Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / d5118c8f78a05ad0b426b6032138d1d934b77c8d / . / docs / LangRef.html
blob: 0ea83dd8c2f84d5cff7efade72c973a4132b8705 [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.">
Daniel Dunbaradea4972012-04-19 20:20:34 +0000[diff] [blame]10 <link rel="stylesheet" href="_static/llvm.css" type="text/css">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]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>
Bill Wendlingf7f06102011-10-11 06:41:28 +0000[diff] [blame]38 <li><a href="#linkage_external">'<tt>external</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]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 Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]95 <li><a href="#poisonvalues">Poison 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>
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]103 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a>
104 <ol>
105 <li><a href="#tbaa">'<tt>tbaa</tt>' Metadata</a></li>
Duncan Sands5e5c5f82012-04-14 12:36:06 +0000[diff] [blame]106 <li><a href="#fpmath">'<tt>fpmath</tt>' Metadata</a></li>
Rafael Espindola39dd3282012-03-24 00:14:51 +0000[diff] [blame]107 <li><a href="#range">'<tt>range</tt>' Metadata</a></li>
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]108 </ol>
109 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]110 </ol>
111 </li>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]112 <li><a href="#module_flags">Module Flags Metadata</a>
113 <ol>
Bill Wendlingf7b367c2012-02-16 01:10:50 +0000[diff] [blame]114 <li><a href="#objc_gc_flags">Objective-C Garbage Collection Module Flags Metadata</a></li>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]115 </ol>
116 </li>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]117 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
118 <ol>
119 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]120 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
121 Global Variable</a></li>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]122 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
123 Global Variable</a></li>
124 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
125 Global Variable</a></li>
126 </ol>
127 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]128 <li><a href="#instref">Instruction Reference</a>
129 <ol>
130 <li><a href="#terminators">Terminator Instructions</a>
131 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]132 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
133 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]134 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]135 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]136 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]137 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]138 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]139 </ol>
140 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]141 <li><a href="#binaryops">Binary Operations</a>
142 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]143 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]144 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]145 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]146 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]147 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]148 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]149 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
150 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
151 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]152 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
153 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
154 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]155 </ol>
156 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]157 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
158 <ol>
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]159 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
160 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
161 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]162 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]163 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]164 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]165 </ol>
166 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]167 <li><a href="#vectorops">Vector Operations</a>
168 <ol>
169 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
170 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
171 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]172 </ol>
173 </li>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]174 <li><a href="#aggregateops">Aggregate Operations</a>
175 <ol>
176 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
177 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
178 </ol>
179 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]180 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]181 <ol>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]182 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
183 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
184 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
185 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
186 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
187 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]188 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]189 </ol>
190 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]191 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]192 <ol>
193 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
194 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
195 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
196 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
197 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]198 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
199 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
200 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
201 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]202 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
203 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]204 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]205 </ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]206 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]207 <li><a href="#otherops">Other Operations</a>
208 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]209 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
210 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]211 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]212 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]213 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]214 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]215 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]216 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]217 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]218 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]219 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]220 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]221 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]222 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
223 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]224 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
225 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
226 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]227 </ol>
228 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]229 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
230 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]231 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
232 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
233 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]234 </ol>
235 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]236 <li><a href="#int_codegen">Code Generator Intrinsics</a>
237 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]238 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
239 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
240 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
241 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
242 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
243 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohman31f1af12010-05-26 21:56:15 +0000[diff] [blame]244 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000[diff] [blame]245 </ol>
246 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]247 <li><a href="#int_libc">Standard C Library Intrinsics</a>
248 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]249 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
250 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
251 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
252 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
253 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]254 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
255 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
256 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohman08b280b2011-05-27 00:36:31 +0000[diff] [blame]257 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
258 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]259 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
Peter Collingbourne168a4c32012-07-03 12:25:40 +0000[diff] [blame]260 <li><a href="#int_fabs">'<tt>llvm.fabs.*</tt>' Intrinsic</a></li>
Dan Gohman27db99f2012-07-26 17:43:27 +0000[diff] [blame]261 <li><a href="#int_floor">'<tt>llvm.floor.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]262 </ol>
263 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]264 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]265 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]266 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]267 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
268 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
269 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]270 </ol>
271 </li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]272 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
273 <ol>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]274 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
275 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
276 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
277 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
278 <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]279 <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]280 </ol>
281 </li>
Lang Hames5afba6f2012-06-05 19:07:46 +0000[diff] [blame]282 <li><a href="#spec_arithmetic">Specialised Arithmetic Intrinsics</a>
283 <ol>
284 <li><a href="#fmuladd">'<tt>llvm.fmuladd</tt> Intrinsic</a></li>
285 </ol>
286 </li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]287 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
288 <ol>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]289 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
290 <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]291 </ol>
292 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]293 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]294 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]295 <li><a href="#int_trampoline">Trampoline Intrinsics</a>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]296 <ol>
297 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]298 <li><a href="#int_at">'<tt>llvm.adjust.trampoline</tt>' Intrinsic</a></li>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]299 </ol>
300 </li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]301 <li><a href="#int_memorymarkers">Memory Use Markers</a>
302 <ol>
Jakub Staszak8e1b12a2011-12-04 20:44:25 +0000[diff] [blame]303 <li><a href="#int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a></li>
304 <li><a href="#int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a></li>
305 <li><a href="#int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a></li>
306 <li><a href="#int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a></li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]307 </ol>
308 </li>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame]309 <li><a href="#int_general">General intrinsics</a>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]310 <ol>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame]311 <li><a href="#int_var_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]312 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]313 <li><a href="#int_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]314 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]315 <li><a href="#int_trap">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]316 '<tt>llvm.trap</tt>' Intrinsic</a></li>
Dan Gohmana6063c62012-05-14 18:58:10 +0000[diff] [blame]317 <li><a href="#int_debugtrap">
318 '<tt>llvm.debugtrap</tt>' Intrinsic</a></li>
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]319 <li><a href="#int_stackprotector">
320 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Nuno Lopesb0c76d72012-07-05 17:37:07 +0000[diff] [blame]321 <li><a href="#int_objectsize">
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]322 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Nuno Lopesb0c76d72012-07-05 17:37:07 +0000[diff] [blame]323 <li><a href="#int_expect">
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]324 '<tt>llvm.expect</tt>' Intrinsic</a></li>
Nuno Lopesb0c76d72012-07-05 17:37:07 +0000[diff] [blame]325 <li><a href="#int_donothing">
326 '<tt>llvm.donothing</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]327 </ol>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]328 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]329 </ol>
330 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]331</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]332
333<div class="doc_author">
334 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
335 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]336</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]337
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]338<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]339<h2><a name="abstract">Abstract</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]340<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]341
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]342<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]343
344<p>This document is a reference manual for the LLVM assembly language. LLVM is
345 a Static Single Assignment (SSA) based representation that provides type
346 safety, low-level operations, flexibility, and the capability of representing
347 'all' high-level languages cleanly. It is the common code representation
348 used throughout all phases of the LLVM compilation strategy.</p>
349
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]350</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]351
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]352<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]353<h2><a name="introduction">Introduction</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]354<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]355
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]356<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]357
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]358<p>The LLVM code representation is designed to be used in three different forms:
359 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
360 for fast loading by a Just-In-Time compiler), and as a human readable
361 assembly language representation. This allows LLVM to provide a powerful
362 intermediate representation for efficient compiler transformations and
363 analysis, while providing a natural means to debug and visualize the
364 transformations. The three different forms of LLVM are all equivalent. This
365 document describes the human readable representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]366
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]367<p>The LLVM representation aims to be light-weight and low-level while being
368 expressive, typed, and extensible at the same time. It aims to be a
369 "universal IR" of sorts, by being at a low enough level that high-level ideas
370 may be cleanly mapped to it (similar to how microprocessors are "universal
371 IR's", allowing many source languages to be mapped to them). By providing
372 type information, LLVM can be used as the target of optimizations: for
373 example, through pointer analysis, it can be proven that a C automatic
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]374 variable is never accessed outside of the current function, allowing it to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]375 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]376
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]377<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]378<h4>
379 <a name="wellformed">Well-Formedness</a>
380</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]381
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]382<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]383
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]384<p>It is important to note that this document describes 'well formed' LLVM
385 assembly language. There is a difference between what the parser accepts and
386 what is considered 'well formed'. For example, the following instruction is
387 syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]388
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]389<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]390%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]391</pre>
392
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]393<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
394 LLVM infrastructure provides a verification pass that may be used to verify
395 that an LLVM module is well formed. This pass is automatically run by the
396 parser after parsing input assembly and by the optimizer before it outputs
397 bitcode. The violations pointed out by the verifier pass indicate bugs in
398 transformation passes or input to the parser.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]399
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]400</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]401
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]402</div>
403
Chris Lattnercc689392007-10-03 17:34:29 +0000[diff] [blame]404<!-- Describe the typesetting conventions here. -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]405
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]406<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]407<h2><a name="identifiers">Identifiers</a></h2>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]408<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]409
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]410<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]411
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]412<p>LLVM identifiers come in two basic types: global and local. Global
413 identifiers (functions, global variables) begin with the <tt>'@'</tt>
414 character. Local identifiers (register names, types) begin with
415 the <tt>'%'</tt> character. Additionally, there are three different formats
416 for identifiers, for different purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]417
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]418<ol>
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]419 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]420 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
421 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
