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4390
4391<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4392<pre>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4393 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]4394 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4395 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4396 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4397 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wangaeb06d22008-11-10 04:46:22 +0000[diff] [blame]4398 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4399 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wangaeb06d22008-11-10 04:46:22 +0000[diff] [blame]4400 &lt;8 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 &gt; <i>; yields &lt;8 x i32&gt;</i>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4401</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4402
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4403</div>
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]4404
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4405</div>
4406
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4407<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4408<h3>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4409 <a name="aggregateops">Aggregate Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4410</h3>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4411
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4412<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4413
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4414<p>LLVM supports several instructions for working with
4415 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4416
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4417<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4418<h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4419 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4420</h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4421
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4422<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4423
4424<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4425<pre>
4426 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4427</pre>
4428
4429<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4430<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4431 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4432
4433<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4434<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4435 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4436 <a href="#t_array">array</a> type. The operands are constant indices to
4437 specify which value to extract in a similar manner as indices in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4438 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Frits van Bommel13242892010-12-05 20:54:38 +0000[diff] [blame]4439 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4440 <ul>
4441 <li>Since the value being indexed is not a pointer, the first index is
4442 omitted and assumed to be zero.</li>
4443 <li>At least one index must be specified.</li>
4444 <li>Not only struct indices but also array indices must be in
4445 bounds.</li>
4446 </ul>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4447
4448<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4449<p>The result is the value at the position in the aggregate specified by the
4450 index operands.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4451
4452<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4453<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4454 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4455</pre>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4456
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4457</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4458
4459<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4460<h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4461 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4462</h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4463
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4464<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4465
4466<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4467<pre>
Bill Wendling194229e2011-07-26 20:42:28 +0000[diff] [blame]4468 &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 Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4469</pre>
4470
4471<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4472<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4473 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4474
4475<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4476<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4477 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4478 <a href="#t_array">array</a> type. The second operand is a first-class
4479 value to insert. The following operands are constant indices indicating
4480 the position at which to insert the value in a similar manner as indices in a
Frits van Bommel13242892010-12-05 20:54:38 +0000[diff] [blame]4481 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4482 value to insert must have the same type as the value identified by the
4483 indices.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4484
4485<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4486<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4487 that of <tt>val</tt> except that the value at the position specified by the
4488 indices is that of <tt>elt</tt>.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4489
4490<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4491<pre>
Chris Lattner8645d1a2011-05-22 07:18:08 +0000[diff] [blame]4492 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4493 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4494 %agg3 = insertvalue {i32, {float}} %agg1, float %val, 1, 0 <i>; yields {i32 1, float %val}</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4495</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4496
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4497</div>
4498
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4499</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4500
4501<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4502<h3>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]4503 <a name="memoryops">Memory Access and Addressing Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4504</h3>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4505
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4506<div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4507
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4508<p>A key design point of an SSA-based representation is how it represents
4509 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]4510 very simple. This section describes how to read, write, and allocate
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4511 memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4512
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4513<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4514<h4>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4515 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4516</h4>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4517
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4518<div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4519
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4520<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4521<pre>
Dan Gohmanf75a7d32010-05-28 01:14:11 +0000[diff] [blame]4522 &lt;result&gt; = alloca &lt;type&gt;[, &lt;ty&gt; &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4523</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4524
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4525<h5>Overview:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]4526<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4527 currently executing function, to be automatically released when this function
4528 returns to its caller. The object is always allocated in the generic address
4529 space (address space zero).</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4530
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4531<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4532<p>The '<tt>alloca</tt>' instruction
4533 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4534 runtime stack, returning a pointer of the appropriate type to the program.
4535 If "NumElements" is specified, it is the number of elements allocated,
4536 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4537 specified, the value result of the allocation is guaranteed to be aligned to
4538 at least that boundary. If not specified, or if zero, the target can choose
4539 to align the allocation on any convenient boundary compatible with the
4540 type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4541
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4542<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4543
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4544<h5>Semantics:</h5>
Bill Wendling871eb0a2009-05-08 20:49:29 +0000[diff] [blame]4545<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4546 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4547 memory is automatically released when the function returns. The
4548 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4549 variables that must have an address available. When the function returns
4550 (either with the <tt><a href="#i_ret">ret</a></tt>
4551 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4552 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4553
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4554<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4555<pre>
Dan Gohman81e21672009-01-04 23:49:44 +0000[diff] [blame]4556 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4557 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4558 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4559 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4560</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4561
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4562</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4563
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4564<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4565<h4>
4566 <a name="i_load">'<tt>load</tt>' Instruction</a>
4567</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4568
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4569<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4570
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4571<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4572<pre>
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]4573 &lt;result&gt; = [volatile] load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4574 &lt;result&gt; = atomic [volatile] load &lt;ty&gt;* &lt;pointer&gt; [singlethread] &lt;ordering&gt;, align &lt;alignment&gt;
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4575 !&lt;index&gt; = !{ i32 1 }
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4576</pre>
4577
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4578<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4579<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4580
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4581<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4582<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4583 from which to load. The pointer must point to
4584 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4585 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]4586 number or order of execution of this <tt>load</tt> with other <a
4587 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4588
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]4589<p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
4590 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4591 argument. The <code>release</code> and <code>acq_rel</code> orderings are
4592 not valid on <code>load</code> instructions. Atomic loads produce <a
4593 href="#memorymodel">defined</a> results when they may see multiple atomic
4594 stores. The type of the pointee must be an integer type whose bit width
4595 is a power of two greater than or equal to eight and less than or equal
4596 to a target-specific size limit. <code>align</code> must be explicitly
4597 specified on atomic loads, and the load has undefined behavior if the
4598 alignment is not set to a value which is at least the size in bytes of
4599 the pointee. <code>!nontemporal</code> does not have any defined semantics
4600 for atomic loads.</p>
4601
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4602<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4603 operation (that is, the alignment of the memory address). A value of 0 or an
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4604 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4605 alignment for the target. It is the responsibility of the code emitter to
4606 ensure that the alignment information is correct. Overestimating the
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4607 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4608 produce less efficient code. An alignment of 1 is always safe.</p>
4609
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4610<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4611 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4612 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4613 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4614 and code generator that this load is not expected to be reused in the cache.
4615 The code generator may select special instructions to save cache bandwidth,
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4616 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4617
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4618<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4619<p>The location of memory pointed to is loaded. If the value being loaded is of
4620 scalar type then the number of bytes read does not exceed the minimum number
4621 of bytes needed to hold all bits of the type. For example, loading an
4622 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4623 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4624 is undefined if the value was not originally written using a store of the
4625 same type.</p>
4626
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4627<h5>Examples:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4628<pre>
4629 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4630 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4631 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4632</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4633
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4634</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4635
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4636<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4637<h4>
4638 <a name="i_store">'<tt>store</tt>' Instruction</a>
4639</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4640
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4641<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4642
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4643<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4644<pre>
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]4645 [volatile] store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;] <i>; yields {void}</i>
4646 atomic [volatile] store &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 Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4647</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4648
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4649<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4650<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4651
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4652<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4653<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4654 and an address at which to store it. The type of the
4655 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4656 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]4657 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
4658 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4659 order of execution of this <tt>store</tt> with other <a
4660 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4661
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]4662<p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
4663 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4664 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
4665 valid on <code>store</code> instructions. Atomic loads produce <a
4666 href="#memorymodel">defined</a> results when they may see multiple atomic
4667 stores. The type of the pointee must be an integer type whose bit width
4668 is a power of two greater than or equal to eight and less than or equal
4669 to a target-specific size limit. <code>align</code> must be explicitly
4670 specified on atomic stores, and the store has undefined behavior if the
4671 alignment is not set to a value which is at least the size in bytes of
4672 the pointee. <code>!nontemporal</code> does not have any defined semantics
4673 for atomic stores.</p>
4674
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4675<p>The optional constant "align" argument specifies the alignment of the
4676 operation (that is, the alignment of the memory address). A value of 0 or an
4677 omitted "align" argument means that the operation has the preferential
4678 alignment for the target. It is the responsibility of the code emitter to
4679 ensure that the alignment information is correct. Overestimating the
4680 alignment results in an undefined behavior. Underestimating the alignment may
4681 produce less efficient code. An alignment of 1 is always safe.</p>
4682
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4683<p>The optional !nontemporal metadata must reference a single metatadata
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]4684 name &lt;index&gt; corresponding to a metadata node with one i32 entry of
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4685 value 1. The existence of the !nontemporal metatadata on the
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4686 instruction tells the optimizer and code generator that this load is
4687 not expected to be reused in the cache. The code generator may
4688 select special instructions to save cache bandwidth, such as the
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4689 MOVNT instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4690
4691
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4692<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4693<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4694 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4695 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4696 does not exceed the minimum number of bytes needed to hold all bits of the
4697 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4698 writing a value of a type like <tt>i20</tt> with a size that is not an
4699 integral number of bytes, it is unspecified what happens to the extra bits
4700 that do not belong to the type, but they will typically be overwritten.</p>
4701
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4702<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4703<pre>
4704 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8c6c72d2007-10-22 05:10:05 +0000[diff] [blame]4705 store i32 3, i32* %ptr <i>; yields {void}</i>
4706 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4707</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4708
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]4709</div>
4710
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4711<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]4712<h4>
4713<a name="i_fence">'<tt>fence</tt>' Instruction</a>
4714</h4>
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]4715
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]4716<div>
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]4717
4718<h5>Syntax:</h5>
4719<pre>
4720 fence [singlethread] &lt;ordering&gt; <i>; yields {void}</i>
4721</pre>
4722
4723<h5>Overview:</h5>
4724<p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
4725between operations.</p>
4726
4727<h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
4728href="#ordering">ordering</a> argument which defines what
4729<i>synchronizes-with</i> edges they add. They can only be given
4730<code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
4731<code>seq_cst</code> orderings.</p>
4732
4733<h5>Semantics:</h5>
4734<p>A fence <var>A</var> which has (at least) <code>release</code> ordering
4735semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
4736<code>acquire</code> ordering semantics if and only if there exist atomic
4737operations <var>X</var> and <var>Y</var>, both operating on some atomic object
4738<var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
4739<var>X</var> modifies <var>M</var> (either directly or through some side effect
4740of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
4741<var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
4742<i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
4743than an explicit <code>fence</code>, one (but not both) of the atomic operations
4744<var>X</var> or <var>Y</var> might provide a <code>release</code> or
4745<code>acquire</code> (resp.) ordering constraint and still
4746<i>synchronize-with</i> the explicit <code>fence</code> and establish the
4747<i>happens-before</i> edge.</p>
4748
4749<p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
4750having both <code>acquire</code> and <code>release</code> semantics specified
4751above, participates in the global program order of other <code>seq_cst</code>
4752operations and/or fences.</p>
4753
4754<p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
4755specifies that the fence only synchronizes with other fences in the same
4756thread. (This is useful for interacting with signal handlers.)</p>
4757
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]4758<h5>Example:</h5>
4759<pre>
4760 fence acquire <i>; yields {void}</i>
4761 fence singlethread seq_cst <i>; yields {void}</i>
4762</pre>
4763
4764</div>
4765
4766<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]4767<h4>
4768<a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
4769</h4>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4770
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]4771<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4772
4773<h5>Syntax:</h5>
4774<pre>
4775 [volatile] cmpxchg &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>
4776</pre>
4777
4778<h5>Overview:</h5>
4779<p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
4780It loads a value in memory and compares it to a given value. If they are
4781equal, it stores a new value into the memory.</p>
4782
4783<h5>Arguments:</h5>
4784<p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
4785address to operate on, a value to compare to the value currently be at that
4786address, and a new value to place at that address if the compared values are
4787equal. The type of '<var>&lt;cmp&gt;</var>' must be an integer type whose
4788bit width is a power of two greater than or equal to eight and less than
4789or equal to a target-specific size limit. '<var>&lt;cmp&gt;</var>' and
4790'<var>&lt;new&gt;</var>' must have the same type, and the type of
4791'<var>&lt;pointer&gt;</var>' must be a pointer to that type. If the
4792<code>cmpxchg</code> is marked as <code>volatile</code>, then the
4793optimizer is not allowed to modify the number or order of execution
4794of this <code>cmpxchg</code> with other <a href="#volatile">volatile
4795operations</a>.</p>
4796
4797<!-- FIXME: Extend allowed types. -->
4798
4799<p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
4800<code>cmpxchg</code> synchronizes with other atomic operations.</p>
4801
4802<p>The optional "<code>singlethread</code>" argument declares that the
4803<code>cmpxchg</code> is only atomic with respect to code (usually signal
4804handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
4805cmpxchg is atomic with respect to all other code in the system.</p>
4806
4807<p>The pointer passed into cmpxchg must have alignment greater than or equal to
4808the size in memory of the operand.