422 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
423 other characters in their names can be surrounded with quotes. Special
424 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
425 ASCII code for the character in hexadecimal. In this way, any character
426 can be used in a name value, even quotes themselves.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]427
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]428 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]429 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]430
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]431 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]432 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]433</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]434
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]435<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]436 don't need to worry about name clashes with reserved words, and the set of
437 reserved words may be expanded in the future without penalty. Additionally,
438 unnamed identifiers allow a compiler to quickly come up with a temporary
439 variable without having to avoid symbol table conflicts.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]440
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]441<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]442 languages. There are keywords for different opcodes
443 ('<tt><a href="#i_add">add</a></tt>',
444 '<tt><a href="#i_bitcast">bitcast</a></tt>',
445 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
446 ('<tt><a href="#t_void">void</a></tt>',
447 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
448 reserved words cannot conflict with variable names, because none of them
449 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]450
451<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]452 '<tt>%X</tt>' by 8:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]453
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]454<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]455
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]456<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]457%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]458</pre>
459
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]460<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]461
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]462<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]463%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]464</pre>
465
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]466<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]467
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]468<pre class="doc_code">
Gabor Greifec58f752009-10-28 13:05:07 +0000[diff] [blame]469%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
470%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]471%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]472</pre>
473
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]474<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
475 lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]476
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]477<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]478 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]479 line.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]480
481 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]482 assigned to a named value.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]483
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]484 <li>Unnamed temporaries are numbered sequentially</li>
485</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]486
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]487<p>It also shows a convention that we follow in this document. When
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]488 demonstrating instructions, we will follow an instruction with a comment that
489 defines the type and name of value produced. Comments are shown in italic
490 text.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]491
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]492</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]493
494<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]495<h2><a name="highlevel">High Level Structure</a></h2>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]496<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]497<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]498<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]499<h3>
500 <a name="modulestructure">Module Structure</a>
501</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]502
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]503<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]504
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]505<p>LLVM programs are composed of <tt>Module</tt>s, each of which is a
506 translation unit of the input programs. Each module consists of functions,
507 global variables, and symbol table entries. Modules may be combined together
508 with the LLVM linker, which merges function (and global variable)
509 definitions, resolves forward declarations, and merges symbol table
510 entries. Here is an example of the "hello world" module:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]511
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]512<pre class="doc_code">
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]513<i>; Declare the string constant as a global constant.</i>&nbsp;
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]514<a href="#identifiers">@.str</a> = <a href="#linkage_private">private</a>&nbsp;<a href="#globalvars">unnamed_addr</a>&nbsp;<a href="#globalvars">constant</a>&nbsp;<a href="#t_array">[13 x i8]</a> c"hello world\0A\00"&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]515
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]516<i>; External declaration of the puts function</i>&nbsp;
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]517<a href="#functionstructure">declare</a> i32 @puts(i8* <a href="#nocapture">nocapture</a>) <a href="#fnattrs">nounwind</a>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]518
519<i>; Definition of main function</i>
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]520define i32 @main() { <i>; i32()* </i>&nbsp;
521 <i>; Convert [13 x i8]* to i8 *...</i>&nbsp;
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]522 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.str, i64 0, i64 0
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]523
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]524 <i>; Call puts function to write out the string to stdout.</i>&nbsp;
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]525 <a href="#i_call">call</a> i32 @puts(i8* %cast210)
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]526 <a href="#i_ret">ret</a> i32 0&nbsp;
527}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]528
529<i>; Named metadata</i>
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]530!1 = metadata !{i32 42}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]531!foo = !{!1, null}
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]532</pre>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]533
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]534<p>This example is made up of a <a href="#globalvars">global variable</a> named
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]535 "<tt>.str</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]536 a <a href="#functionstructure">function definition</a> for
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]537 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]538 "<tt>foo</tt>".</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]539
Bill Wendling4cc2be62012-03-14 08:07:43 +0000[diff] [blame]540<p>In general, a module is made up of a list of global values (where both
541 functions and global variables are global values). Global values are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]542 represented by a pointer to a memory location (in this case, a pointer to an
543 array of char, and a pointer to a function), and have one of the
544 following <a href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]545
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]546</div>
547
548<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]549<h3>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]550 <a name="linkage">Linkage Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]551</h3>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]552
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]553<div>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]554
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]555<p>All Global Variables and Functions have one of the following types of
556 linkage:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]557
558<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]559 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]560 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
561 by objects in the current module. In particular, linking code into a
562 module with an private global value may cause the private to be renamed as
563 necessary to avoid collisions. Because the symbol is private to the
564 module, all references can be updated. This doesn't show up in any symbol
565 table in the object file.</dd>
Rafael Espindolabb46f522009-01-15 20:18:42 +0000[diff] [blame]566
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]567 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]568 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
569 assembler and evaluated by the linker. Unlike normal strong symbols, they
570 are removed by the linker from the final linked image (executable or
571 dynamic library).</dd>
572
573 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
574 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
575 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
576 linker. The symbols are removed by the linker from the final linked image
577 (executable or dynamic library).</dd>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000[diff] [blame]578
Bill Wendling55ae5152010-08-20 22:05:50 +0000[diff] [blame]579 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
580 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
581 of the object is not taken. For instance, functions that had an inline
582 definition, but the compiler decided not to inline it. Note,
583 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
584 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
585 visibility. The symbols are removed by the linker from the final linked
586 image (executable or dynamic library).</dd>
587
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]588 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling07d31772010-06-29 22:34:52 +0000[diff] [blame]589 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]590 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
591 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]592
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]593 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000[diff] [blame]594 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]595 into the object file corresponding to the LLVM module. They exist to
596 allow inlining and other optimizations to take place given knowledge of
597 the definition of the global, which is known to be somewhere outside the
598 module. Globals with <tt>available_externally</tt> linkage are allowed to
599 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
600 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000[diff] [blame]601
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]602 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattner4887bd82007-01-14 06:51:48 +0000[diff] [blame]603 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner873187c2010-01-09 19:15:14 +0000[diff] [blame]604 the same name when linkage occurs. This can be used to implement
605 some forms of inline functions, templates, or other code which must be
606 generated in each translation unit that uses it, but where the body may
607 be overridden with a more definitive definition later. Unreferenced
608 <tt>linkonce</tt> globals are allowed to be discarded. Note that
609 <tt>linkonce</tt> linkage does not actually allow the optimizer to
610 inline the body of this function into callers because it doesn't know if
611 this definition of the function is the definitive definition within the
612 program or whether it will be overridden by a stronger definition.
613 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
614 linkage.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]615
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]616 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]617 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
618 <tt>linkonce</tt> linkage, except that unreferenced globals with
619 <tt>weak</tt> linkage may not be discarded. This is used for globals that
620 are declared "weak" in C source code.</dd>
621
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]622 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]623 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
624 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
625 global scope.
626 Symbols with "<tt>common</tt>" linkage are merged in the same way as
627 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000[diff] [blame]628 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]629 must have a zero initializer, and may not be marked '<a
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000[diff] [blame]630 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
631 have common linkage.</dd>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]632
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_appending">appending</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]635 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]636 pointer to array type. When two global variables with appending linkage
637 are linked together, the two global arrays are appended together. This is
638 the LLVM, typesafe, equivalent of having the system linker append together
639 "sections" with identical names when .o files are linked.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]640
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]641 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]642 <dd>The semantics of this linkage follow the ELF object file model: the symbol
643 is weak until linked, if not linked, the symbol becomes null instead of
644 being an undefined reference.</dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]645
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]646 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
647 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]648 <dd>Some languages allow differing globals to be merged, such as two functions
649 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]650 that only equivalent globals are ever merged (the "one definition rule"
651 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]652 and <tt>weak_odr</tt> linkage types to indicate that the global will only
653 be merged with equivalent globals. These linkage types are otherwise the
654 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]655
Bill Wendling5c3a9f72011-11-04 20:40:41 +0000[diff] [blame]656 <dt><tt><b><a name="linkage_external">external</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]657 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]658 visible, meaning that it participates in linkage and can be used to
659 resolve external symbol references.</dd>
Reid Spencerc8910842007-04-11 23:49:50 +0000[diff] [blame]660</dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]661
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]662<p>The next two types of linkage are targeted for Microsoft Windows platform
663 only. They are designed to support importing (exporting) symbols from (to)
664 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]665
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]666<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]667 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]668 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]669 or variable via a global pointer to a pointer that is set up by the DLL
670 exporting the symbol. On Microsoft Windows targets, the pointer name is
671 formed by combining <code>__imp_</code> and the function or variable
672 name.</dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]673
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]674 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]675 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]676 pointer to a pointer in a DLL, so that it can be referenced with the
677 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
678 name is formed by combining <code>__imp_</code> and the function or
679 variable name.</dd>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]680</dl>
681
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]682<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
683 another module defined a "<tt>.LC0</tt>" variable and was linked with this
684 one, one of the two would be renamed, preventing a collision. Since
685 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
686 declarations), they are accessible outside of the current module.</p>
687
688<p>It is illegal for a function <i>declaration</i> to have any linkage type
Bill Wendlingf7f06102011-10-11 06:41:28 +0000[diff] [blame]689 other than <tt>external</tt>, <tt>dllimport</tt>
690 or <tt>extern_weak</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]691
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]692<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]693 or <tt>weak_odr</tt> linkages.</p>
694
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]695</div>
696
697<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]698<h3>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]699 <a name="callingconv">Calling Conventions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]700</h3>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]701
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]702<div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]703
704<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]705 and <a href="#i_invoke">invokes</a> can all have an optional calling
706 convention specified for the call. The calling convention of any pair of
707 dynamic caller/callee must match, or the behavior of the program is
708 undefined. The following calling conventions are supported by LLVM, and more
709 may be added in the future:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]710
711<dl>
712 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]713 <dd>This calling convention (the default if no other calling convention is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]714 specified) matches the target C calling conventions. This calling
715 convention supports varargs function calls and tolerates some mismatch in
716 the declared prototype and implemented declaration of the function (as
717 does normal C).</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]718
719 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]720 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]721 (e.g. by passing things in registers). This calling convention allows the
722 target to use whatever tricks it wants to produce fast code for the
723 target, without having to conform to an externally specified ABI
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]724 (Application Binary Interface).
725 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattner29689432010-03-11 00:22:57 +0000[diff] [blame]726 when this or the GHC convention is used.</a> This calling convention
727 does not support varargs and requires the prototype of all callees to
728 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]729
730 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]731 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]732 as possible under the assumption that the call is not commonly executed.
733 As such, these calls often preserve all registers so that the call does
734 not break any live ranges in the caller side. This calling convention
735 does not support varargs and requires the prototype of all callees to
736 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]737
Chris Lattner29689432010-03-11 00:22:57 +0000[diff] [blame]738 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
739 <dd>This calling convention has been implemented specifically for use by the
740 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
741 It passes everything in registers, going to extremes to achieve this by
742 disabling callee save registers. This calling convention should not be
743 used lightly but only for specific situations such as an alternative to
744 the <em>register pinning</em> performance technique often used when
745 implementing functional programming languages.At the moment only X86
746 supports this convention and it has the following limitations:
747 <ul>
748 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
749 floating point types are supported.</li>
750 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
751 6 floating point parameters.</li>
752 </ul>
753 This calling convention supports
754 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
755 requires both the caller and callee are using it.
756 </dd>
757
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]758 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]759 <dd>Any calling convention may be specified by number, allowing
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]760 target-specific calling conventions to be used. Target specific calling
761 conventions start at 64.</dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]762</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]763
764<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]765 support Pascal conventions or any other well-known target-independent
766 convention.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]767
768</div>
769
770<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]771<h3>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]772 <a name="visibility">Visibility Styles</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]773</h3>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]774
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]775<div>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]776
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]777<p>All Global Variables and Functions have one of the following visibility
778 styles:</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]779
780<dl>
781 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]782 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]783 that the declaration is visible to other modules and, in shared libraries,
784 means that the declared entity may be overridden. On Darwin, default
785 visibility means that the declaration is visible to other modules. Default
786 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]787
788 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]789 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]790 object if they are in the same shared object. Usually, hidden visibility
791 indicates that the symbol will not be placed into the dynamic symbol
792 table, so no other module (executable or shared library) can reference it
793 directly.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]794
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]795 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]796 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]797 the dynamic symbol table, but that references within the defining module
798 will bind to the local symbol. That is, the symbol cannot be overridden by
799 another module.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]800</dl>
801
802</div>
803
804<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]805<h3>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]806 <a name="namedtypes">Named Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]807</h3>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]808
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]809<div>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]810
811<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]812 it easier to read the IR and make the IR more condensed (particularly when
813 recursive types are involved). An example of a name specification is:</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]814
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]815<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]816%mytype = type { %mytype*, i32 }
817</pre>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]818
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]819<p>You may give a name to any <a href="#typesystem">type</a> except
Chris Lattnerdc65f222010-08-17 23:26:04 +0000[diff] [blame]820 "<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]821 is expected with the syntax "%mytype".</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]822
823<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]824 and that you can therefore specify multiple names for the same type. This
825 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
826 uses structural typing, the name is not part of the type. When printing out
827 LLVM IR, the printer will pick <em>one name</em> to render all types of a
828 particular shape. This means that if you have code where two different
829 source types end up having the same LLVM type, that the dumper will sometimes
830 print the "wrong" or unexpected type. This is an important design point and
831 isn't going to change.</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]832
833</div>
834
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]835<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]836<h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]837 <a name="globalvars">Global Variables</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]838</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]839
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]840<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]841
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]842<p>Global variables define regions of memory allocated at compilation time
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]843 instead of run-time. Global variables may optionally be initialized, may
844 have an explicit section to be placed in, and may have an optional explicit
Hans Wennborgce718ff2012-06-23 11:37:03 +0000[diff] [blame]845 alignment specified.</p>
846
847<p>A variable may be defined as <tt>thread_local</tt>, which
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]848 means that it will not be shared by threads (each thread will have a
Hans Wennborgce718ff2012-06-23 11:37:03 +0000[diff] [blame]849 separated copy of the variable). Not all targets support thread-local
850 variables. Optionally, a TLS model may be specified:</p>
851
852<dl>
853 <dt><b><tt>localdynamic</tt></b>:</dt>
854 <dd>For variables that are only used within the current shared library.</dd>
855
856 <dt><b><tt>initialexec</tt></b>:</dt>
857 <dd>For variables in modules that will not be loaded dynamically.</dd>
858
859 <dt><b><tt>localexec</tt></b>:</dt>
860 <dd>For variables defined in the executable and only used within it.</dd>
861</dl>
862
863<p>The models correspond to the ELF TLS models; see
864 <a href="http://people.redhat.com/drepper/tls.pdf">ELF
865 Handling For Thread-Local Storage</a> for more information on under which
866 circumstances the different models may be used. The target may choose a
867 different TLS model if the specified model is not supported, or if a better
868 choice of model can be made.</p>
869
870<p>A variable may be defined as a global
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]871 "constant," which indicates that the contents of the variable
872 will <b>never</b> be modified (enabling better optimization, allowing the
873 global data to be placed in the read-only section of an executable, etc).
874 Note that variables that need runtime initialization cannot be marked
875 "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]876
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]877<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
878 constant, even if the final definition of the global is not. This capability
879 can be used to enable slightly better optimization of the program, but
880 requires the language definition to guarantee that optimizations based on the
881 'constantness' are valid for the translation units that do not include the
882 definition.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]883
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]884<p>As SSA values, global variables define pointer values that are in scope
885 (i.e. they dominate) all basic blocks in the program. Global variables
886 always define a pointer to their "content" type because they describe a
887 region of memory, and all memory objects in LLVM are accessed through
888 pointers.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]889
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]890<p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
891 that the address is not significant, only the content. Constants marked
Rafael Espindolaa5eaa862011-01-15 08:20:57 +0000[diff] [blame]892 like this can be merged with other constants if they have the same
893 initializer. Note that a constant with significant address <em>can</em>
894 be merged with a <tt>unnamed_addr</tt> constant, the result being a
895 constant whose address is significant.</p>
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]896
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]897<p>A global variable may be declared to reside in a target-specific numbered
898 address space. For targets that support them, address spaces may affect how
899 optimizations are performed and/or what target instructions are used to
900 access the variable. The default address space is zero. The address space
901 qualifier must precede any other attributes.</p>
Christopher Lamb284d9922007-12-11 09:31:00 +0000[diff] [blame]902
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]903<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]904 supports it, it will emit globals to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]905
Chris Lattnerce99fa92010-04-28 00:13:42 +0000[diff] [blame]906<p>An explicit alignment may be specified for a global, which must be a power
907 of 2. If not present, or if the alignment is set to zero, the alignment of
908 the global is set by the target to whatever it feels convenient. If an
909 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner2d4b8ee2010-04-28 00:31:12 +0000[diff] [blame]910 alignment. Targets and optimizers are not allowed to over-align the global
911 if the global has an assigned section. In this case, the extra alignment
912 could be observable: for example, code could assume that the globals are
913 densely packed in their section and try to iterate over them as an array,
914 alignment padding would break this iteration.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]915
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]916<p>For example, the following defines a global in a numbered address space with
917 an initializer, section, and alignment:</p>
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]918
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]919<pre class="doc_code">
Dan Gohman398873c2009-01-11 00:40:00 +0000[diff] [blame]920@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]921</pre>
922
Hans Wennborgce718ff2012-06-23 11:37:03 +0000[diff] [blame]923<p>The following example defines a thread-local global with
924 the <tt>initialexec</tt> TLS model:</p>
925
926<pre class="doc_code">
927@G = thread_local(initialexec) global i32 0, align 4
928</pre>
929
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]930</div>
931
932
933<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]934<h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]935 <a name="functionstructure">Functions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]936</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]937
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]938<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]939
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]940<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]941 optional <a href="#linkage">linkage type</a>, an optional
942 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]943 <a href="#callingconv">calling convention</a>,
944 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]945 <a href="#paramattrs">parameter attribute</a> for the return type, a function
946 name, a (possibly empty) argument list (each with optional
947 <a href="#paramattrs">parameter attributes</a>), optional
948 <a href="#fnattrs">function attributes</a>, an optional section, an optional
949 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
950 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]951
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]952<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
953 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]954 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]955 <a href="#callingconv">calling convention</a>,
956 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]957 <a href="#paramattrs">parameter attribute</a> for the return type, a function
958 name, a possibly empty list of arguments, an optional alignment, and an
959 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]960
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]961<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]962 (Control Flow Graph) for the function. Each basic block may optionally start
963 with a label (giving the basic block a symbol table entry), contains a list
964 of instructions, and ends with a <a href="#terminators">terminator</a>
965 instruction (such as a branch or function return).</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]966
Chris Lattner4a3c9012007-06-08 16:52:14 +0000[diff] [blame]967<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]968 executed on entrance to the function, and it is not allowed to have
969 predecessor basic blocks (i.e. there can not be any branches to the entry
970 block of a function). Because the block can have no predecessors, it also
971 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]972
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]973<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]974 supports it, it will emit functions to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]975
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]976<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]977 the alignment is set to zero, the alignment of the function is set by the
978 target to whatever it feels convenient. If an explicit alignment is
979 specified, the function is forced to have at least that much alignment. All
980 alignments must be a power of 2.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]981
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]982<p>If the <tt>unnamed_addr</tt> attribute is given, the address is know to not
Bill Wendling5c3a9f72011-11-04 20:40:41 +0000[diff] [blame]983 be significant and two identical functions can be merged.</p>
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]984
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]985<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]986<pre class="doc_code">
Chris Lattner50ad45c2008-10-13 16:55:18 +0000[diff] [blame]987define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]988 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
989 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
990 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
991 [<a href="#gc">gc</a>] { ... }
992</pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]993
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]994</div>
995
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]996<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]997<h3>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]998 <a name="aliasstructure">Aliases</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]999</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1000
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1001<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1002
1003<p>Aliases act as "second name" for the aliasee value (which can be either
1004 function, global variable, another alias or bitcast of global value). Aliases
1005 may have an optional <a href="#linkage">linkage type</a>, and an
1006 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]1007
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1008<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1009<pre class="doc_code">
Duncan Sands0b23ac12008-09-12 20:48:21 +0000[diff] [blame]1010@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]1011</pre>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]1012
1013</div>
1014
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1015<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1016<h3>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1017 <a name="namedmetadatastructure">Named Metadata</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1018</h3>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1019
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1020<div>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1021
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]1022<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]1023 nodes</a> (but not metadata strings) are the only valid operands for
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]1024 a named metadata.</p>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1025
1026<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1027<pre class="doc_code">
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]1028; Some unnamed metadata nodes, which are referenced by the named metadata.