4809
4810<h5>Semantics:</h5>
4811<p>The contents of memory at the location specified by the
4812'<tt>&lt;pointer&gt;</tt>' operand is read and compared to
4813'<tt>&lt;cmp&gt;</tt>'; if the read value is the equal,
4814'<tt>&lt;new&gt;</tt>' is written. The original value at the location
4815is returned.
4816
4817<p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
4818purpose of identifying <a href="#release_sequence">release sequences</a>. A
4819failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
4820parameter determined by dropping any <code>release</code> part of the
4821<code>cmpxchg</code>'s ordering.</p>
4822
4823<!--
4824FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
4825optimization work on ARM.)
4826
4827FIXME: Is a weaker ordering constraint on failure helpful in practice?
4828-->
4829
4830<h5>Example:</h5>
4831<pre>
4832entry:
4833 %orig = atomic <a href="#i_load">load</a> i32* %ptr unordered <i>; yields {i32}</i>
4834 <a href="#i_br">br</a> label %loop
4835
4836loop:
4837 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
4838 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
4839 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
4840 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
4841 <a href="#i_br">br</a> i1 %success, label %done, label %loop
4842
4843done:
4844 ...
4845</pre>
4846
4847</div>
4848
4849<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]4850<h4>
4851<a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
4852</h4>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4853
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame^]4854<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4855
4856<h5>Syntax:</h5>
4857<pre>
4858 [volatile] atomicrmw &lt;operation&gt; &lt;ty&gt;* &lt;pointer&gt;, &lt;ty&gt; &lt;value&gt; [singlethread] &lt;ordering&gt; <i>; yields {ty}</i>
4859</pre>
4860
4861<h5>Overview:</h5>
4862<p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
4863
4864<h5>Arguments:</h5>
4865<p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
4866operation to apply, an address whose value to modify, an argument to the
4867operation. The operation must be one of the following keywords:</p>
4868<ul>
4869 <li>xchg</li>
4870 <li>add</li>
4871 <li>sub</li>
4872 <li>and</li>
4873 <li>nand</li>
4874 <li>or</li>
4875 <li>xor</li>
4876 <li>max</li>
4877 <li>min</li>
4878 <li>umax</li>
4879 <li>umin</li>
4880</ul>
4881
4882<p>The type of '<var>&lt;value&gt;</var>' must be an integer type whose
4883bit width is a power of two greater than or equal to eight and less than
4884or equal to a target-specific size limit. The type of the
4885'<code>&lt;pointer&gt;</code>' operand must be a pointer to that type.
4886If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
4887optimizer is not allowed to modify the number or order of execution of this
4888<code>atomicrmw</code> with other <a href="#volatile">volatile
4889 operations</a>.</p>
4890
4891<!-- FIXME: Extend allowed types. -->
4892
4893<h5>Semantics:</h5>
4894<p>The contents of memory at the location specified by the
4895'<tt>&lt;pointer&gt;</tt>' operand are atomically read, modified, and written
4896back. The original value at the location is returned. The modification is
4897specified by the <var>operation</var> argument:</p>
4898
4899<ul>
4900 <li>xchg: <code>*ptr = val</code></li>
4901 <li>add: <code>*ptr = *ptr + val</code></li>
4902 <li>sub: <code>*ptr = *ptr - val</code></li>
4903 <li>and: <code>*ptr = *ptr &amp; val</code></li>
4904 <li>nand: <code>*ptr = ~(*ptr &amp; val)</code></li>
4905 <li>or: <code>*ptr = *ptr | val</code></li>
4906 <li>xor: <code>*ptr = *ptr ^ val</code></li>
4907 <li>max: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using a signed comparison)</li>
4908 <li>min: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using a signed comparison)</li>
4909 <li>umax: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using an unsigned comparison)</li>
4910 <li>umin: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using an unsigned comparison)</li>
4911</ul>
4912
4913<h5>Example:</h5>
4914<pre>
4915 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
4916</pre>
4917
4918</div>
4919
4920<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4921<h4>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4922 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4923</h4>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4924
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4925<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4926
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]4927<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4928<pre>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]4929 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]4930 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4931</pre>
4932
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]4933<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4934<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4935 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4936 It performs address calculation only and does not access memory.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4937
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]4938<h5>Arguments:</h5>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]4939<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnerc8eef442009-07-29 06:44:13 +0000[diff] [blame]4940 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4941 elements of the aggregate object are indexed. The interpretation of each
4942 index is dependent on the type being indexed into. The first index always
4943 indexes the pointer value given as the first argument, the second index
4944 indexes a value of the type pointed to (not necessarily the value directly
4945 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4946 indexed into must be a pointer value, subsequent types can be arrays,
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4947 vectors, and structs. Note that subsequent types being indexed into
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4948 can never be pointers, since that would require loading the pointer before
4949 continuing calculation.</p>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]4950
4951<p>The type of each index argument depends on the type it is indexing into.
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4952 When indexing into a (optionally packed) structure, only <tt>i32</tt>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4953 integer <b>constants</b> are allowed. When indexing into an array, pointer
4954 or vector, integers of any width are allowed, and they are not required to be
Chris Lattnerc8eef442009-07-29 06:44:13 +0000[diff] [blame]4955 constant.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4956
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4957<p>For example, let's consider a C code fragment and how it gets compiled to
4958 LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4959
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]4960<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]4961struct RT {
4962 char A;
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]4963 int B[10][20];
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]4964 char C;
4965};
4966struct ST {
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]4967 int X;
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]4968 double Y;
4969 struct RT Z;
4970};
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4971
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]4972int *foo(struct ST *s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]4973 return &amp;s[1].Z.B[5][13];
4974}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4975</pre>
4976
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4977<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4978
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]4979<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]4980%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4981%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4982
Dan Gohman4df605b2009-07-25 02:23:48 +0000[diff] [blame]4983define i32* @foo(%ST* %s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]4984entry:
4985 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4986 ret i32* %reg
4987}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]4988</pre>
4989
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]4990<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4991<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4992 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4993 }</tt>' type, a structure. The second index indexes into the third element
4994 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4995 i8 }</tt>' type, another structure. The third index indexes into the second
4996 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4997 array. The two dimensions of the array are subscripted into, yielding an
4998 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4999 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5000
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5001<p>Note that it is perfectly legal to index partially through a structure,
5002 returning a pointer to an inner element. Because of this, the LLVM code for
5003 the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5004
5005<pre>
Dan Gohman4df605b2009-07-25 02:23:48 +0000[diff] [blame]5006 define i32* @foo(%ST* %s) {
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5007 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]5008 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
5009 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5010 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5011 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5012 ret i32* %t5
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5013 }
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5014</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]5015
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5016<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman27ef9972010-04-23 15:23:32 +0000[diff] [blame]5017 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
5018 base pointer is not an <i>in bounds</i> address of an allocated object,
5019 or if any of the addresses that would be formed by successive addition of
5020 the offsets implied by the indices to the base address with infinitely
5021 precise arithmetic are not an <i>in bounds</i> address of that allocated
5022 object. The <i>in bounds</i> addresses for an allocated object are all
5023 the addresses that point into the object, plus the address one byte past
5024 the end.</p>
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5025
5026<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
5027 the base address with silently-wrapping two's complement arithmetic, and
5028 the result value of the <tt>getelementptr</tt> may be outside the object
5029 pointed to by the base pointer. The result value may not necessarily be
5030 used to access memory though, even if it happens to point into allocated
5031 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
5032 section for more information.</p>
5033
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5034<p>The getelementptr instruction is often confusing. For some more insight into
5035 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]5036
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5037<h5>Example:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5038<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5039 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5040 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5041 <i>; yields i8*:vptr</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5042 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5043 <i>; yields i8*:eptr</i>
5044 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta9f805c22009-04-25 07:27:44 +0000[diff] [blame]5045 <i>; yields i32*:iptr</i>
Sanjiv Gupta16ffa802009-04-24 16:38:13 +0000[diff] [blame]5046 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5047</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5048
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5049</div>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]5050
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5051</div>
5052
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5053<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5054<h3>
5055 <a name="convertops">Conversion Operations</a>
5056</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5057
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5058<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5059
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5060<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5061 which all take a single operand and a type. They perform various bit
5062 conversions on the operand.</p>
5063
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5064<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5065<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5066 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5067</h4>
5068
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5069<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5070
5071<h5>Syntax:</h5>
5072<pre>
5073 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5074</pre>
5075
5076<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5077<p>The '<tt>trunc</tt>' instruction truncates its operand to the
5078 type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5079
5080<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5081<p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5082 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5083 of the same number of integers.
5084 The bit size of the <tt>value</tt> must be larger than
5085 the bit size of the destination type, <tt>ty2</tt>.
5086 Equal sized types are not allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5087
5088<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5089<p>The '<tt>trunc</tt>' instruction truncates the high order bits
5090 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5091 source size must be larger than the destination size, <tt>trunc</tt> cannot
5092 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5093
5094<h5>Example:</h5>
5095<pre>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5096 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5097 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5098 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5099 %W = trunc &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i8&gt; <i>; yields &lt;i8 8, i8 7&gt;</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5100</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5101
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5102</div>
5103
5104<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5105<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5106 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5107</h4>
5108
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5109<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5110
5111<h5>Syntax:</h5>
5112<pre>
5113 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5114</pre>
5115
5116<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5117<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5118 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5119
5120
5121<h5>Arguments:</h5>
Nadav Rotemed9b9342011-02-20 12:37:50 +0000[diff] [blame]5122<p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5123 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5124 of the same number of integers.
5125 The bit size of the <tt>value</tt> must be smaller than
5126 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5127 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5128
5129<h5>Semantics:</h5>
5130<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5131 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5132
Reid Spencerb5929522007-01-12 15:46:11 +0000[diff] [blame]5133<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5134
5135<h5>Example:</h5>
5136<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5137 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5138 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Nadav Rotemed9b9342011-02-20 12:37:50 +0000[diff] [blame]5139 %Z = zext &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i32&gt; <i>; yields &lt;i32 8, i32 7&gt;</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5140</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5141
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5142</div>
5143
5144<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5145<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5146 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5147</h4>
5148
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5149<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5150
5151<h5>Syntax:</h5>
5152<pre>
5153 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5154</pre>
5155
5156<h5>Overview:</h5>
5157<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5158
5159<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5160<p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5161 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5162 of the same number of integers.