1029!0 = metadata !{metadata !"zero"}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1030!1 = metadata !{metadata !"one"}
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]1031!2 = metadata !{metadata !"two"}
Dan Gohman1005bc52010-07-13 19:48:13 +0000[diff] [blame]1032; A named metadata.
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]1033!name = !{!0, !1, !2}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1034</pre>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]1035
1036</div>
1037
1038<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1039<h3>
1040 <a name="paramattrs">Parameter Attributes</a>
1041</h3>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1042
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1043<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1044
1045<p>The return type and each parameter of a function type may have a set of
1046 <i>parameter attributes</i> associated with them. Parameter attributes are
1047 used to communicate additional information about the result or parameters of
1048 a function. Parameter attributes are considered to be part of the function,
1049 not of the function type, so functions with different parameter attributes
1050 can have the same function type.</p>
1051
1052<p>Parameter attributes are simple keywords that follow the type specified. If
1053 multiple parameter attributes are needed, they are space separated. For
1054 example:</p>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1055
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1056<pre class="doc_code">
Nick Lewyckyb6a7d252009-02-15 23:06:14 +0000[diff] [blame]1057declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattner66d922c2008-10-04 18:33:34 +0000[diff] [blame]1058declare i32 @atoi(i8 zeroext)
1059declare signext i8 @returns_signed_char()
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1060</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1061
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1062<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1063 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1064
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1065<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1066
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1067<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1068 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1069 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarichebe81732011-03-16 22:20:18 +0000[diff] [blame]1070 should be zero-extended to the extent required by the target's ABI (which
1071 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1072 parameter) or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1073
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1074 <dt><tt><b>signext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1075 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarich9e69ff92011-03-17 14:21:58 +0000[diff] [blame]1076 should be sign-extended to the extent required by the target's ABI (which
1077 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1078 return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1079
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1080 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1081 <dd>This indicates that this parameter or return value should be treated in a
1082 special target-dependent fashion during while emitting code for a function
1083 call or return (usually, by putting it in a register as opposed to memory,
1084 though some targets use it to distinguish between two different kinds of
1085 registers). Use of this attribute is target-specific.</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1086
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1087 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Chris Lattnera6fd81d2010-11-20 23:49:06 +0000[diff] [blame]1088 <dd><p>This indicates that the pointer parameter should really be passed by
1089 value to the function. The attribute implies that a hidden copy of the
1090 pointee
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1091 is made between the caller and the callee, so the callee is unable to
Chris Lattneref097052012-05-30 00:40:23 +0000[diff] [blame]1092 modify the value in the caller. This attribute is only valid on LLVM
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1093 pointer arguments. It is generally used to pass structs and arrays by
1094 value, but is also valid on pointers to scalars. The copy is considered
1095 to belong to the caller not the callee (for example,
1096 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1097 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnera6fd81d2010-11-20 23:49:06 +0000[diff] [blame]1098 values.</p>
1099
1100 <p>The byval attribute also supports specifying an alignment with
1101 the align attribute. It indicates the alignment of the stack slot to
1102 form and the known alignment of the pointer specified to the call site. If
1103 the alignment is not specified, then the code generator makes a
1104 target-specific assumption.</p></dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1105
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1106 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1107 <dd>This indicates that the pointer parameter specifies the address of a
1108 structure that is the return value of the function in the source program.
1109 This pointer must be guaranteed by the caller to be valid: loads and
1110 stores to the structure may be assumed by the callee to not to trap. This
1111 may only be applied to the first parameter. This is not a valid attribute
1112 for return values. </dd>
1113
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1114 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1115 <dd>This indicates that pointer values
1116 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmanefca7f92010-07-02 23:46:54 +0000[diff] [blame]1117 value do not alias pointer values which are not <i>based</i> on it,
1118 ignoring certain "irrelevant" dependencies.
1119 For a call to the parent function, dependencies between memory
1120 references from before or after the call and from those during the call
1121 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1122 return value used in that call.
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1123 The caller shares the responsibility with the callee for ensuring that
1124 these requirements are met.
1125 For further details, please see the discussion of the NoAlias response in
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]1126 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1127<br>
John McCall191d4ee2010-07-06 21:07:14 +0000[diff] [blame]1128 Note that this definition of <tt>noalias</tt> is intentionally
1129 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner211244a2010-07-06 20:51:35 +0000[diff] [blame]1130 arguments, though it is slightly weaker.
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]1131<br>
1132 For function return values, C99's <tt>restrict</tt> is not meaningful,
1133 while LLVM's <tt>noalias</tt> is.
1134 </dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1135
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1136 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1137 <dd>This indicates that the callee does not make any copies of the pointer
1138 that outlive the callee itself. This is not a valid attribute for return
1139 values.</dd>
1140
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1141 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1142 <dd>This indicates that the pointer parameter can be excised using the
1143 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1144 attribute for return values.</dd>
1145</dl>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1146
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1147</div>
1148
1149<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1150<h3>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1151 <a name="gc">Garbage Collector Names</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1152</h3>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1153
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1154<div>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1155
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1156<p>Each function may specify a garbage collector name, which is simply a
1157 string:</p>
1158
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1159<pre class="doc_code">
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1160define void @f() gc "name" { ... }
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1161</pre>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1162
1163<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1164 collector which will cause the compiler to alter its output in order to
1165 support the named garbage collection algorithm.</p>
1166
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1167</div>
1168
1169<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1170<h3>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1171 <a name="fnattrs">Function Attributes</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1172</h3>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1173
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1174<div>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1175
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1176<p>Function attributes are set to communicate additional information about a
1177 function. Function attributes are considered to be part of the function, not
1178 of the function type, so functions with different parameter attributes can
1179 have the same function type.</p>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1180
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1181<p>Function attributes are simple keywords that follow the type specified. If
1182 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1183
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1184<pre class="doc_code">
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1185define void @f() noinline { ... }
1186define void @f() alwaysinline { ... }
1187define void @f() alwaysinline optsize { ... }
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1188define void @f() optsize { ... }
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1189</pre>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1190
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1191<dl>
Kostya Serebryany164b86b2012-01-20 17:56:17 +0000[diff] [blame]1192 <dt><tt><b>address_safety</b></tt></dt>
1193 <dd>This attribute indicates that the address safety analysis
1194 is enabled for this function. </dd>
1195
Charles Davis1e063d12010-02-12 00:31:15 +0000[diff] [blame]1196 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1197 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1198 the backend should forcibly align the stack pointer. Specify the
1199 desired alignment, which must be a power of two, in parentheses.
1200
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1201 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1202 <dd>This attribute indicates that the inliner should attempt to inline this
1203 function into callers whenever possible, ignoring any active inlining size
1204 threshold for this caller.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1205
Dan Gohman129bd562011-06-16 16:03:13 +0000[diff] [blame]1206 <dt><tt><b>nonlazybind</b></tt></dt>
1207 <dd>This attribute suppresses lazy symbol binding for the function. This
1208 may make calls to the function faster, at the cost of extra program
1209 startup time if the function is not called during program startup.</dd>
1210
Chad Rosier249d6702012-08-10 00:00:22 +0000[diff] [blame]1211 <dt><tt><b>ia_nsdialect</b></tt></dt>
1212 <dd>This attribute indicates the associated inline assembly call is using a
1213 non-standard assembly dialect. The standard dialect is ATT, which is
1214 assumed when this attribute is not present. When present, the dialect
1215 is assumed to be Intel. Currently, ATT and Intel are the only supported
1216 dialects.</dd>
1217
Jakob Stoklund Olesen570a4a52010-02-06 01:16:28 +0000[diff] [blame]1218 <dt><tt><b>inlinehint</b></tt></dt>
1219 <dd>This attribute indicates that the source code contained a hint that inlining
1220 this function is desirable (such as the "inline" keyword in C/C++). It
1221 is just a hint; it imposes no requirements on the inliner.</dd>
1222
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1223 <dt><tt><b>naked</b></tt></dt>
1224 <dd>This attribute disables prologue / epilogue emission for the function.
1225 This can have very system-specific consequences.</dd>
1226
1227 <dt><tt><b>noimplicitfloat</b></tt></dt>
1228 <dd>This attributes disables implicit floating point instructions.</dd>
1229
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1230 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1231 <dd>This attribute indicates that the inliner should never inline this
1232 function in any situation. This attribute may not be used together with
1233 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1234
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1235 <dt><tt><b>noredzone</b></tt></dt>
1236 <dd>This attribute indicates that the code generator should not use a red
1237 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1238
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1239 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1240 <dd>This function attribute indicates that the function never returns
1241 normally. This produces undefined behavior at runtime if the function
1242 ever does dynamically return.</dd>
Bill Wendling31359ba2008-11-13 01:02:51 +0000[diff] [blame]1243
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1244 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1245 <dd>This function attribute indicates that the function never returns with an
1246 unwind or exceptional control flow. If the function does unwind, its
1247 runtime behavior is undefined.</dd>
Bill Wendlingfbaa7ed2008-11-26 19:07:40 +0000[diff] [blame]1248
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1249 <dt><tt><b>optsize</b></tt></dt>
1250 <dd>This attribute suggests that optimization passes and code generator passes
1251 make choices that keep the code size of this function low, and otherwise
1252 do optimizations specifically to reduce code size.</dd>
1253
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1254 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1255 <dd>This attribute indicates that the function computes its result (or decides
1256 to unwind an exception) based strictly on its arguments, without
1257 dereferencing any pointer arguments or otherwise accessing any mutable
1258 state (e.g. memory, control registers, etc) visible to caller functions.
1259 It does not write through any pointer arguments
1260 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1261 changes any state visible to callers. This means that it cannot unwind
Bill Wendling7b9e5392012-02-06 21:57:33 +0000[diff] [blame]1262 exceptions by calling the <tt>C++</tt> exception throwing methods.</dd>
Devang Patel5d96fda2009-06-12 19:45:19 +0000[diff] [blame]1263
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1264 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1265 <dd>This attribute indicates that the function does not write through any
1266 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1267 arguments) or otherwise modify any state (e.g. memory, control registers,
1268 etc) visible to caller functions. It may dereference pointer arguments
1269 and read state that may be set in the caller. A readonly function always
1270 returns the same value (or unwinds an exception identically) when called
1271 with the same set of arguments and global state. It cannot unwind an
Bill Wendling7b9e5392012-02-06 21:57:33 +0000[diff] [blame]1272 exception by calling the <tt>C++</tt> exception throwing methods.</dd>
Anton Korobeynikovc5ec8a72009-07-17 18:07:26 +0000[diff] [blame]1273
Bill Wendling9bd5d042011-12-05 21:27:54 +0000[diff] [blame]1274 <dt><tt><b><a name="returns_twice">returns_twice</a></b></tt></dt>
1275 <dd>This attribute indicates that this function can return twice. The
1276 C <code>setjmp</code> is an example of such a function. The compiler
1277 disables some optimizations (like tail calls) in the caller of these
1278 functions.</dd>
1279
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1280 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1281 <dd>This attribute indicates that the function should emit a stack smashing
1282 protector. It is in the form of a "canary"&mdash;a random value placed on
1283 the stack before the local variables that's checked upon return from the
1284 function to see if it has been overwritten. A heuristic is used to
1285 determine if a function needs stack protectors or not.<br>
1286<br>
1287 If a function that has an <tt>ssp</tt> attribute is inlined into a
1288 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1289 function will have an <tt>ssp</tt> attribute.</dd>
1290
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1291 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1292 <dd>This attribute indicates that the function should <em>always</em> emit a
1293 stack smashing protector. This overrides
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1294 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1295<br>
1296 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1297 function that doesn't have an <tt>sspreq</tt> attribute or which has
1298 an <tt>ssp</tt> attribute, then the resulting function will have
1299 an <tt>sspreq</tt> attribute.</dd>
Rafael Espindolafbff0ec2011-07-25 15:27:59 +0000[diff] [blame]1300
1301 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1302 <dd>This attribute indicates that the ABI being targeted requires that
1303 an unwind table entry be produce for this function even if we can
1304 show that no exceptions passes by it. This is normally the case for
1305 the ELF x86-64 abi, but it can be disabled for some compilation
1306 units.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1307</dl>
1308
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1309</div>
1310
1311<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1312<h3>
Chris Lattner1eeeb0c2006-04-08 04:40:53 +0000[diff] [blame]1313 <a name="moduleasm">Module-Level Inline Assembly</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1314</h3>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1315
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1316<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1317
1318<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1319 the GCC "file scope inline asm" blocks. These blocks are internally
1320 concatenated by LLVM and treated as a single unit, but may be separated in
1321 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1322
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1323<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1324module asm "inline asm code goes here"
1325module asm "more can go here"
1326</pre>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1327
1328<p>The strings can contain any character by escaping non-printable characters.
1329 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]1330 for the number.</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1331
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1332<p>The inline asm code is simply printed to the machine code .s file when
1333 assembly code is generated.</p>
1334
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1335</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1336
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1337<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1338<h3>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1339 <a name="datalayout">Data Layout</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1340</h3>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1341
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1342<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1343
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1344<p>A module may specify a target specific data layout string that specifies how
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1345 data is to be laid out in memory. The syntax for the data layout is
1346 simply:</p>
1347
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1348<pre class="doc_code">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1349target datalayout = "<i>layout specification</i>"
1350</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1351
1352<p>The <i>layout specification</i> consists of a list of specifications
1353 separated by the minus sign character ('-'). Each specification starts with
1354 a letter and may include other information after the letter to define some
1355 aspect of the data layout. The specifications accepted are as follows:</p>
1356
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1357<dl>
1358 <dt><tt>E</tt></dt>
1359 <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]1360 bits with the most significance have the lowest address location.</dd>
1361
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1362 <dt><tt>e</tt></dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]1363 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1364 the bits with the least significance have the lowest address
1365 location.</dd>
1366
Lang Hamesbb5b3f32011-10-10 23:42:08 +0000[diff] [blame]1367 <dt><tt>S<i>size</i></tt></dt>
1368 <dd>Specifies the natural alignment of the stack in bits. Alignment promotion
1369 of stack variables is limited to the natural stack alignment to avoid
1370 dynamic stack realignment. The stack alignment must be a multiple of
Lang Hames5f119a62011-10-11 17:50:14 +0000[diff] [blame]1371 8-bits. If omitted, the natural stack alignment defaults to "unspecified",
1372 which does not prevent any alignment promotions.</dd>
Lang Hamesbb5b3f32011-10-10 23:42:08 +0000[diff] [blame]1373
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1374 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1375 <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]1376 <i>preferred</i> alignments. All sizes are in bits. Specifying
1377 the <i>pref</i> alignment is optional. If omitted, the
1378 preceding <tt>:</tt> should be omitted too.</dd>
1379
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1380 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1381 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1382 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1383
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1384 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1385 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1386 <i>size</i>.</dd>
1387
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1388 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1389 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesen9d8d2212010-05-28 18:54:47 +0000[diff] [blame]1390 <i>size</i>. Only values of <i>size</i> that are supported by the target
1391 will work. 32 (float) and 64 (double) are supported on all targets;
1392 80 or 128 (different flavors of long double) are also supported on some
1393 targets.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1394
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1395 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1396 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1397 <i>size</i>.</dd>
1398
Daniel Dunbar87bde0b2009-06-08 22:17:53 +0000[diff] [blame]1399 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1400 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1401 <i>size</i>.</dd>
Chris Lattnere82bdc42009-11-07 09:35:34 +0000[diff] [blame]1402
1403 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1404 <dd>This specifies a set of native integer widths for the target CPU
1405 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1406 "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]1407 this set are considered to support most general arithmetic
Chris Lattnere82bdc42009-11-07 09:35:34 +0000[diff] [blame]1408 operations efficiently.</dd>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1409</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1410
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1411<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman1c70c002010-04-28 00:36:01 +0000[diff] [blame]1412 default set of specifications which are then (possibly) overridden by the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1413 specifications in the <tt>datalayout</tt> keyword. The default specifications
1414 are given in this list:</p>
1415
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1416<ul>
1417 <li><tt>E</tt> - big endian</li>
Dan Gohmanfdf2e8c2010-02-23 02:44:03 +0000[diff] [blame]1418 <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]1419 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1420 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1421 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1422 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]1423 <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]1424 alignment of 64-bits</li>
1425 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1426 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1427 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1428 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1429 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar87bde0b2009-06-08 22:17:53 +0000[diff] [blame]1430 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1431</ul>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1432
1433<p>When LLVM is determining the alignment for a given type, it uses the
1434 following rules:</p>
1435
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1436<ol>
1437 <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]1438 specification is used.</li>
1439
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1440 <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]1441 smallest integer type that is larger than the bitwidth of the sought type
1442 is used. If none of the specifications are larger than the bitwidth then
Sylvestre Ledruc8e41c52012-07-23 08:51:15 +0000[diff] [blame]1443 the largest integer type is used. For example, given the default
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1444 specifications above, the i7 type will use the alignment of i8 (next
1445 largest) while both i65 and i256 will use the alignment of i64 (largest
1446 specified).</li>
1447
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1448 <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]1449 largest vector type that is smaller than the sought vector type will be
1450 used as a fall back. This happens because &lt;128 x double&gt; can be
1451 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1452</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1453
Chris Lattner6509f502011-10-11 23:01:39 +0000[diff] [blame]1454<p>The function of the data layout string may not be what you expect. Notably,
1455 this is not a specification from the frontend of what alignment the code
1456 generator should use.</p>
1457
1458<p>Instead, if specified, the target data layout is required to match what the
1459 ultimate <em>code generator</em> expects. This string is used by the
1460 mid-level optimizers to
1461 improve code, and this only works if it matches what the ultimate code
1462 generator uses. If you would like to generate IR that does not embed this
1463 target-specific detail into the IR, then you don't have to specify the
1464 string. This will disable some optimizations that require precise layout
1465 information, but this also prevents those optimizations from introducing
1466 target specificity into the IR.</p>
1467
1468
1469
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1470</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1471
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1472<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1473<h3>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1474 <a name="pointeraliasing">Pointer Aliasing Rules</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1475</h3>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1476
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1477<div>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1478
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1479<p>Any memory access must be done through a pointer value associated
Andreas Bolka99a82052009-07-27 20:37:10 +0000[diff] [blame]1480with an address range of the memory access, otherwise the behavior
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1481is undefined. Pointer values are associated with address ranges
1482according to the following rules:</p>
1483
1484<ul>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1485 <li>A pointer value is associated with the addresses associated with
1486 any value it is <i>based</i> on.