5163 The bit size of the <tt>value</tt> must be smaller than
5164 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5165 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5166
5167<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5168<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5169 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5170 of the type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5171
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5172<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5173
5174<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5175<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5176 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5177 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5178 %Z = sext &lt;2 x i16&gt; &lt;i16 8, i16 7&gt; to &lt;2 x i32&gt; <i>; yields &lt;i32 8, i32 7&gt;</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5179</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5180
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5181</div>
5182
5183<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5184<h4>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5185 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5186</h4>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5187
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5188<div>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5189
5190<h5>Syntax:</h5>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5191<pre>
5192 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5193</pre>
5194
5195<h5>Overview:</h5>
5196<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5197 <tt>ty2</tt>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5198
5199<h5>Arguments:</h5>
5200<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5201 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5202 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5203 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5204 <i>no-op cast</i>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5205
5206<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5207<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5208 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5209 <a href="#t_floating">floating point</a> type. If the value cannot fit
5210 within the destination type, <tt>ty2</tt>, then the results are
5211 undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5212
5213<h5>Example:</h5>
5214<pre>
5215 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5216 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5217</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5218
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5219</div>
5220
5221<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5222<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5223 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5224</h4>
5225
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5226<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5227
5228<h5>Syntax:</h5>
5229<pre>
5230 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5231</pre>
5232
5233<h5>Overview:</h5>
5234<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5235 floating point value.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5236
5237<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5238<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5239 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5240 a <a href="#t_floating">floating point</a> type to cast it to. The source
5241 type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5242
5243<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5244<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5245 <a href="#t_floating">floating point</a> type to a larger
5246 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5247 used to make a <i>no-op cast</i> because it always changes bits. Use
5248 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5249
5250<h5>Example:</h5>
5251<pre>
Nick Lewycky5bb3ece2011-03-31 18:20:19 +0000[diff] [blame]5252 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5253 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5254</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5255
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5256</div>
5257
5258<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5259<h4>
Reid Spencer24d6da52007-01-21 00:29:26 +0000[diff] [blame]5260 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5261</h4>
5262
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5263<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5264
5265<h5>Syntax:</h5>
5266<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5267 &lt;result&gt; = fptoui &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5268</pre>
5269
5270<h5>Overview:</h5>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5271<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5272 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5273
5274<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5275<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5276 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5277 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5278 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5279 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5280
5281<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5282<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5283 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5284 towards zero) unsigned integer value. If the value cannot fit
5285 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5286
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5287<h5>Example:</h5>
5288<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5289 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner88519042007-09-22 03:17:52 +0000[diff] [blame]5290 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5291 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5292</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5293
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5294</div>
5295
5296<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5297<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5298 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5299</h4>
5300
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5301<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5302
5303<h5>Syntax:</h5>
5304<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5305 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5306</pre>
5307
5308<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5309<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5310 <a href="#t_floating">floating point</a> <tt>value</tt> to
5311 type <tt>ty2</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5312
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5313<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5314<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5315 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5316 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5317 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5318 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5319
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5320<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5321<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5322 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5323 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5324 the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5325
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]5326<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5327<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5328 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner88519042007-09-22 03:17:52 +0000[diff] [blame]5329 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5330 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5331</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5332
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5333</div>
5334
5335<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5336<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5337 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5338</h4>
5339
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5340<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5341
5342<h5>Syntax:</h5>
5343<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5344 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5345</pre>
5346
5347<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5348<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5349 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5350
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5351<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +0000[diff] [blame]5352<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5353 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5354 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5355 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5356 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5357
5358<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5359<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5360 integer quantity and converts it to the corresponding floating point
5361 value. If the value cannot fit in the floating point value, the results are
5362 undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5363
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5364<h5>Example:</h5>
5365<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5366 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5367 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5368</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5369
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5370</div>
5371
5372<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5373<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5374 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5375</h4>
5376
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5377<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5378
5379<h5>Syntax:</h5>
5380<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5381 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5382</pre>
5383
5384<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5385<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5386 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5387
5388<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +0000[diff] [blame]5389<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5390 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5391 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5392 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5393 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5394
5395<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5396<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5397 quantity and converts it to the corresponding floating point value. If the
5398 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5399
5400<h5>Example:</h5>
5401<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5402 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5403 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5404</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5405
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5406</div>
5407
5408<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5409<h4>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5410 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5411</h4>
5412
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5413<div>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5414
5415<h5>Syntax:</h5>
5416<pre>
5417 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5418</pre>
5419
5420<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5421<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
5422 the integer type <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5423
5424<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5425<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
5426 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
5427 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5428
5429<h5>Semantics:</h5>
5430<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5431 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5432 truncating or zero extending that value to the size of the integer type. If
5433 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5434 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5435 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5436 change.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5437
5438<h5>Example:</h5>
5439<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5440 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
5441 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5442</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5443
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5444</div>
5445
5446<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5447<h4>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5448 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5449</h4>
5450
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5451<div>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5452
5453<h5>Syntax:</h5>
5454<pre>
5455 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5456</pre>
5457
5458<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5459<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5460 pointer type, <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5461
5462<h5>Arguments:</h5>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]5463<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5464 value to cast, and a type to cast it to, which must be a
5465 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5466
5467<h5>Semantics:</h5>
5468<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5469 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5470 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5471 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5472 than the size of a pointer then a zero extension is done. If they are the
5473 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5474
5475<h5>Example:</h5>
5476<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5477 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5478 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5479 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5480</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5481
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5482</div>
5483
5484<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5485<h4>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5486 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5487</h4>
5488
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5489<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5490
5491<h5>Syntax:</h5>
5492<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5493 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5494</pre>
5495
5496<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5497<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5498 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5499
5500<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5501<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5502 non-aggregate first class value, and a type to cast it to, which must also be
5503 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5504 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5505 identical. If the source type is a pointer, the destination type must also be
5506 a pointer. This instruction supports bitwise conversion of vectors to
5507 integers and to vectors of other types (as long as they have the same
5508 size).</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5509
5510<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5511<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5512 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5513 this conversion. The conversion is done as if the <tt>value</tt> had been
5514 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
5515 be converted to other pointer types with this instruction. To convert
5516 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5517 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5518
5519<h5>Example:</h5>
5520<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5521 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5522 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5523 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]5524</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5525
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5526</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5527
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5528</div>
5529
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5530<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5531<h3>
5532 <a name="otherops">Other Operations</a>
5533</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5534
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5535<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5536
5537<p>The instructions in this category are the "miscellaneous" instructions, which
5538 defy better classification.</p>
5539
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5540<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5541<h4>
5542 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5543</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5544
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5545<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5546
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5547<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5548<pre>
5549 &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5550</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5551
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5552<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5553<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5554 boolean values based on comparison of its two integer, integer vector, or
5555 pointer operands.</p>
5556
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5557<h5>Arguments:</h5>
5558<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5559 the condition code indicating the kind of comparison to perform. It is not a
5560 value, just a keyword. The possible condition code are:</p>
5561
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5562<ol>
5563 <li><tt>eq</tt>: equal</li>
5564 <li><tt>ne</tt>: not equal </li>
5565 <li><tt>ugt</tt>: unsigned greater than</li>
5566 <li><tt>uge</tt>: unsigned greater or equal</li>
5567 <li><tt>ult</tt>: unsigned less than</li>
5568 <li><tt>ule</tt>: unsigned less or equal</li>
5569 <li><tt>sgt</tt>: signed greater than</li>
5570 <li><tt>sge</tt>: signed greater or equal</li>
5571 <li><tt>slt</tt>: signed less than</li>
5572 <li><tt>sle</tt>: signed less or equal</li>
5573</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5574
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]5575<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5576 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5577 typed. They must also be identical types.</p>
5578
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5579<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5580<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5581 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]5582 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5583 result, as follows:</p>
5584
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5585<ol>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5586 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5587 <tt>false</tt> otherwise. No sign interpretation is necessary or
5588 performed.</li>
5589
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5590 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5591 <tt>false</tt> otherwise. No sign interpretation is necessary or
5592 performed.</li>
5593
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5594 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5595 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5596
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5597 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5598 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5599 to <tt>op2</tt>.</li>
5600
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5601 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5602 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5603
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5604 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5605 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5606
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5607 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5608 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5609
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5610 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5611 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5612 to <tt>op2</tt>.</li>
5613
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5614 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5615 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5616
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5617 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5618 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5619</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5620
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5621<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5622 values are compared as if they were integers.</p>
5623
5624<p>If the operands are integer vectors, then they are compared element by
5625 element. The result is an <tt>i1</tt> vector with the same number of elements
5626 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5627
5628<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5629<pre>
5630 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5631 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
5632 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
5633 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5634 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5635 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5636</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]5637
5638<p>Note that the code generator does not yet support vector types with
5639 the <tt>icmp</tt> instruction.</p>
5640
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5641</div>
5642
5643<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5644<h4>
5645 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5646</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5647
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5648<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5649
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5650<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5651<pre>
5652 &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5653</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5654
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5655<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5656<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5657 values based on comparison of its operands.</p>
5658
5659<p>If the operands are floating point scalars, then the result type is a boolean
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]5660(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5661
5662<p>If the operands are floating point vectors, then the result type is a vector
5663 of boolean with the same number of elements as the operands being
5664 compared.</p>
5665
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5666<h5>Arguments:</h5>
5667<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5668 the condition code indicating the kind of comparison to perform. It is not a
5669 value, just a keyword. The possible condition code are:</p>
5670
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5671<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5672 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5673 <li><tt>oeq</tt>: ordered and equal</li>
5674 <li><tt>ogt</tt>: ordered and greater than </li>
5675 <li><tt>oge</tt>: ordered and greater than or equal</li>
5676 <li><tt>olt</tt>: ordered and less than </li>
5677 <li><tt>ole</tt>: ordered and less than or equal</li>
5678 <li><tt>one</tt>: ordered and not equal</li>
5679 <li><tt>ord</tt>: ordered (no nans)</li>
5680 <li><tt>ueq</tt>: unordered or equal</li>
5681 <li><tt>ugt</tt>: unordered or greater than </li>
5682 <li><tt>uge</tt>: unordered or greater than or equal</li>
5683 <li><tt>ult</tt>: unordered or less than </li>
5684 <li><tt>ule</tt>: unordered or less than or equal</li>
5685 <li><tt>une</tt>: unordered or not equal</li>
5686 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5687 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5688</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5689
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5690<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5691 <i>unordered</i> means that either operand may be a QNAN.</p>
5692
5693<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5694 a <a href="#t_floating">floating point</a> type or
5695 a <a href="#t_vector">vector</a> of floating point type. They must have
5696 identical types.</p>
5697
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5698<h5>Semantics:</h5>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]5699<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5700 according to the condition code given as <tt>cond</tt>. If the operands are
5701 vectors, then the vectors are compared element by element. Each comparison
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]5702 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5703 follows:</p>
5704
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5705<ol>
5706 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5707
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5708 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5709 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5710
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5711 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]5712 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5713
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5714 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5715 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5716
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5717 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5718 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5719
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5720 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5721 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5722
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5723 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5724 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5725
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5726 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5727
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5728 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5729 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5730
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5731 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5732 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5733
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5734 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5735 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5736
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5737 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5738 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5739
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5740 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5741 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5742
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5743 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5744 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5745
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5746 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5747
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5748 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5749</ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5750
5751<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5752<pre>
5753 &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanf72fb672008-09-09 01:02:47 +0000[diff] [blame]5754 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5755 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5756 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5757</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]5758
5759<p>Note that the code generator does not yet support vector types with
5760 the <tt>fcmp</tt> instruction.</p>
5761
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5762</div>
5763
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5764<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5765<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5766 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5767</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5768
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5769<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5770
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5771<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5772<pre>
5773 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
5774</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5775
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5776<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5777<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
5778 SSA graph representing the function.</p>
5779
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5780<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5781<p>The type of the incoming values is specified with the first type field. After
5782 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5783 one pair for each predecessor basic block of the current block. Only values
5784 of <a href="#t_firstclass">first class</a> type may be used as the value
5785 arguments to the PHI node. Only labels may be used as the label
5786 arguments.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5787
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5788<p>There must be no non-phi instructions between the start of a basic block and
5789 the PHI instructions: i.e. PHI instructions must be first in a basic
5790 block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5791
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5792<p>For the purposes of the SSA form, the use of each incoming value is deemed to
5793 occur on the edge from the corresponding predecessor block to the current
5794 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5795 value on the same edge).</p>
Jay Foadd2449092009-06-03 10:20:10 +0000[diff] [blame]5796
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5797<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5798<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5799 specified by the pair corresponding to the predecessor basic block that
5800 executed just prior to the current block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5801
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5802<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5803<pre>
5804Loop: ; Infinite loop that counts from 0 on up...
5805 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5806 %nextindvar = add i32 %indvar, 1
5807 br label %Loop
5808</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5809
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5810</div>
5811
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5812<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5813<h4>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5814 <a name="i_select">'<tt>select</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5815</h4>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5816
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5817<div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5818
5819<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5820<pre>
Dan Gohmanf72fb672008-09-09 01:02:47 +0000[diff] [blame]5821 &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>
5822
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5823 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5824</pre>
5825
5826<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5827<p>The '<tt>select</tt>' instruction is used to choose one value based on a
5828 condition, without branching.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5829
5830
5831<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5832<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5833 values indicating the condition, and two values of the
5834 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5835 vectors and the condition is a scalar, then entire vectors are selected, not
5836 individual elements.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5837
5838<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5839<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5840 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5841
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5842<p>If the condition is a vector of i1, then the value arguments must be vectors
5843 of the same size, and the selection is done element by element.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5844
5845<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5846<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5847 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5848</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]5849
5850<p>Note that the code generator does not yet support conditions
5851 with vector type.</p>
5852
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5853</div>
5854
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]5855<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5856<h4>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5857 <a name="i_call">'<tt>call</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5858</h4>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5859
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5860<div>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5861
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5862<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5863<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]5864 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5865</pre>
5866
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5867<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5868<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5869
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5870<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5871<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5872
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5873<ol>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]5874 <li>The optional "tail" marker indicates that the callee function does not
5875 access any allocas or varargs in the caller. Note that calls may be
5876 marked "tail" even if they do not occur before
5877 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5878 present, the function call is eligible for tail call optimization,
5879 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Chengdc444e92010-03-08 21:05:02 +0000[diff] [blame]5880 optimized into a jump</a>. The code generator may optimize calls marked
5881 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5882 sibling call optimization</a> when the caller and callee have
5883 matching signatures, or 2) forced tail call optimization when the
5884 following extra requirements are met:
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]5885 <ul>
5886 <li>Caller and callee both have the calling
5887 convention <tt>fastcc</tt>.</li>
5888 <li>The call is in tail position (ret immediately follows call and ret
5889 uses value of call or is void).</li>
5890 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohmanfbbee8d2010-03-02 01:08:11 +0000[diff] [blame]5891 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]5892 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5893 constraints are met.</a></li>
5894 </ul>
5895 </li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]5896
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5897 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5898 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]5899 defaults to using C calling conventions. The calling convention of the
5900 call must match the calling convention of the target function, or else the
5901 behavior is undefined.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]5902
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5903 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5904 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5905 '<tt>inreg</tt>' attributes are valid here.</li>
5906
5907 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5908 type of the return value. Functions that return no value are marked
5909 <tt><a href="#t_void">void</a></tt>.</li>
5910
5911 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5912 being invoked. The argument types must match the types implied by this
5913 signature. This type can be omitted if the function is not varargs and if
5914 the function type does not return a pointer to a function.</li>
5915
5916 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5917 be invoked. In most cases, this is a direct function invocation, but
5918 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5919 to function value.</li>
5920
5921 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]5922 signature argument types and parameter attributes. All arguments must be
5923 of <a href="#t_firstclass">first class</a> type. If the function
5924 signature indicates the function accepts a variable number of arguments,
5925 the extra arguments can be specified.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5926
5927 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5928 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5929 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5930</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5931
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5932<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5933<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5934 a specified function, with its incoming arguments bound to the specified
5935 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5936 function, control flow continues with the instruction after the function
5937 call, and the return value of the function is bound to the result
5938 argument.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5939
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5940<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5941<pre>
Nick Lewyckydb7e3c92007-09-08 13:57:50 +0000[diff] [blame]5942 %retval = call i32 @test(i32 %argc)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]5943 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattner772fccf2008-03-21 17:24:17 +0000[diff] [blame]5944 %X = tail call i32 @foo() <i>; yields i32</i>
5945 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5946 call void %foo(i8 97 signext)
Devang Patelc3fc6df2008-03-10 20:49:15 +0000[diff] [blame]5947
5948 %struct.A = type { i32, i8 }
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]5949 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]5950 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5951 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner85a350f2008-10-08 06:26:11 +0000[diff] [blame]5952 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmancb73d192008-10-07 10:03:45 +0000[diff] [blame]5953 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]5954</pre>
5955
Dale Johannesen07de8d12009-09-24 18:38:21 +0000[diff] [blame]5956<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen9f8380b2009-09-25 17:04:42 +0000[diff] [blame]5957standard C99 library as being the C99 library functions, and may perform
5958optimizations or generate code for them under that assumption. This is
5959something we'd like to change in the future to provide better support for
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]5960freestanding environments and non-C-based languages.</p>
Dale Johannesen07de8d12009-09-24 18:38:21 +0000[diff] [blame]5961
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5962</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5963
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]5964<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5965<h4>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]5966 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5967</h4>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5968
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5969<div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5970
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]5971<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5972<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]5973 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5974</pre>
5975
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]5976<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]5977<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5978 the "variable argument" area of a function call. It is used to implement the
5979 <tt>va_arg</tt> macro in C.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5980
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]5981<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5982<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5983 argument. It returns a value of the specified argument type and increments
5984 the <tt>va_list</tt> to point to the next argument. The actual type
5985 of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5986
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]5987<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5988<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5989 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5990 to the next argument. For more information, see the variable argument
5991 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5992
5993<p>It is legal for this instruction to be called in a function which does not
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5994 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5995 function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]5996
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5997<p><tt>va_arg</tt> is an LLVM instruction instead of
5998 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5999 argument.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6000
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6001<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6002<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6003
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6004<p>Note that the code generator does not yet fully support va_arg on many
6005 targets. Also, it does not currently support va_arg with aggregate types on
6006 any target.</p>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +0000[diff] [blame]6007
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6008</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6009
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6010<!-- _______________________________________________________________________ -->
6011<h4>
6012 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6013</h4>
6014
6015<div>
6016
6017<h5>Syntax:</h5>
6018<pre>
Bill Wendlingbf13ee12011-08-08 08:06:05 +0000[diff] [blame]6019 &lt;resultval&gt; = landingpad &lt;somety&gt; personality &lt;type&gt; &lt;pers_fn&gt; &lt;clause&gt;+
6020 &lt;resultval&gt; = landingpad &lt;somety&gt; personality &lt;type&gt; &lt;pers_fn&gt; cleanup &lt;clause&gt;*
6021
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6022 &lt;clause&gt; := catch &lt;type&gt; &lt;value&gt;
Bill Wendlingbf13ee12011-08-08 08:06:05 +0000[diff] [blame]6023 &lt;clause&gt; := filter &lt;type&gt; &lt;value&gt; {, &lt;type&gt; &lt;value&gt;}*
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6024</pre>
6025
6026<h5>Overview:</h5>
6027<p>The '<tt>landingpad</tt>' instruction is used by
6028 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6029 system</a> to specify that a basic block is a landing pad &mdash; one where
6030 the exception lands, and corresponds to the code found in the
6031 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6032 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6033 re-entry to the function. The <tt>resultval</tt> has the
6034 type <tt>somety</tt>.</p>
6035
6036<h5>Arguments:</h5>
6037<p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6038 function associated with the unwinding mechanism. The optional
6039 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6040
6041<p>A <tt>clause</tt> begins with the clause type &mdash; <tt>catch</tt>
6042 or <tt>filter</tt> &mdash; and contains a list of global variables
6043 representing the "types" that may be caught or filtered respectively. The
6044 '<tt>landingpad</tt>' instruction must contain <em>at least</em>
6045 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6046
6047<h5>Semantics:</h5>
6048<p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6049 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6050 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6051 calling conventions, how the personality function results are represented in
6052 LLVM IR is target specific.</p>
6053
Bill Wendlingb7a01352011-08-03 17:17:06 +0000[diff] [blame]6054<p>The clauses are applied in order from top to bottom. If two
6055 <tt>landingpad</tt> instructions are merged together through inlining, the
Bill Wendling2905c322011-08-08 07:58:58 +0000[diff] [blame]6056 clauses from the calling function are appended to the list of clauses.</p>
Bill Wendlingb7a01352011-08-03 17:17:06 +0000[diff] [blame]6057
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6058<p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6059
6060<ul>
6061 <li>A landing pad block is a basic block which is the unwind destination of an
6062 '<tt>invoke</tt>' instruction.</li>
6063 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6064 first non-PHI instruction.</li>
6065 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6066 pad block.</li>
6067 <li>A basic block that is not a landing pad block may not include a
6068 '<tt>landingpad</tt>' instruction.</li>
6069 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6070 personality function.</li>
6071</ul>
6072
6073<h5>Example:</h5>
6074<pre>
6075 ;; A landing pad which can catch an integer.
6076 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6077 catch i8** @_ZTIi
6078 ;; A landing pad that is a cleanup.
6079 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6080 cleanup
6081 ;; A landing pad which can catch an integer and can only throw a double.
6082 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6083 catch i8** @_ZTIi
6084 filter i8** @_ZTId
6085</pre>
6086
6087</div>
6088
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6089</div>
6090
6091</div>
6092
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6093<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6094<h2><a name="intrinsics">Intrinsic Functions</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]6095<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6096
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6097<div>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6098
6099<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6100 well known names and semantics and are required to follow certain
6101 restrictions. Overall, these intrinsics represent an extension mechanism for
6102 the LLVM language that does not require changing all of the transformations
6103 in LLVM when adding to the language (or the bitcode reader/writer, the
6104 parser, etc...).</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6105
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]6106<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6107 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6108 begin with this prefix. Intrinsic functions must always be external
6109 functions: you cannot define the body of intrinsic functions. Intrinsic
6110 functions may only be used in call or invoke instructions: it is illegal to
6111 take the address of an intrinsic function. Additionally, because intrinsic
6112 functions are part of the LLVM language, it is required if any are added that
6113 they be documented here.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6114
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6115<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6116 family of functions that perform the same operation but on different data
6117 types. Because LLVM can represent over 8 million different integer types,
6118 overloading is used commonly to allow an intrinsic function to operate on any
6119 integer type. One or more of the argument types or the result type can be
6120 overloaded to accept any integer type. Argument types may also be defined as
6121 exactly matching a previous argument's type or the result type. This allows
6122 an intrinsic function which accepts multiple arguments, but needs all of them
6123 to be of the same type, to only be overloaded with respect to a single
6124 argument or the result.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6125
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6126<p>Overloaded intrinsics will have the names of its overloaded argument types
6127 encoded into its function name, each preceded by a period. Only those types
6128 which are overloaded result in a name suffix. Arguments whose type is matched
6129 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6130 can take an integer of any width and returns an integer of exactly the same
6131 integer width. This leads to a family of functions such as
6132 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6133 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6134 suffix is required. Because the argument's type is matched against the return
6135 type, it does not require its own name suffix.</p>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]6136
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6137<p>To learn how to add an intrinsic function, please see the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6138 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6139
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6140<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6141<h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6142 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6143</h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6144
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6145<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6146
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6147<p>Variable argument support is defined in LLVM with
6148 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6149 intrinsic functions. These functions are related to the similarly named
6150 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6151
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6152<p>All of these functions operate on arguments that use a target-specific value
6153 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6154 not define what this type is, so all transformations should be prepared to
6155 handle these functions regardless of the type used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6156
Chris Lattner374ab302006-05-15 17:26:46 +0000[diff] [blame]6157<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6158 instruction and the variable argument handling intrinsic functions are
6159 used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6160
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]6161<pre class="doc_code">
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6162define i32 @test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6163 ; Initialize variable argument processing
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6164 %ap = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6165 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6166 call void @llvm.va_start(i8* %ap2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6167
6168 ; Read a single integer argument
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6169 %tmp = va_arg i8** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6170
6171 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6172 %aq = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6173 %aq2 = bitcast i8** %aq to i8*
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6174 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6175 call void @llvm.va_end(i8* %aq2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6176
6177 ; Stop processing of arguments.
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6178 call void @llvm.va_end(i8* %ap2)
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]6179 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6180}
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6181
6182declare void @llvm.va_start(i8*)
6183declare void @llvm.va_copy(i8*, i8*)
6184declare void @llvm.va_end(i8*)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6185</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6186
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6187<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6188<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6189 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6190</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6191
6192
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6193<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6194
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6195<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6196<pre>
6197 declare void %llvm.va_start(i8* &lt;arglist&gt;)
6198</pre>
6199
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6200<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6201<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
6202 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6203
6204<h5>Arguments:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6205<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6206
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6207<h5>Semantics:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6208<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6209 macro available in C. In a target-dependent way, it initializes
6210 the <tt>va_list</tt> element to which the argument points, so that the next
6211 call to <tt>va_arg</tt> will produce the first variable argument passed to
6212 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6213 need to know the last argument of the function as the compiler can figure
6214 that out.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6215
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6216</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6217
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6218<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6219<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6220 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6221</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6222
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6223<div>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6224
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6225<h5>Syntax:</h5>
6226<pre>
6227 declare void @llvm.va_end(i8* &lt;arglist&gt;)
6228</pre>
6229
6230<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6231<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6232 which has been initialized previously
6233 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6234 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6235
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6236<h5>Arguments:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6237<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6238
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6239<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6240<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6241 macro available in C. In a target-dependent way, it destroys
6242 the <tt>va_list</tt> element to which the argument points. Calls
6243 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6244 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6245 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6246
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6247</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6248
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6249<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6250<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6251 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6252</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6253
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6254<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6255
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6256<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6257<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]6258 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6259</pre>
6260
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6261<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6262<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6263 from the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6264
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6265<h5>Arguments:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6266<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6267 The second argument is a pointer to a <tt>va_list</tt> element to copy
6268 from.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6269
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6270<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6271<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6272 macro available in C. In a target-dependent way, it copies the
6273 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6274 element. This intrinsic is necessary because
6275 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6276 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6277
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6278</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6279
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6280</div>
6281
Bill Wendling0246bb72011-07-31 06:45:03 +0000[diff] [blame]6282</div>
6283
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6284<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6285<h3>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6286 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6287</h3>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6288
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6289<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6290
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6291<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]6292Collection</a> (GC) requires the implementation and generation of these
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6293intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6294roots on the stack</a>, as well as garbage collector implementations that
6295require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6296barriers. Front-ends for type-safe garbage collected languages should generate
6297these intrinsics to make use of the LLVM garbage collectors. For more details,
6298see <a href="GarbageCollection.html">Accurate Garbage Collection with
6299LLVM</a>.</p>
Christopher Lamb303dae92007-12-17 01:00:21 +0000[diff] [blame]6300
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6301<p>The garbage collection intrinsics only operate on objects in the generic
6302 address space (address space zero).</p>
Christopher Lamb303dae92007-12-17 01:00:21 +0000[diff] [blame]6303
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6304<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6305<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6306 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6307</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6308
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6309<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6310
6311<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6312<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6313 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6314</pre>
6315
6316<h5>Overview:</h5>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]6317<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6318 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6319
6320<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6321<p>The first argument specifies the address of a stack object that contains the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6322 root pointer. The second pointer (which must be either a constant or a
6323 global value address) contains the meta-data to be associated with the
6324 root.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6325
6326<h5>Semantics:</h5>
Chris Lattner05d67092008-04-24 05:59:56 +0000[diff] [blame]6327<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6328 location. At compile-time, the code generator generates information to allow
6329 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6330 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6331 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6332
6333</div>
6334
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6335<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6336<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6337 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6338</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6339
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6340<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6341
6342<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6343<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6344 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6345</pre>
6346
6347<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6348<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6349 locations, allowing garbage collector implementations that require read
6350 barriers.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6351
6352<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]6353<p>The second argument is the address to read from, which should be an address
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6354 allocated from the garbage collector. The first object is a pointer to the
6355 start of the referenced object, if needed by the language runtime (otherwise
6356 null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6357
6358<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6359<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6360 instruction, but may be replaced with substantially more complex code by the
6361 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6362 may only be used in a function which <a href="#gc">specifies a GC
6363 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6364
6365</div>
6366
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6367<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6368<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6369 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6370</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6371
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6372<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6373
6374<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6375<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6376 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6377</pre>
6378
6379<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6380<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6381 locations, allowing garbage collector implementations that require write
6382 barriers (such as generational or reference counting collectors).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6383
6384<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]6385<p>The first argument is the reference to store, the second is the start of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6386 object to store it to, and the third is the address of the field of Obj to
6387 store to. If the runtime does not require a pointer to the object, Obj may
6388 be null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6389
6390<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6391<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6392 instruction, but may be replaced with substantially more complex code by the
6393 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6394 may only be used in a function which <a href="#gc">specifies a GC
6395 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6396
6397</div>
6398
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6399</div>
6400
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6401<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6402<h3>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6403 <a name="int_codegen">Code Generator Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6404</h3>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6405
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6406<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6407
6408<p>These intrinsics are provided by LLVM to expose special features that may
6409 only be implemented with code generator support.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6410
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6411<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6412<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6413 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6414</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6415
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6416<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6417
6418<h5>Syntax:</h5>
6419<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]6420 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6421</pre>
6422
6423<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6424<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6425 target-specific value indicating the return address of the current function
6426 or one of its callers.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6427
6428<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6429<p>The argument to this intrinsic indicates which function to return the address
6430 for. Zero indicates the calling function, one indicates its caller, etc.