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1487 <li>An address of a global variable is associated with the address
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1488 range of the variable's storage.</li>
1489 <li>The result value of an allocation instruction is associated with
1490 the address range of the allocated storage.</li>
1491 <li>A null pointer in the default address-space is associated with
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1492 no address.</li>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1493 <li>An integer constant other than zero or a pointer value returned
1494 from a function not defined within LLVM may be associated with address
1495 ranges allocated through mechanisms other than those provided by
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1496 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1497 allocated by mechanisms provided by LLVM.</li>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1498</ul>
1499
1500<p>A pointer value is <i>based</i> on another pointer value according
1501 to the following rules:</p>
1502
1503<ul>
1504 <li>A pointer value formed from a
1505 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1506 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1507 <li>The result value of a
1508 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1509 of the <tt>bitcast</tt>.</li>
1510 <li>A pointer value formed by an
1511 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1512 pointer values that contribute (directly or indirectly) to the
1513 computation of the pointer's value.</li>
1514 <li>The "<i>based</i> on" relationship is transitive.</li>
1515</ul>
1516
1517<p>Note that this definition of <i>"based"</i> is intentionally
1518 similar to the definition of <i>"based"</i> in C99, though it is
1519 slightly weaker.</p>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1520
1521<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1522<tt><a href="#i_load">load</a></tt> merely indicates the size and
1523alignment of the memory from which to load, as well as the
Dan Gohmanc22c0f32010-06-17 19:23:50 +0000[diff] [blame]1524interpretation of the value. The first operand type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1525<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1526and alignment of the store.</p>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1527
1528<p>Consequently, type-based alias analysis, aka TBAA, aka
1529<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1530LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1531additional information which specialized optimization passes may use
1532to implement type-based alias analysis.</p>
1533
1534</div>
1535
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1536<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1537<h3>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1538 <a name="volatile">Volatile Memory Accesses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1539</h3>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1540
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1541<div>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1542
1543<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1544href="#i_store"><tt>store</tt></a>s, and <a
1545href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1546The optimizers must not change the number of volatile operations or change their
1547order of execution relative to other volatile operations. The optimizers
1548<i>may</i> change the order of volatile operations relative to non-volatile
1549operations. This is not Java's "volatile" and has no cross-thread
1550synchronization behavior.</p>
1551
1552</div>
1553
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1554<!-- ======================================================================= -->
1555<h3>
1556 <a name="memmodel">Memory Model for Concurrent Operations</a>
1557</h3>
1558
1559<div>
1560
1561<p>The LLVM IR does not define any way to start parallel threads of execution
1562or to register signal handlers. Nonetheless, there are platform-specific
1563ways to create them, and we define LLVM IR's behavior in their presence. This
1564model is inspired by the C++0x memory model.</p>
1565
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1566<p>For a more informal introduction to this model, see the
1567<a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.
1568
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1569<p>We define a <i>happens-before</i> partial order as the least partial order
1570that</p>
1571<ul>
1572 <li>Is a superset of single-thread program order, and</li>
1573 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1574 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1575 by platform-specific techniques, like pthread locks, thread
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1576 creation, thread joining, etc., and by atomic instructions.
1577 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1578 </li>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1579</ul>
1580
1581<p>Note that program order does not introduce <i>happens-before</i> edges
1582between a thread and signals executing inside that thread.</p>
1583
1584<p>Every (defined) read operation (load instructions, memcpy, atomic
1585loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1586(defined) write operations (store instructions, atomic
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1587stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1588initialized globals are considered to have a write of the initializer which is
1589atomic and happens before any other read or write of the memory in question.
1590For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1591any write to the same byte, except:</p>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1592
1593<ul>
1594 <li>If <var>write<sub>1</sub></var> happens before
1595 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1596 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1597 does not see <var>write<sub>1</sub></var>.
Bill Wendling0246bb72011-07-31 06:45:03 +0000[diff] [blame]1598 <li>If <var>R<sub>byte</sub></var> happens before
1599 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1600 see <var>write<sub>3</sub></var>.
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1601</ul>
1602
1603<p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1604<ul>
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1605 <li>If <var>R</var> is volatile, the result is target-dependent. (Volatile
1606 is supposed to give guarantees which can support
1607 <code>sig_atomic_t</code> in C/C++, and may be used for accesses to
1608 addresses which do not behave like normal memory. It does not generally
1609 provide cross-thread synchronization.)
1610 <li>Otherwise, if there is no write to the same byte that happens before
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1611 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1612 <tt>undef</tt> for that byte.
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1613 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1614 <var>R<sub>byte</sub></var> returns the value written by that
1615 write.</li>
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1616 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1617 <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]1618 values written. See the <a href="#ordering">Atomic Memory Ordering
1619 Constraints</a> section for additional constraints on how the choice
1620 is made.
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1621 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1622</ul>
1623
1624<p><var>R</var> returns the value composed of the series of bytes it read.
1625This implies that some bytes within the value may be <tt>undef</tt>
1626<b>without</b> the entire value being <tt>undef</tt>. Note that this only
1627defines the semantics of the operation; it doesn't mean that targets will
1628emit more than one instruction to read the series of bytes.</p>
1629
1630<p>Note that in cases where none of the atomic intrinsics are used, this model
1631places only one restriction on IR transformations on top of what is required
1632for single-threaded execution: introducing a store to a byte which might not
Eli Friedman101c81d2011-08-02 01:15:34 +0000[diff] [blame]1633otherwise be stored is not allowed in general. (Specifically, in the case
1634where another thread might write to and read from an address, introducing a
1635store can change a load that may see exactly one write into a load that may
1636see multiple writes.)</p>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1637
1638<!-- FIXME: This model assumes all targets where concurrency is relevant have
1639a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1640none of the backends currently in the tree fall into this category; however,
1641there might be targets which care. If there are, we want a paragraph
1642like the following:
1643
1644Targets may specify that stores narrower than a certain width are not
1645available; on such a target, for the purposes of this model, treat any
1646non-atomic write with an alignment or width less than the minimum width
1647as if it writes to the relevant surrounding bytes.
1648-->
1649
1650</div>
1651
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1652<!-- ======================================================================= -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]1653<h3>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1654 <a name="ordering">Atomic Memory Ordering Constraints</a>
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]1655</h3>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1656
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]1657<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1658
1659<p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]1660<a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1661<a href="#i_fence"><code>fence</code></a>,
1662<a href="#i_load"><code>atomic load</code></a>, and
Eli Friedman8fa281a2011-08-09 23:26:12 +0000[diff] [blame]1663<a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1664that determines which other atomic instructions on the same address they
1665<i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1666but are somewhat more colloquial. If these descriptions aren't precise enough,
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1667check those specs (see spec references in the
Nick Lewycky300a2632012-01-23 08:47:21 +0000[diff] [blame]1668<a href="Atomics.html#introduction">atomics guide</a>).
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1669<a href="#i_fence"><code>fence</code></a> instructions
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1670treat these orderings somewhat differently since they don't take an address.
1671See that instruction's documentation for details.</p>
1672
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1673<p>For a simpler introduction to the ordering constraints, see the
1674<a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.</p>
1675
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1676<dl>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1677<dt><code>unordered</code></dt>
1678<dd>The set of values that can be read is governed by the happens-before
1679partial order. A value cannot be read unless some operation wrote it.