6431 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6432
6433<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6434<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6435 indicating the return address of the specified call frame, or zero if it
6436 cannot be identified. The value returned by this intrinsic is likely to be
6437 incorrect or 0 for arguments other than zero, so it should only be used for
6438 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6439
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6440<p>Note that calling this intrinsic does not prevent function inlining or other
6441 aggressive transformations, so the value returned may not be that of the
6442 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6443
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6444</div>
6445
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6446<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6447<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6448 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6449</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6450
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6451<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6452
6453<h5>Syntax:</h5>
6454<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6455 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6456</pre>
6457
6458<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6459<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6460 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6461
6462<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6463<p>The argument to this intrinsic indicates which function to return the frame
6464 pointer for. Zero indicates the calling function, one indicates its caller,
6465 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6466
6467<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6468<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6469 indicating the frame address of the specified call frame, or zero if it
6470 cannot be identified. The value returned by this intrinsic is likely to be
6471 incorrect or 0 for arguments other than zero, so it should only be used for
6472 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6473
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6474<p>Note that calling this intrinsic does not prevent function inlining or other
6475 aggressive transformations, so the value returned may not be that of the
6476 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6477
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6478</div>
6479
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6480<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6481<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6482 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6483</h4>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6484
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6485<div>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6486
6487<h5>Syntax:</h5>
6488<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6489 declare i8* @llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6490</pre>
6491
6492<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6493<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6494 of the function stack, for use
6495 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6496 useful for implementing language features like scoped automatic variable
6497 sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6498
6499<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6500<p>This intrinsic returns a opaque pointer value that can be passed
6501 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6502 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6503 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6504 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6505 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6506 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6507
6508</div>
6509
6510<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6511<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6512 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6513</h4>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6514
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6515<div>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6516
6517<h5>Syntax:</h5>
6518<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6519 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6520</pre>
6521
6522<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6523<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6524 the function stack to the state it was in when the
6525 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6526 executed. This is useful for implementing language features like scoped
6527 automatic variable sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6528
6529<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6530<p>See the description
6531 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6532
6533</div>
6534
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6535<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6536<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6537 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6538</h4>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6539
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6540<div>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6541
6542<h5>Syntax:</h5>
6543<pre>
Bruno Cardoso Lopes9a767332011-06-14 04:58:37 +0000[diff] [blame]6544 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;, i32 &lt;cache type&gt;)
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6545</pre>
6546
6547<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6548<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6549 insert a prefetch instruction if supported; otherwise, it is a noop.
6550 Prefetches have no effect on the behavior of the program but can change its
6551 performance characteristics.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6552
6553<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6554<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6555 specifier determining if the fetch should be for a read (0) or write (1),
6556 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Bruno Cardoso Lopes9a767332011-06-14 04:58:37 +0000[diff] [blame]6557 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
6558 specifies whether the prefetch is performed on the data (1) or instruction (0)
6559 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
6560 must be constant integers.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6561
6562<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6563<p>This intrinsic does not modify the behavior of the program. In particular,
6564 prefetches cannot trap and do not produce a value. On targets that support
6565 this intrinsic, the prefetch can provide hints to the processor cache for
6566 better performance.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6567
6568</div>
6569
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6570<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6571<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6572 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6573</h4>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6574
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6575<div>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6576
6577<h5>Syntax:</h5>
6578<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6579 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6580</pre>
6581
6582<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6583<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6584 Counter (PC) in a region of code to simulators and other tools. The method
6585 is target specific, but it is expected that the marker will use exported
6586 symbols to transmit the PC of the marker. The marker makes no guarantees
6587 that it will remain with any specific instruction after optimizations. It is
6588 possible that the presence of a marker will inhibit optimizations. The
6589 intended use is to be inserted after optimizations to allow correlations of
6590 simulation runs.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6591
6592<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6593<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6594
6595<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6596<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6597 not support this intrinsic may ignore it.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6598
6599</div>
6600
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6601<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6602<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6603 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6604</h4>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6605
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6606<div>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6607
6608<h5>Syntax:</h5>
6609<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]6610 declare i64 @llvm.readcyclecounter()
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6611</pre>
6612
6613<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6614<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6615 counter register (or similar low latency, high accuracy clocks) on those
6616 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6617 should map to RPCC. As the backing counters overflow quickly (on the order
6618 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6619
6620<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6621<p>When directly supported, reading the cycle counter should not modify any
6622 memory. Implementations are allowed to either return a application specific
6623 value or a system wide value. On backends without support, this is lowered
6624 to a constant 0.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6625
6626</div>
6627
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6628</div>
6629
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6630<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6631<h3>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6632 <a name="int_libc">Standard C Library Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6633</h3>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6634
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6635<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6636
6637<p>LLVM provides intrinsics for a few important standard C library functions.
6638 These intrinsics allow source-language front-ends to pass information about
6639 the alignment of the pointer arguments to the code generator, providing
6640 opportunity for more efficient code generation.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6641
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6642<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6643<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6644 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6645</h4>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6646
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6647<div>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6648
6649<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6650<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
Mon P Wange88909b2010-04-07 06:35:53 +0000[diff] [blame]6651 integer bit width and for different address spaces. Not all targets support
6652 all bit widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6653
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6654<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6655 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6656 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6657 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6658 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6659</pre>
6660
6661<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6662<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6663 source location to the destination location.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6664
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6665<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6666 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6667 and the pointers can be in specified address spaces.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6668
6669<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6670
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6671<p>The first argument is a pointer to the destination, the second is a pointer
6672 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6673 number of bytes to copy, the fourth argument is the alignment of the
6674 source and destination locations, and the fifth is a boolean indicating a
6675 volatile access.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6676
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6677<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6678 then the caller guarantees that both the source and destination pointers are
6679 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]6680
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]6681<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6682 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6683 The detailed access behavior is not very cleanly specified and it is unwise
6684 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6685
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6686<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6687
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6688<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6689 source location to the destination location, which are not allowed to
6690 overlap. It copies "len" bytes of memory over. If the argument is known to
6691 be aligned to some boundary, this can be specified as the fourth argument,
6692 otherwise it should be set to 0 or 1.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6693
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6694</div>
6695
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6696<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6697<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6698 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6699</h4>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6700
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6701<div>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6702
6703<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +0000[diff] [blame]6704<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Mon P Wange88909b2010-04-07 06:35:53 +0000[diff] [blame]6705 width and for different address space. Not all targets support all bit
6706 widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6707
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6708<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6709 declare void @llvm.memmove.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6710 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6711 declare void @llvm.memmove.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6712 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6713</pre>
6714
6715<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6716<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6717 source location to the destination location. It is similar to the
6718 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6719 overlap.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6720
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6721<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6722 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6723 and the pointers can be in specified address spaces.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6724
6725<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6726
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6727<p>The first argument is a pointer to the destination, the second is a pointer
6728 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6729 number of bytes to copy, the fourth argument is the alignment of the
6730 source and destination locations, and the fifth is a boolean indicating a
6731 volatile access.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6732
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6733<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6734 then the caller guarantees that the source and destination pointers are
6735 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]6736
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]6737<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6738 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6739 The detailed access behavior is not very cleanly specified and it is unwise
6740 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6741
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6742<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6743
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6744<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6745 source location to the destination location, which may overlap. It copies
6746 "len" bytes of memory over. If the argument is known to be aligned to some
6747 boundary, this can be specified as the fourth argument, otherwise it should
6748 be set to 0 or 1.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6749
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6750</div>
6751
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6752<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6753<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6754 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6755</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6756
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6757<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6758
6759<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +0000[diff] [blame]6760<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]6761 width and for different address spaces. However, not all targets support all
6762 bit widths.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6763
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6764<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6765 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +0000[diff] [blame]6766 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6767 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +0000[diff] [blame]6768 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6769</pre>
6770
6771<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6772<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6773 particular byte value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6774
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6775<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]6776 intrinsic does not return a value and takes extra alignment/volatile
6777 arguments. Also, the destination can be in an arbitrary address space.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6778
6779<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6780<p>The first argument is a pointer to the destination to fill, the second is the
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]6781 byte value with which to fill it, the third argument is an integer argument
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6782 specifying the number of bytes to fill, and the fourth argument is the known
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]6783 alignment of the destination location.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6784
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6785<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6786 then the caller guarantees that the destination pointer is aligned to that
6787 boundary.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6788
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]6789<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6790 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6791 The detailed access behavior is not very cleanly specified and it is unwise
6792 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6793
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6794<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6795<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6796 at the destination location. If the argument is known to be aligned to some
6797 boundary, this can be specified as the fourth argument, otherwise it should
6798 be set to 0 or 1.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6799
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6800</div>
6801
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]6802<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6803<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6804 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6805</h4>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6806
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6807<div>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6808
6809<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6810<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6811 floating point or vector of floating point type. Not all targets support all
6812 types however.</p>
6813
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6814<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +0000[diff] [blame]6815 declare float @llvm.sqrt.f32(float %Val)
6816 declare double @llvm.sqrt.f64(double %Val)
6817 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6818 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6819 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6820</pre>
6821
6822<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6823<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6824 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6825 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6826 behavior for negative numbers other than -0.0 (which allows for better
6827 optimization, because there is no need to worry about errno being
6828 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6829
6830<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6831<p>The argument and return value are floating point numbers of the same
6832 type.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6833
6834<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6835<p>This function returns the sqrt of the specified operand if it is a
6836 nonnegative floating point number.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6837
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6838</div>
6839
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6840<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6841<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6842 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6843</h4>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6844
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6845<div>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6846
6847<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6848<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6849 floating point or vector of floating point type. Not all targets support all
6850 types however.</p>
6851
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6852<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +0000[diff] [blame]6853 declare float @llvm.powi.f32(float %Val, i32 %power)
6854 declare double @llvm.powi.f64(double %Val, i32 %power)
6855 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6856 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6857 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6858</pre>
6859
6860<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6861<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6862 specified (positive or negative) power. The order of evaluation of
6863 multiplications is not defined. When a vector of floating point type is
6864 used, the second argument remains a scalar integer value.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6865
6866<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6867<p>The second argument is an integer power, and the first is a value to raise to
6868 that power.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6869
6870<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6871<p>This function returns the first value raised to the second power with an
6872 unspecified sequence of rounding operations.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6873
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]6874</div>
6875
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6876<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6877<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6878 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6879</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6880
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6881<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6882
6883<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6884<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6885 floating point or vector of floating point type. Not all targets support all
6886 types however.</p>
6887
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6888<pre>
6889 declare float @llvm.sin.f32(float %Val)
6890 declare double @llvm.sin.f64(double %Val)
6891 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6892 declare fp128 @llvm.sin.f128(fp128 %Val)
6893 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6894</pre>
6895
6896<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6897<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6898
6899<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6900<p>The argument and return value are floating point numbers of the same
6901 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6902
6903<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6904<p>This function returns the sine of the specified operand, returning the same
6905 values as the libm <tt>sin</tt> functions would, and handles error conditions
6906 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6907
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6908</div>
6909
6910<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6911<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6912 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6913</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6914
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6915<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6916
6917<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6918<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6919 floating point or vector of floating point type. Not all targets support all
6920 types however.</p>