1680This is intended to provide a guarantee strong enough to model Java's
1681non-volatile shared variables. This ordering cannot be specified for
1682read-modify-write operations; it is not strong enough to make them atomic
1683in any interesting way.</dd>
1684<dt><code>monotonic</code></dt>
1685<dd>In addition to the guarantees of <code>unordered</code>, there is a single
1686total order for modifications by <code>monotonic</code> operations on each
1687address. All modification orders must be compatible with the happens-before
1688order. There is no guarantee that the modification orders can be combined to
1689a global total order for the whole program (and this often will not be
1690possible). The read in an atomic read-modify-write operation
1691(<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1692<a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1693reads the value in the modification order immediately before the value it
1694writes. If one atomic read happens before another atomic read of the same
1695address, the later read must see the same value or a later value in the
1696address's modification order. This disallows reordering of
1697<code>monotonic</code> (or stronger) operations on the same address. If an
1698address is written <code>monotonic</code>ally by one thread, and other threads
1699<code>monotonic</code>ally read that address repeatedly, the other threads must
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1700eventually see the write. This corresponds to the C++0x/C1x
1701<code>memory_order_relaxed</code>.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1702<dt><code>acquire</code></dt>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1703<dd>In addition to the guarantees of <code>monotonic</code>,
Eli Friedmanc264b2f2011-08-24 20:28:39 +0000[diff] [blame]1704a <i>synchronizes-with</i> edge may be formed with a <code>release</code>
1705operation. This is intended to model C++'s <code>memory_order_acquire</code>.</dd>
1706<dt><code>release</code></dt>
1707<dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1708writes a value which is subsequently read by an <code>acquire</code> operation,
1709it <i>synchronizes-with</i> that operation. (This isn't a complete
1710description; see the C++0x definition of a release sequence.) This corresponds
1711to the C++0x/C1x <code>memory_order_release</code>.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1712<dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1713<code>acquire</code> and <code>release</code> operation on its address.
1714This corresponds to the C++0x/C1x <code>memory_order_acq_rel</code>.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1715<dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1716<dd>In addition to the guarantees of <code>acq_rel</code>
1717(<code>acquire</code> for an operation which only reads, <code>release</code>
1718for an operation which only writes), there is a global total order on all
1719sequentially-consistent operations on all addresses, which is consistent with
1720the <i>happens-before</i> partial order and with the modification orders of
1721all the affected addresses. Each sequentially-consistent read sees the last
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1722preceding write to the same address in this global order. This corresponds
1723to the C++0x/C1x <code>memory_order_seq_cst</code> and Java volatile.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1724</dl>
1725
1726<p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1727it only <i>synchronizes with</i> or participates in modification and seq_cst
1728total orderings with other operations running in the same thread (for example,
1729in signal handlers).</p>
1730
1731</div>
1732
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1733</div>
1734
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1735<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1736<h2><a name="typesystem">Type System</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1737<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1738
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1739<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1740
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1741<p>The LLVM type system is one of the most important features of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1742 intermediate representation. Being typed enables a number of optimizations
1743 to be performed on the intermediate representation directly, without having
1744 to do extra analyses on the side before the transformation. A strong type
1745 system makes it easier to read the generated code and enables novel analyses
1746 and transformations that are not feasible to perform on normal three address
1747 code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1748
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1749<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1750<h3>
1751 <a name="t_classifications">Type Classifications</a>
1752</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1753
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1754<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1755
1756<p>The types fall into a few useful classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1757
1758<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1759 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1760 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1761 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1762 <td><a href="#t_integer">integer</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1763 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1764 </tr>
1765 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1766 <td><a href="#t_floating">floating point</a></td>
Dan Gohmance163392011-12-17 00:04:22 +0000[diff] [blame]1767 <td><tt>half, float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1768 </tr>
1769 <tr>
1770 <td><a name="t_firstclass">first class</a></td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1771 <td><a href="#t_integer">integer</a>,
1772 <a href="#t_floating">floating point</a>,
1773 <a href="#t_pointer">pointer</a>,
Dan Gohman0066db62008-06-18 18:42:13 +0000[diff] [blame]1774 <a href="#t_vector">vector</a>,
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]1775 <a href="#t_struct">structure</a>,
1776 <a href="#t_array">array</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1777 <a href="#t_label">label</a>,
1778 <a href="#t_metadata">metadata</a>.
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1779 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1780 </tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1781 <tr>
1782 <td><a href="#t_primitive">primitive</a></td>
1783 <td><a href="#t_label">label</a>,
1784 <a href="#t_void">void</a>,
Tobias Grosser05387292010-12-28 20:29:31 +0000[diff] [blame]1785 <a href="#t_integer">integer</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1786 <a href="#t_floating">floating point</a>,
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1787 <a href="#t_x86mmx">x86mmx</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1788 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1789 </tr>
1790 <tr>
1791 <td><a href="#t_derived">derived</a></td>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1792 <td><a href="#t_array">array</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1793 <a href="#t_function">function</a>,
1794 <a href="#t_pointer">pointer</a>,
1795 <a href="#t_struct">structure</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1796 <a href="#t_vector">vector</a>,
1797 <a href="#t_opaque">opaque</a>.
Dan Gohman01ac1012008-10-14 16:32:04 +0000[diff] [blame]1798 </td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1799 </tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1800 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1801</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1802
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1803<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1804 important. Values of these types are the only ones which can be produced by
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1805 instructions.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1806
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1807</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1808
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1809<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1810<h3>
1811 <a name="t_primitive">Primitive Types</a>
1812</h3>
Chris Lattner8f8c7b72008-01-04 04:34:14 +0000[diff] [blame]1813
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1814<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1815
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1816<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1817 system.</p>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1818
1819<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1820<h4>
1821 <a name="t_integer">Integer Type</a>
1822</h4>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1823
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1824<div>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1825
1826<h5>Overview:</h5>
1827<p>The integer type is a very simple type that simply specifies an arbitrary
1828 bit width for the integer type desired. Any bit width from 1 bit to
1829 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1830
1831<h5>Syntax:</h5>
1832<pre>
1833 iN
1834</pre>
1835
1836<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1837 value.</p>
1838
1839<h5>Examples:</h5>
1840<table class="layout">
1841 <tr class="layout">
1842 <td class="left"><tt>i1</tt></td>
1843 <td class="left">a single-bit integer.</td>
1844 </tr>
1845 <tr class="layout">
1846 <td class="left"><tt>i32</tt></td>
1847 <td class="left">a 32-bit integer.</td>
1848 </tr>
1849 <tr class="layout">
1850 <td class="left"><tt>i1942652</tt></td>
1851 <td class="left">a really big integer of over 1 million bits.</td>
1852 </tr>
1853</table>
1854
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1855</div>
1856
1857<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1858<h4>
1859 <a name="t_floating">Floating Point Types</a>
1860</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1861
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1862<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1863
1864<table>
1865 <tbody>
1866 <tr><th>Type</th><th>Description</th></tr>
Dan Gohmance163392011-12-17 00:04:22 +0000[diff] [blame]1867 <tr><td><tt>half</tt></td><td>16-bit floating point value</td></tr>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1868 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1869 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1870 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1871 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1872 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1873 </tbody>
1874</table>
1875
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1876</div>
1877
1878<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1879<h4>
1880 <a name="t_x86mmx">X86mmx Type</a>
1881</h4>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1882
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1883<div>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1884
1885<h5>Overview:</h5>
1886<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>
1887
1888<h5>Syntax:</h5>
1889<pre>
Dale Johannesen473a8c82010-10-01 01:07:02 +0000[diff] [blame]1890 x86mmx
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1891</pre>
1892
1893</div>
1894
1895<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1896<h4>
1897 <a name="t_void">Void Type</a>
1898</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1899
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1900<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1901
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1902<h5>Overview:</h5>
1903<p>The void type does not represent any value and has no size.</p>
1904
1905<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1906<pre>
1907 void
1908</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1909
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1910</div>
1911
1912<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1913<h4>
1914 <a name="t_label">Label Type</a>
1915</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1916
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1917<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1918
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1919<h5>Overview:</h5>
1920<p>The label type represents code labels.</p>
1921
1922<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1923<pre>
1924 label
1925</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1926
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1927</div>
1928
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1929<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1930<h4>
1931 <a name="t_metadata">Metadata Type</a>
1932</h4>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1933
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1934<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1935
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1936<h5>Overview:</h5>
Nick Lewyckyc261df92009-09-27 23:27:42 +0000[diff] [blame]1937<p>The metadata type represents embedded metadata. No derived types may be
1938 created from metadata except for <a href="#t_function">function</a>
1939 arguments.
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1940
1941<h5>Syntax:</h5>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1942<pre>
1943 metadata
1944</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1945
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1946</div>
1947
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1948</div>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1949
1950<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1951<h3>
1952 <a name="t_derived">Derived Types</a>
1953</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1954
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1955<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1956
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1957<p>The real power in LLVM comes from the derived types in the system. This is
1958 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1959 useful types. Each of these types contain one or more element types which
1960 may be a primitive type, or another derived type. For example, it is
1961 possible to have a two dimensional array, using an array as the element type
1962 of another array.</p>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]1963
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1964<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1965<h4>
1966 <a name="t_aggregate">Aggregate Types</a>
1967</h4>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1968
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1969<div>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1970
1971<p>Aggregate Types are a subset of derived types that can contain multiple
Duncan Sands20536b52011-12-14 15:44:20 +0000[diff] [blame]1972 member types. <a href="#t_array">Arrays</a> and
1973 <a href="#t_struct">structs</a> are aggregate types.