6921
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6922<pre>
6923 declare float @llvm.cos.f32(float %Val)
6924 declare double @llvm.cos.f64(double %Val)
6925 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6926 declare fp128 @llvm.cos.f128(fp128 %Val)
6927 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6928</pre>
6929
6930<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6931<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6932
6933<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6934<p>The argument and return value are floating point numbers of the same
6935 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6936
6937<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6938<p>This function returns the cosine of the specified operand, returning the same
6939 values as the libm <tt>cos</tt> functions would, and handles error conditions
6940 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6941
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6942</div>
6943
6944<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6945<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6946 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6947</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6948
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6949<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6950
6951<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6952<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6953 floating point or vector of floating point type. Not all targets support all
6954 types however.</p>
6955
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6956<pre>
6957 declare float @llvm.pow.f32(float %Val, float %Power)
6958 declare double @llvm.pow.f64(double %Val, double %Power)
6959 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6960 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6961 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6962</pre>
6963
6964<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6965<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6966 specified (positive or negative) power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6967
6968<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6969<p>The second argument is a floating point power, and the first is a value to
6970 raise to that power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6971
6972<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6973<p>This function returns the first value raised to the second power, returning
6974 the same values as the libm <tt>pow</tt> functions would, and handles error
6975 conditions in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6976
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]6977</div>
6978
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6979</div>
6980
Dan Gohman4e9011c2011-05-23 21:13:03 +0000[diff] [blame]6981<!-- _______________________________________________________________________ -->
6982<h4>
6983 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
6984</h4>
6985
6986<div>
6987
6988<h5>Syntax:</h5>
6989<p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
6990 floating point or vector of floating point type. Not all targets support all
6991 types however.</p>
6992
6993<pre>
6994 declare float @llvm.exp.f32(float %Val)
6995 declare double @llvm.exp.f64(double %Val)
6996 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
6997 declare fp128 @llvm.exp.f128(fp128 %Val)
6998 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
6999</pre>
7000
7001<h5>Overview:</h5>
7002<p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7003
7004<h5>Arguments:</h5>
7005<p>The argument and return value are floating point numbers of the same
7006 type.</p>
7007
7008<h5>Semantics:</h5>
7009<p>This function returns the same values as the libm <tt>exp</tt> functions
7010 would, and handles error conditions in the same way.</p>
7011
7012</div>
7013
7014<!-- _______________________________________________________________________ -->
7015<h4>
7016 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7017</h4>
7018
7019<div>
7020
7021<h5>Syntax:</h5>
7022<p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7023 floating point or vector of floating point type. Not all targets support all
7024 types however.</p>
7025
7026<pre>
7027 declare float @llvm.log.f32(float %Val)
7028 declare double @llvm.log.f64(double %Val)
7029 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7030 declare fp128 @llvm.log.f128(fp128 %Val)
7031 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7032</pre>
7033
7034<h5>Overview:</h5>
7035<p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7036
7037<h5>Arguments:</h5>
7038<p>The argument and return value are floating point numbers of the same
7039 type.</p>
7040
7041<h5>Semantics:</h5>
7042<p>This function returns the same values as the libm <tt>log</tt> functions
7043 would, and handles error conditions in the same way.</p>
7044
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]7045<h4>
7046 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7047</h4>
7048
7049<div>
7050
7051<h5>Syntax:</h5>
7052<p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7053 floating point or vector of floating point type. Not all targets support all
7054 types however.</p>
7055
7056<pre>
7057 declare float @llvm.fma.f32(float %a, float %b, float %c)
7058 declare double @llvm.fma.f64(double %a, double %b, double %c)
7059 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7060 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7061 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7062</pre>
7063
7064<h5>Overview:</h5>
Cameron Zwarichabc43e62011-07-08 22:13:55 +0000[diff] [blame]7065<p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]7066 operation.</p>
7067
7068<h5>Arguments:</h5>
7069<p>The argument and return value are floating point numbers of the same
7070 type.</p>
7071
7072<h5>Semantics:</h5>
7073<p>This function returns the same values as the libm <tt>fma</tt> functions
7074 would.</p>
7075
Dan Gohman4e9011c2011-05-23 21:13:03 +0000[diff] [blame]7076</div>
7077
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7078<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7079<h3>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7080 <a name="int_manip">Bit Manipulation Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7081</h3>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7082
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7083<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7084
7085<p>LLVM provides intrinsics for a few important bit manipulation operations.
7086 These allow efficient code generation for some algorithms.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7087
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7088<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7089<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7090 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7091</h4>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7092
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7093<div>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7094
7095<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]7096<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7097 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7098
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7099<pre>
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7100 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
7101 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
7102 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7103</pre>
7104
7105<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7106<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7107 values with an even number of bytes (positive multiple of 16 bits). These
7108 are useful for performing operations on data that is not in the target's
7109 native byte order.</p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7110
7111<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7112<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7113 and low byte of the input i16 swapped. Similarly,
7114 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7115 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7116 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7117 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7118 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7119 more, respectively).</p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7120
7121</div>
7122
7123<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7124<h4>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]7125 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7126</h4>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7127
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7128<div>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7129
7130<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]7131<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7132 width, or on any vector with integer elements. Not all targets support all
7133 bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7134
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7135<pre>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7136 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7137 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]7138 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7139 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
7140 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7141 declare &lt;2 x i32&gt; @llvm.ctpop.v2i32(&lt;2 x i32&gt; &lt;src&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7142</pre>
7143
7144<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7145<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7146 in a value.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7147
7148<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7149<p>The only argument is the value to be counted. The argument may be of any
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7150 integer type, or a vector with integer elements.
7151 The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7152
7153<h5>Semantics:</h5>
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7154<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7155 element of a vector.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7156
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7157</div>
7158
7159<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7160<h4>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]7161 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7162</h4>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7163
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7164<div>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7165
7166<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7167<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7168 integer bit width, or any vector whose elements are integers. Not all
7169 targets support all bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7170
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7171<pre>
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7172 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
7173 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]7174 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7175 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
7176 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7177 declare &lt;2 x i32&gt; @llvm.ctlz.v2i32(&lt;2 x i32&gt; &lt;src;gt)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7178</pre>
7179
7180<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7181<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7182 leading zeros in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7183
7184<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7185<p>The only argument is the value to be counted. The argument may be of any
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7186 integer type, or any vector type with integer element type.
7187 The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7188
7189<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7190<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7191 zeros in a variable, or within each element of the vector if the operation
7192 is of vector type. If the src == 0 then the result is the size in bits of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7193 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7194
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7195</div>
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]7196
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7197<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7198<h4>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]7199 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7200</h4>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7201
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7202<div>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7203
7204<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7205<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7206 integer bit width, or any vector of integer elements. Not all targets
7207 support all bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7208
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7209<pre>
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7210 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
7211 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]7212 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7213 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
7214 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7215 declase &lt;2 x i32&gt; @llvm.cttz.v2i32(&lt;2 x i32&gt; &lt;src&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7216</pre>
7217
7218<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7219<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7220 trailing zeros.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7221
7222<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7223<p>The only argument is the value to be counted. The argument may be of any
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7224 integer type, or a vectory with integer element type.. The return type
7225 must match the argument type.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7226
7227<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7228<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7229 zeros in a variable, or within each element of a vector.
7230 If the src == 0 then the result is the size in bits of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7231 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7232
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7233</div>
7234
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7235</div>
7236
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7237<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7238<h3>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7239 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7240</h3>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7241
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7242<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7243
7244<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7245
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7246<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7247<h4>
7248 <a name="int_sadd_overflow">
7249 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7250 </a>
7251</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7252
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7253<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7254
7255<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7256<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7257 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7258
7259<pre>
7260 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7261 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7262 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7263</pre>
7264
7265<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7266<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7267 a signed addition of the two arguments, and indicate whether an overflow
7268 occurred during the signed summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7269
7270<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7271<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7272 be of integer types of any bit width, but they must have the same bit
7273 width. The second element of the result structure must be of
7274 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7275 undergo signed addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7276
7277<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7278<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7279 a signed addition of the two variables. They return a structure &mdash; the
7280 first element of which is the signed summation, and the second element of
7281 which is a bit specifying if the signed summation resulted in an
7282 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7283
7284<h5>Examples:</h5>
7285<pre>
7286 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7287 %sum = extractvalue {i32, i1} %res, 0
7288 %obit = extractvalue {i32, i1} %res, 1
7289 br i1 %obit, label %overflow, label %normal
7290</pre>
7291
7292</div>
7293
7294<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7295<h4>
7296 <a name="int_uadd_overflow">
7297 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7298 </a>
7299</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7300
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7301<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7302
7303<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7304<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7305 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7306
7307<pre>
7308 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7309 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7310 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7311</pre>
7312
7313<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7314<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7315 an unsigned addition of the two arguments, and indicate whether a carry
7316 occurred during the unsigned summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7317
7318<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7319<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7320 be of integer types of any bit width, but they must have the same bit
7321 width. The second element of the result structure must be of
7322 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7323 undergo unsigned addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7324
7325<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7326<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7327 an unsigned addition of the two arguments. They return a structure &mdash;
7328 the first element of which is the sum, and the second element of which is a
7329 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7330
7331<h5>Examples:</h5>
7332<pre>
7333 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7334 %sum = extractvalue {i32, i1} %res, 0
7335 %obit = extractvalue {i32, i1} %res, 1
7336 br i1 %obit, label %carry, label %normal
7337</pre>
7338
7339</div>
7340
7341<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7342<h4>
7343 <a name="int_ssub_overflow">
7344 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7345 </a>
7346</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7347
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7348<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7349
7350<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7351<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7352 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7353
7354<pre>
7355 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7356 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7357 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7358</pre>
7359
7360<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7361<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7362 a signed subtraction of the two arguments, and indicate whether an overflow
7363 occurred during the signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7364
7365<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7366<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7367 be of integer types of any bit width, but they must have the same bit
7368 width. The second element of the result structure must be of
7369 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7370 undergo signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7371
7372<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7373<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7374 a signed subtraction of the two arguments. They return a structure &mdash;
7375 the first element of which is the subtraction, and the second element of
7376 which is a bit specifying if the signed subtraction resulted in an
7377 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7378
7379<h5>Examples:</h5>
7380<pre>
7381 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7382 %sum = extractvalue {i32, i1} %res, 0
7383 %obit = extractvalue {i32, i1} %res, 1
7384 br i1 %obit, label %overflow, label %normal
7385</pre>
7386
7387</div>
7388
7389<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7390<h4>
7391 <a name="int_usub_overflow">
7392 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7393 </a>
7394</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7395
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7396<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7397
7398<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7399<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7400 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7401
7402<pre>
7403 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7404 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7405 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7406</pre>
7407
7408<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7409<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7410 an unsigned subtraction of the two arguments, and indicate whether an
7411 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7412
7413<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7414<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7415 be of integer types of any bit width, but they must have the same bit
7416 width. The second element of the result structure must be of
7417 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7418 undergo unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7419
7420<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7421<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7422 an unsigned subtraction of the two arguments. They return a structure &mdash;
7423 the first element of which is the subtraction, and the second element of
7424 which is a bit specifying if the unsigned subtraction resulted in an
7425 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7426
7427<h5>Examples:</h5>
7428<pre>
7429 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7430 %sum = extractvalue {i32, i1} %res, 0
7431 %obit = extractvalue {i32, i1} %res, 1
7432 br i1 %obit, label %overflow, label %normal
7433</pre>
7434
7435</div>
7436
7437<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7438<h4>
7439 <a name="int_smul_overflow">
7440 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7441 </a>
7442</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7443
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7444<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7445
7446<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7447<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7448 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7449
7450<pre>
7451 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7452 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7453 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7454</pre>
7455
7456<h5>Overview:</h5>
7457
7458<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7459 a signed multiplication of the two arguments, and indicate whether an
7460 overflow occurred during the signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7461
7462<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7463<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7464 be of integer types of any bit width, but they must have the same bit
7465 width. The second element of the result structure must be of
7466 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7467 undergo signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7468
7469<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7470<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7471 a signed multiplication of the two arguments. They return a structure &mdash;
7472 the first element of which is the multiplication, and the second element of
7473 which is a bit specifying if the signed multiplication resulted in an
7474 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7475
7476<h5>Examples:</h5>
7477<pre>
7478 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7479 %sum = extractvalue {i32, i1} %res, 0
7480 %obit = extractvalue {i32, i1} %res, 1
7481 br i1 %obit, label %overflow, label %normal
7482</pre>
7483
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]7484</div>
7485
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7486<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7487<h4>
7488 <a name="int_umul_overflow">
7489 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
7490 </a>
7491</h4>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7492
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7493<div>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7494
7495<h5>Syntax:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7496<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7497 on any integer bit width.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7498
7499<pre>
7500 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
7501 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7502 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
7503</pre>
7504
7505<h5>Overview:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7506<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7507 a unsigned multiplication of the two arguments, and indicate whether an
7508 overflow occurred during the unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7509
7510<h5>Arguments:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7511<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7512 be of integer types of any bit width, but they must have the same bit
7513 width. The second element of the result structure must be of
7514 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7515 undergo unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7516
7517<h5>Semantics:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7518<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7519 an unsigned multiplication of the two arguments. They return a structure
7520 &mdash; the first element of which is the multiplication, and the second
7521 element of which is a bit specifying if the unsigned multiplication resulted
7522 in an overflow.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7523
7524<h5>Examples:</h5>
7525<pre>
7526 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7527 %sum = extractvalue {i32, i1} %res, 0
7528 %obit = extractvalue {i32, i1} %res, 1
7529 br i1 %obit, label %overflow, label %normal
7530</pre>
7531
7532</div>
7533
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7534</div>
7535
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7536<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7537<h3>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7538 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7539</h3>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7540
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7541<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7542
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7543<p>Half precision floating point is a storage-only format. This means that it is
7544 a dense encoding (in memory) but does not support computation in the
7545 format.</p>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7546
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7547<p>This means that code must first load the half-precision floating point
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7548 value as an i16, then convert it to float with <a
7549 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
7550 Computation can then be performed on the float value (including extending to
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7551 double etc). To store the value back to memory, it is first converted to
7552 float if needed, then converted to i16 with
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7553 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
7554 storing as an i16 value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7555
7556<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7557<h4>
7558 <a name="int_convert_to_fp16">
7559 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
7560 </a>
7561</h4>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7562
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7563<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7564
7565<h5>Syntax:</h5>
7566<pre>
7567 declare i16 @llvm.convert.to.fp16(f32 %a)
7568</pre>
7569
7570<h5>Overview:</h5>
7571<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7572 a conversion from single precision floating point format to half precision
7573 floating point format.</p>
7574
7575<h5>Arguments:</h5>
7576<p>The intrinsic function contains single argument - the value to be
7577 converted.</p>
7578
7579<h5>Semantics:</h5>
7580<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7581 a conversion from single precision floating point format to half precision
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7582 floating point format. The return value is an <tt>i16</tt> which
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7583 contains the converted number.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7584
7585<h5>Examples:</h5>
7586<pre>
7587 %res = call i16 @llvm.convert.to.fp16(f32 %a)
7588 store i16 %res, i16* @x, align 2
7589</pre>
7590
7591</div>
7592
7593<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7594<h4>
7595 <a name="int_convert_from_fp16">
7596 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
7597 </a>
7598</h4>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7599
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7600<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7601
7602<h5>Syntax:</h5>
7603<pre>
7604 declare f32 @llvm.convert.from.fp16(i16 %a)
7605</pre>
7606
7607<h5>Overview:</h5>
7608<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
7609 a conversion from half precision floating point format to single precision
7610 floating point format.</p>
7611
7612<h5>Arguments:</h5>
7613<p>The intrinsic function contains single argument - the value to be
7614 converted.</p>
7615
7616<h5>Semantics:</h5>
7617<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7618 conversion from half single precision floating point format to single
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7619 precision floating point format. The input half-float value is represented by
7620 an <tt>i16</tt> value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7621
7622<h5>Examples:</h5>
7623<pre>
7624 %a = load i16* @x, align 2
7625 %res = call f32 @llvm.convert.from.fp16(i16 %a)
7626</pre>
7627
7628</div>
7629
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7630</div>
7631
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7632<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7633<h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7634 <a name="int_debugger">Debugger Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7635</h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7636
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7637<div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7638
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7639<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
7640 prefix), are described in
7641 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
7642 Level Debugging</a> document.</p>
7643
7644</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7645
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7646<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7647<h3>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7648 <a name="int_eh">Exception Handling Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7649</h3>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7650
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7651<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7652
7653<p>The LLVM exception handling intrinsics (which all start with
7654 <tt>llvm.eh.</tt> prefix), are described in
7655 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
7656 Handling</a> document.</p>
7657
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7658</div>
7659
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]7660<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7661<h3>
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]7662 <a name="int_trampoline">Trampoline Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7663</h3>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7664
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7665<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7666
7667<p>This intrinsic makes it possible to excise one parameter, marked with
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]7668 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
7669 The result is a callable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7670 function pointer lacking the nest parameter - the caller does not need to
7671 provide a value for it. Instead, the value to use is stored in advance in a
7672 "trampoline", a block of memory usually allocated on the stack, which also
7673 contains code to splice the nest value into the argument list. This is used
7674 to implement the GCC nested function address extension.</p>
7675
7676<p>For example, if the function is
7677 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
7678 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
7679 follows:</p>
7680
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]7681<pre class="doc_code">
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]7682 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
7683 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7684 %p = call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval)
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]7685 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7686</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7687
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7688<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
7689 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7690
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7691<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7692<h4>
7693 <a name="int_it">
7694 '<tt>llvm.init.trampoline</tt>' Intrinsic
7695 </a>
7696</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7697
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7698<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7699
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7700<h5>Syntax:</h5>
7701<pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7702 declare i8* @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7703</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7704
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7705<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7706<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
7707 function pointer suitable for executing it.</p>
7708
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7709<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7710<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7711 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7712 sufficiently aligned block of memory; this memory is written to by the
7713 intrinsic. Note that the size and the alignment are target-specific - LLVM
7714 currently provides no portable way of determining them, so a front-end that
7715 generates this intrinsic needs to have some target-specific knowledge.
7716 The <tt>func</tt> argument must hold a function bitcast to
7717 an <tt>i8*</tt>.</p>
7718
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7719<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7720<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
7721 dependent code, turning it into a function. A pointer to this function is
7722 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
7723 function pointer type</a> before being called. The new function's signature
7724 is the same as that of <tt>func</tt> with any arguments marked with
7725 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
7726 is allowed, and it must be of pointer type. Calling the new function is
7727 equivalent to calling <tt>func</tt> with the same argument list, but
7728 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
7729 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
7730 by <tt>tramp</tt> is modified, then the effect of any later call to the
7731 returned function pointer is undefined.</p>
7732
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7733</div>
7734
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7735</div>
7736
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7737<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7738<h3>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7739 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7740</h3>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7741
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7742<div>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7743
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7744<p>These intrinsic functions expand the "universal IR" of LLVM to represent
7745 hardware constructs for atomic operations and memory synchronization. This
7746 provides an interface to the hardware, not an interface to the programmer. It
7747 is aimed at a low enough level to allow any programming models or APIs
7748 (Application Programming Interfaces) which need atomic behaviors to map
7749 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
7750 hardware provides a "universal IR" for source languages, it also provides a
7751 starting point for developing a "universal" atomic operation and
7752 synchronization IR.</p>
7753
7754<p>These do <em>not</em> form an API such as high-level threading libraries,
7755 software transaction memory systems, atomic primitives, and intrinsic
7756 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
7757 application libraries. The hardware interface provided by LLVM should allow
7758 a clean implementation of all of these APIs and parallel programming models.
7759 No one model or paradigm should be selected above others unless the hardware
7760 itself ubiquitously does so.</p>
7761
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7762<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7763<h4>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7764 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7765</h4>
7766
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7767<div>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7768<h5>Syntax:</h5>
7769<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7770 declare void @llvm.memory.barrier(i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;, i1 &lt;device&gt;)
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7771</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7772
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7773<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7774<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7775 specific pairs of memory access types.</p>
7776
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7777<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7778<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7779 The first four arguments enables a specific barrier as listed below. The
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]7780 fifth argument specifies that the barrier applies to io or device or uncached
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7781 memory.</p>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7782
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7783<ul>
7784 <li><tt>ll</tt>: load-load barrier</li>
7785 <li><tt>ls</tt>: load-store barrier</li>
7786 <li><tt>sl</tt>: store-load barrier</li>
7787 <li><tt>ss</tt>: store-store barrier</li>
7788 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7789</ul>
7790
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7791<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7792<p>This intrinsic causes the system to enforce some ordering constraints upon
7793 the loads and stores of the program. This barrier does not
7794 indicate <em>when</em> any events will occur, it only enforces
7795 an <em>order</em> in which they occur. For any of the specified pairs of load
7796 and store operations (f.ex. load-load, or store-load), all of the first
7797 operations preceding the barrier will complete before any of the second
7798 operations succeeding the barrier begin. Specifically the semantics for each
7799 pairing is as follows:</p>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7800
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7801<ul>
7802 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7803 after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]7804 <li><tt>ls</tt>: All loads before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7805 store after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]7806 <li><tt>ss</tt>: All stores before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7807 store after the barrier begins.</li>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]7808 <li><tt>sl</tt>: All stores before the barrier must complete before any
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7809 load after the barrier begins.</li>
7810</ul>
7811
7812<p>These semantics are applied with a logical "and" behavior when more than one
7813 is enabled in a single memory barrier intrinsic.</p>
7814
7815<p>Backends may implement stronger barriers than those requested when they do
7816 not support as fine grained a barrier as requested. Some architectures do
7817 not need all types of barriers and on such architectures, these become
7818 noops.</p>
7819
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7820<h5>Example:</h5>
7821<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]7822%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7823%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7824 store i32 4, %ptr
7825
7826%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
Evan Cheng0b0669a2011-06-29 17:14:00 +0000[diff] [blame]7827 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false, i1 true)
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7828 <i>; guarantee the above finishes</i>
7829 store i32 8, %ptr <i>; before this begins</i>
7830</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7831
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]7832</div>
7833
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7834<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7835<h4>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]7836 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7837</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7838
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7839<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7840
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7841<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7842<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7843 any integer bit width and for different address spaces. Not all targets
7844 support all bit widths however.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7845
7846<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7847 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt;)
7848 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt;)
7849 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt;)
7850 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7851</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7852
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7853<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7854<p>This loads a value in memory and compares it to a given value. If they are
7855 equal, it stores a new value into the memory.</p>
7856
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7857<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7858<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7859 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7860 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7861 this integer type. While any bit width integer may be used, targets may only
7862 lower representations they support in hardware.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7863
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7864<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7865<p>This entire intrinsic must be executed atomically. It first loads the value
7866 in memory pointed to by <tt>ptr</tt> and compares it with the
7867 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7868 memory. The loaded value is yielded in all cases. This provides the
7869 equivalent of an atomic compare-and-swap operation within the SSA
7870 framework.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7871
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7872<h5>Examples:</h5>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7873<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]7874%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7875%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7876 store i32 4, %ptr
7877
7878%val1 = add i32 4, 4
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7879%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7880 <i>; yields {i32}:result1 = 4</i>
7881%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7882%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7883
7884%val2 = add i32 1, 1
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7885%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7886 <i>; yields {i32}:result2 = 8</i>
7887%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7888
7889%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7890</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7891
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7892</div>
7893
7894<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7895<h4>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7896 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7897</h4>
7898
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7899<div>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7900<h5>Syntax:</h5>
7901
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7902<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7903 integer bit width. Not all targets support all bit widths however.</p>
7904
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7905<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7906 declare i8 @llvm.atomic.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;val&gt;)
7907 declare i16 @llvm.atomic.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;val&gt;)
7908 declare i32 @llvm.atomic.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;val&gt;)
7909 declare i64 @llvm.atomic.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;val&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7910</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7911
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7912<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7913<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7914 the value from memory. It then stores the value in <tt>val</tt> in the memory
7915 at <tt>ptr</tt>.</p>
7916
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7917<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7918<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7919 the <tt>val</tt> argument and the result must be integers of the same bit
7920 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7921 integer type. The targets may only lower integer representations they
7922 support.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7923
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7924<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7925<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7926 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7927 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7928
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7929<h5>Examples:</h5>
7930<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]7931%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7932%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7933 store i32 4, %ptr
7934
7935%val1 = add i32 4, 4
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7936%result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7937 <i>; yields {i32}:result1 = 4</i>
7938%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7939%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7940
7941%val2 = add i32 1, 1
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7942%result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7943 <i>; yields {i32}:result2 = 8</i>
7944
7945%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7946%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7947</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7948
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7949</div>
7950
7951<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7952<h4>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]7953 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7954</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7955
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7956<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7957
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7958<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7959<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7960 any integer bit width. Not all targets support all bit widths however.</p>
7961
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7962<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7963 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7964 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7965 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7966 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7967</pre>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7968
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7969<h5>Overview:</h5>
7970<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7971 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7972
7973<h5>Arguments:</h5>
7974<p>The intrinsic takes two arguments, the first a pointer to an integer value
7975 and the second an integer value. The result is also an integer value. These
7976 integer types can have any bit width, but they must all have the same bit
7977 width. The targets may only lower integer representations they support.</p>
7978
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7979<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7980<p>This intrinsic does a series of operations atomically. It first loads the
7981 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7982 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7983
7984<h5>Examples:</h5>
7985<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]7986%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7987%ptr = bitcast i8* %mallocP to i32*
7988 store i32 4, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7989%result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7990 <i>; yields {i32}:result1 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7991%result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7992 <i>; yields {i32}:result2 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7993%result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7994 <i>; yields {i32}:result3 = 10</i>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]7995%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7996</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7997
Andrew Lenharthab0b9492008-02-21 06:45:13 +0000[diff] [blame]7998</div>
7999
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8000<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8001<h4>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8002 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8003</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8004
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8005<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8006
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8007<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8008<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
8009 any integer bit width and for different address spaces. Not all targets
8010 support all bit widths however.</p>
8011
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8012<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8013 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8014 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8015 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8016 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8017</pre>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8018
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8019<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]8020<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8021 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
8022
8023<h5>Arguments:</h5>
8024<p>The intrinsic takes two arguments, the first a pointer to an integer value
8025 and the second an integer value. The result is also an integer value. These
8026 integer types can have any bit width, but they must all have the same bit
8027 width. The targets may only lower integer representations they support.</p>
8028
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8029<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8030<p>This intrinsic does a series of operations atomically. It first loads the
8031 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
8032 result to <tt>ptr</tt>. It yields the original value stored
8033 at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8034
8035<h5>Examples:</h5>
8036<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]8037%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8038%ptr = bitcast i8* %mallocP to i32*
8039 store i32 8, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8040%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8041 <i>; yields {i32}:result1 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8042%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8043 <i>; yields {i32}:result2 = 4</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8044%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8045 <i>; yields {i32}:result3 = 2</i>
8046%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
8047</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8048
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8049</div>
8050
8051<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8052<h4>
8053 <a name="int_atomic_load_and">
8054 '<tt>llvm.atomic.load.and.*</tt>' Intrinsic
8055 </a>
8056 <br>
8057 <a name="int_atomic_load_nand">
8058 '<tt>llvm.atomic.load.nand.*</tt>' Intrinsic
8059 </a>
8060 <br>
8061 <a name="int_atomic_load_or">
8062 '<tt>llvm.atomic.load.or.*</tt>' Intrinsic
8063 </a>
8064 <br>
8065 <a name="int_atomic_load_xor">
8066 '<tt>llvm.atomic.load.xor.*</tt>' Intrinsic
8067 </a>
8068</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8069
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8070<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8071
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8072<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8073<p>These are overloaded intrinsics. You can
8074 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
8075 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
8076 bit width and for different address spaces. Not all targets support all bit
8077 widths however.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8078
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8079<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8080 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8081 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8082 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8083 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8084</pre>
8085
8086<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8087 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8088 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8089 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8090 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8091</pre>
8092
8093<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8094 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8095 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8096 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8097 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8098</pre>
8099
8100<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8101 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8102 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8103 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8104 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8105</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8106
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8107<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8108<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
8109 the value stored in memory at <tt>ptr</tt>. It yields the original value
8110 at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8111
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8112<h5>Arguments:</h5>
8113<p>These intrinsics take two arguments, the first a pointer to an integer value
8114 and the second an integer value. The result is also an integer value. These
8115 integer types can have any bit width, but they must all have the same bit
8116 width. The targets may only lower integer representations they support.</p>
8117
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8118<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8119<p>These intrinsics does a series of operations atomically. They first load the
8120 value stored at <tt>ptr</tt>. They then do the bitwise
8121 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
8122 original value stored at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8123
8124<h5>Examples:</h5>
8125<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]8126%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8127%ptr = bitcast i8* %mallocP to i32*
8128 store i32 0x0F0F, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8129%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8130 <i>; yields {i32}:result0 = 0x0F0F</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8131%result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8132 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8133%result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8134 <i>; yields {i32}:result2 = 0xF0</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8135%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8136 <i>; yields {i32}:result3 = FF</i>
8137%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
8138</pre>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8139
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8140</div>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8141
8142<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8143<h4>
8144 <a name="int_atomic_load_max">
8145 '<tt>llvm.atomic.load.max.*</tt>' Intrinsic
8146 </a>
8147 <br>
8148 <a name="int_atomic_load_min">
8149 '<tt>llvm.atomic.load.min.*</tt>' Intrinsic
8150 </a>
8151 <br>
8152 <a name="int_atomic_load_umax">
8153 '<tt>llvm.atomic.load.umax.*</tt>' Intrinsic
8154 </a>
8155 <br>
8156 <a name="int_atomic_load_umin">
8157 '<tt>llvm.atomic.load.umin.*</tt>' Intrinsic
8158 </a>
8159</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8160
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8161<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8162
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8163<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8164<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
8165 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
8166 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
8167 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8168
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8169<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8170 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8171 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8172 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8173 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8174</pre>
8175
8176<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8177 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8178 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8179 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8180 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8181</pre>
8182
8183<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8184 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8185 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8186 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8187 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8188</pre>
8189
8190<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8191 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
8192 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
8193 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
8194 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8195</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8196
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8197<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]8198<p>These intrinsics takes the signed or unsigned minimum or maximum of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8199 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
8200 original value at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8201
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8202<h5>Arguments:</h5>
8203<p>These intrinsics take two arguments, the first a pointer to an integer value
8204 and the second an integer value. The result is also an integer value. These
8205 integer types can have any bit width, but they must all have the same bit
8206 width. The targets may only lower integer representations they support.</p>
8207
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8208<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8209<p>These intrinsics does a series of operations atomically. They first load the
8210 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
8211 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
8212 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8213
8214<h5>Examples:</h5>
8215<pre>
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]8216%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
8217%ptr = bitcast i8* %mallocP to i32*
8218 store i32 7, %ptr
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8219%result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8220 <i>; yields {i32}:result0 = 7</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8221%result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8222 <i>; yields {i32}:result1 = -2</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8223%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8224 <i>; yields {i32}:result2 = 8</i>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8225%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8226 <i>; yields {i32}:result3 = 8</i>
8227%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
8228</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8229
Mon P Wang28873102008-06-25 08:15:39 +0000[diff] [blame]8230</div>
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]8231
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8232</div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8233
8234<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8235<h3>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8236 <a name="int_memorymarkers">Memory Use Markers</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8237</h3>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8238
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8239<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8240
8241<p>This class of intrinsics exists to information about the lifetime of memory
8242 objects and ranges where variables are immutable.</p>
8243
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8244<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8245<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8246 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8247</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8248
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8249<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8250
8251<h5>Syntax:</h5>
8252<pre>
8253 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8254</pre>
8255
8256<h5>Overview:</h5>
8257<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
8258 object's lifetime.</p>
8259
8260<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8261<p>The first argument is a constant integer representing the size of the
8262 object, or -1 if it is variable sized. The second argument is a pointer to
8263 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8264
8265<h5>Semantics:</h5>
8266<p>This intrinsic indicates that before this point in the code, the value of the
8267 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
Nick Lewycky8d336592009-10-27 16:56:58 +0000[diff] [blame]8268 never be used and has an undefined value. A load from the pointer that
8269 precedes this intrinsic can be replaced with
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8270 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
8271
8272</div>
8273
8274<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8275<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8276 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8277</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8278
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8279<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8280
8281<h5>Syntax:</h5>
8282<pre>
8283 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8284</pre>
8285
8286<h5>Overview:</h5>
8287<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
8288 object's lifetime.</p>
8289
8290<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8291<p>The first argument is a constant integer representing the size of the
8292 object, or -1 if it is variable sized. The second argument is a pointer to
8293 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8294
8295<h5>Semantics:</h5>
8296<p>This intrinsic indicates that after this point in the code, the value of the
8297 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8298 never be used and has an undefined value. Any stores into the memory object
8299 following this intrinsic may be removed as dead.
8300
8301</div>
8302
8303<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8304<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8305 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8306</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8307
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8308<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8309
8310<h5>Syntax:</h5>
8311<pre>
Nick Lewycky29b6cb42010-11-30 04:13:41 +0000[diff] [blame]8312 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8313</pre>
8314
8315<h5>Overview:</h5>
8316<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
8317 a memory object will not change.</p>
8318
8319<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8320<p>The first argument is a constant integer representing the size of the
8321 object, or -1 if it is variable sized. The second argument is a pointer to
8322 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8323
8324<h5>Semantics:</h5>
8325<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
8326 the return value, the referenced memory location is constant and
8327 unchanging.</p>
8328
8329</div>
8330
8331<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8332<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8333 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8334</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8335
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8336<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8337
8338<h5>Syntax:</h5>
8339<pre>
8340 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8341</pre>
8342
8343<h5>Overview:</h5>
8344<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8345 a memory object are mutable.</p>
8346
8347<h5>Arguments:</h5>
8348<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8349 The second argument is a constant integer representing the size of the
8350 object, or -1 if it is variable sized and the third argument is a pointer
8351 to the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8352
8353<h5>Semantics:</h5>
8354<p>This intrinsic indicates that the memory is mutable again.</p>
8355
8356</div>
8357
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8358</div>
8359
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]8360<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8361<h3>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8362 <a name="int_general">General Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8363</h3>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8364
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8365<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8366
8367<p>This class of intrinsics is designed to be generic and has no specific
8368 purpose.</p>
8369
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8370<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8371<h4>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8372 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8373</h4>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8374
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8375<div>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8376
8377<h5>Syntax:</h5>
8378<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8379 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8380</pre>
8381
8382<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8383<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8384
8385<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8386<p>The first argument is a pointer to a value, the second is a pointer to a
8387 global string, the third is a pointer to a global string which is the source
8388 file name, and the last argument is the line number.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8389
8390<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8391<p>This intrinsic allows annotation of local variables with arbitrary strings.
8392 This can be useful for special purpose optimizations that want to look for
8393 these annotations. These have no other defined use, they are ignored by code
8394 generation and optimization.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8395
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8396</div>
8397
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8398<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8399<h4>
Tanya Lattnere1a8da02007-09-21 23:57:59 +0000[diff] [blame]8400 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8401</h4>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8402
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8403<div>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8404
8405<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8406<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8407 any integer bit width.</p>
8408
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8409<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8410 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8411 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8412 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8413 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8414 declare i256 @llvm.annotation.i256(i256 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8415</pre>
8416
8417<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8418<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8419
8420<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8421<p>The first argument is an integer value (result of some expression), the
8422 second is a pointer to a global string, the third is a pointer to a global
8423 string which is the source file name, and the last argument is the line
8424 number. It returns the value of the first argument.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8425
8426<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8427<p>This intrinsic allows annotations to be put on arbitrary expressions with
8428 arbitrary strings. This can be useful for special purpose optimizations that
8429 want to look for these annotations. These have no other defined use, they
8430 are ignored by code generation and optimization.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8431
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8432</div>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8433
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8434<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8435<h4>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8436 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8437</h4>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8438
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8439<div>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8440
8441<h5>Syntax:</h5>
8442<pre>
8443 declare void @llvm.trap()
8444</pre>
8445
8446<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8447<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8448
8449<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8450<p>None.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8451
8452<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8453<p>This intrinsics is lowered to the target dependent trap instruction. If the
8454 target does not have a trap instruction, this intrinsic will be lowered to
8455 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8456
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8457</div>
8458
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8459<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8460<h4>
Misha Brukmandccb0252008-11-22 23:55:29 +0000[diff] [blame]8461 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8462</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8463
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8464<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8465
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8466<h5>Syntax:</h5>
8467<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8468 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8469</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8470
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8471<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8472<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8473 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8474 ensure that it is placed on the stack before local variables.</p>
8475
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8476<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8477<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8478 arguments. The first argument is the value loaded from the stack
8479 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8480 that has enough space to hold the value of the guard.</p>
8481
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8482<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8483<p>This intrinsic causes the prologue/epilogue inserter to force the position of
8484 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8485 stack. This is to ensure that if a local variable on the stack is
8486 overwritten, it will destroy the value of the guard. When the function exits,
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8487 the guard on the stack is checked against the original guard. If they are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8488 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8489 function.</p>
8490
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8491</div>
8492
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8493<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8494<h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8495 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8496</h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8497
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8498<div>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8499
8500<h5>Syntax:</h5>
8501<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8502 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;type&gt;)
8503 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;type&gt;)
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8504</pre>
8505
8506<h5>Overview:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8507<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8508 the optimizers to determine at compile time whether a) an operation (like
8509 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8510 runtime check for overflow isn't necessary. An object in this context means
8511 an allocation of a specific class, structure, array, or other object.</p>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8512
8513<h5>Arguments:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8514<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher8295a0a2009-12-23 00:29:49 +0000[diff] [blame]8515 argument is a pointer to or into the <tt>object</tt>. The second argument
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8516 is a boolean 0 or 1. This argument determines whether you want the
8517 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
Eric Christopher8295a0a2009-12-23 00:29:49 +0000[diff] [blame]8518 1, variables are not allowed.</p>
8519
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8520<h5>Semantics:</h5>
8521<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8522 representing the size of the object concerned, or <tt>i32/i64 -1 or 0</tt>,
8523 depending on the <tt>type</tt> argument, if the size cannot be determined at
8524 compile time.</p>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8525
8526</div>
8527
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8528</div>
8529
8530</div>
8531
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]8532<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]8533<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8534<address>
8535 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]8536 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8537 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]8538 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8539
8540 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
NAKAMURA Takumib9a33632011-04-09 02:13:37 +0000[diff] [blame]8541 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8542 Last modified: $Date$
8543</address>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +0000[diff] [blame]8544
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]8545</body>
8546</html>