1974 <a href="#t_vector">Vectors</a> are not considered to be aggregate types.</p>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1975
1976</div>
1977
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1978<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1979<h4>
1980 <a name="t_array">Array Type</a>
1981</h4>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1982
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1983<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1984
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1985<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1986<p>The array type is a very simple derived type that arranges elements
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1987 sequentially in memory. The array type requires a size (number of elements)
1988 and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1989
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1990<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1991<pre>
1992 [&lt;# elements&gt; x &lt;elementtype&gt;]
1993</pre>
1994
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1995<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1996 be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1997
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1998<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1999<table class="layout">
2000 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2001 <td class="left"><tt>[40 x i32]</tt></td>
2002 <td class="left">Array of 40 32-bit integer values.</td>
2003 </tr>
2004 <tr class="layout">
2005 <td class="left"><tt>[41 x i32]</tt></td>
2006 <td class="left">Array of 41 32-bit integer values.</td>
2007 </tr>
2008 <tr class="layout">
2009 <td class="left"><tt>[4 x i8]</tt></td>
2010 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2011 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2012</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2013<p>Here are some examples of multidimensional arrays:</p>
2014<table class="layout">
2015 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2016 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
2017 <td class="left">3x4 array of 32-bit integer values.</td>
2018 </tr>
2019 <tr class="layout">
2020 <td class="left"><tt>[12 x [10 x float]]</tt></td>
2021 <td class="left">12x10 array of single precision floating point values.</td>
2022 </tr>
2023 <tr class="layout">
2024 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
2025 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2026 </tr>
2027</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2028
Dan Gohman7657f6b2009-11-09 19:01:53 +0000[diff] [blame]2029<p>There is no restriction on indexing beyond the end of the array implied by
2030 a static type (though there are restrictions on indexing beyond the bounds
2031 of an allocated object in some cases). This means that single-dimension
2032 'variable sized array' addressing can be implemented in LLVM with a zero
2033 length array type. An implementation of 'pascal style arrays' in LLVM could
2034 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]2035
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2036</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2037
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2038<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2039<h4>
2040 <a name="t_function">Function Type</a>
2041</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2042
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2043<div>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]2044
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2045<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2046<p>The function type can be thought of as a function signature. It consists of
2047 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]2048 function type is a first class type or a void type.</p>
Devang Patelc3fc6df2008-03-10 20:49:15 +0000[diff] [blame]2049
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2050<h5>Syntax:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]2051<pre>
Nick Lewycky51386942009-09-27 07:55:32 +0000[diff] [blame]2052 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]2053</pre>
2054
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]2055<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]2056 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
2057 which indicates that the function takes a variable number of arguments.
2058 Variable argument functions can access their arguments with
2059 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]2060 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewyckyc261df92009-09-27 23:27:42 +0000[diff] [blame]2061 <a href="#t_label">label</a>.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]2062
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2063<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2064<table class="layout">
2065 <tr class="layout">
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2066 <td class="left"><tt>i32 (i32)</tt></td>
2067 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2068 </td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2069 </tr><tr class="layout">
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]2070 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencerf17a0b72006-12-31 07:20:23 +0000[diff] [blame]2071 </tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2072 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]2073 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
2074 returning <tt>float</tt>.
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2075 </td>
2076 </tr><tr class="layout">
2077 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2078 <td class="left">A vararg function that takes at least one
2079 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2080 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2081 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2082 </td>
Devang Patela582f402008-03-24 05:35:41 +0000[diff] [blame]2083 </tr><tr class="layout">
2084 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky51386942009-09-27 07:55:32 +0000[diff] [blame]2085 <td class="left">A function taking an <tt>i32</tt>, returning a
2086 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patela582f402008-03-24 05:35:41 +0000[diff] [blame]2087 </td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2088 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2089</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2090
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2091</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2092
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2093<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2094<h4>
2095 <a name="t_struct">Structure Type</a>
2096</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2097
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2098<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2099
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2100<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2101<p>The structure type is used to represent a collection of data members together
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2102 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]2103
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +0000[diff] [blame]2104<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2105 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2106 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2107 Structures in registers are accessed using the
2108 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2109 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2110
2111<p>Structures may optionally be "packed" structures, which indicate that the
2112 alignment of the struct is one byte, and that there is no padding between
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2113 the elements. In non-packed structs, padding between field types is inserted
2114 as defined by the TargetData string in the module, which is required to match
Chris Lattnere4617b02011-10-11 23:02:17 +0000[diff] [blame]2115 what the underlying code generator expects.</p>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2116
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2117<p>Structures can either be "literal" or "identified". A literal structure is
2118 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) whereas identified
2119 types are always defined at the top level with a name. Literal types are
2120 uniqued by their contents and can never be recursive or opaque since there is
Chris Lattneraa175c32011-08-12 18:12:40 +0000[diff] [blame]2121 no way to write one. Identified types can be recursive, can be opaqued, and are
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2122 never uniqued.
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2123</p>
2124
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2125<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2126<pre>
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2127 %T1 = type { &lt;type list&gt; } <i>; Identified normal struct type</i>
2128 %T2 = type &lt;{ &lt;type list&gt; }&gt; <i>; Identified packed struct type</i>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2129</pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2130
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2131<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2132<table class="layout">
2133 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2134 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2135 <td class="left">A triple of three <tt>i32</tt> values</td>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2136 </tr>
2137 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2138 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
2139 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2140 second element is a <a href="#t_pointer">pointer</a> to a
2141 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2142 an <tt>i32</tt>.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2143 </tr>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2144 <tr class="layout">
2145 <td class="left"><tt>&lt;{ i8, i32 }&gt;</tt></td>
2146 <td class="left">A packed struct known to be 5 bytes in size.</td>
2147 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2148</table>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]2149
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2150</div>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2151
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2152<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2153<h4>
Chris Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]2154 <a name="t_opaque">Opaque Structure Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2155</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2156
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2157<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2158
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2159<h5>Overview:</h5>
Chris Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]2160<p>Opaque structure types are used to represent named structure types that do
2161 not have a body specified. This corresponds (for example) to the C notion of
2162 a forward declared structure.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2163
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2164<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2165<pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2166 %X = type opaque
2167 %52 = type opaque
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2168</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2169
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2170<h5>Examples:</h5>
2171<table class="layout">
2172 <tr class="layout">
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2173 <td class="left"><tt>opaque</tt></td>
2174 <td class="left">An opaque type.</td>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2175 </tr>
2176</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2177
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2178</div>
2179
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2180
2181
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2182<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2183<h4>
2184 <a name="t_pointer">Pointer Type</a>
2185</h4>
Chris Lattner0fd4a272009-02-08 19:53:29 +0000[diff] [blame]2186
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2187<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2188
2189<h5>Overview:</h5>
Dan Gohmanff3ef322010-02-25 16:50:07 +0000[diff] [blame]2190<p>The pointer type is used to specify memory locations.
2191 Pointers are commonly used to reference objects in memory.</p>
2192
2193<p>Pointer types may have an optional address space attribute defining the
2194 numbered address space where the pointed-to object resides. The default
2195 address space is number zero. The semantics of non-zero address
2196 spaces are target-specific.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2197
2198<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2199 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner0fd4a272009-02-08 19:53:29 +0000[diff] [blame]2200
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2201<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2202<pre>
2203 &lt;type&gt; *
2204</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2205
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2206<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2207<table class="layout">
2208 <tr class="layout">
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]2209 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2210 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2211 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2212 </tr>
2213 <tr class="layout">
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]2214 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2215 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2216 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2217 <tt>i32</tt>.</td>
2218 </tr>
2219 <tr class="layout">
2220 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2221 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2222 that resides in address space #5.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2223 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2224</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2225
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2226</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2227
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2228<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2229<h4>
2230 <a name="t_vector">Vector Type</a>
2231</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2232
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2233<div>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2234
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2235<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2236<p>A vector type is a simple derived type that represents a vector of elements.
2237 Vector types are used when multiple primitive data are operated in parallel
2238 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sandsd40d14e2009-11-27 13:38:03 +0000[diff] [blame]2239 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2240 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2241
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2242<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2243<pre>
2244 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
2245</pre>
2246
Chris Lattner7d2e7be2010-10-10 18:20:35 +0000[diff] [blame]2247<p>The number of elements is a constant integer value larger than 0; elementtype
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]2248 may be any integer or floating point type, or a pointer to these types.
2249 Vectors of size zero are not allowed. </p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2250
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2251<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2252<table class="layout">
2253 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2254 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
2255 <td class="left">Vector of 4 32-bit integer values.</td>
2256 </tr>
2257 <tr class="layout">
2258 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
2259 <td class="left">Vector of 8 32-bit floating-point values.</td>
2260 </tr>
2261 <tr class="layout">
2262 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
2263 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2264 </tr>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]2265 <tr class="layout">
2266 <td class="left"><tt>&lt;4 x i64*&gt;</tt></td>
2267 <td class="left">Vector of 4 pointers to 64-bit integer values.</td>
2268 </tr>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2269</table>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]2270
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2271</div>
2272
Bill Wendlingaf75f0c2011-07-31 06:47:33 +0000[diff] [blame]2273</div>
2274
NAKAMURA Takumi4b2e07a2011-10-31 13:04:26 +0000[diff] [blame]2275</div>
2276
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2277<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2278<h2><a name="constants">Constants</a></h2>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2279<!-- *********************************************************************** -->
2280
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2281<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2282
2283<p>LLVM has several different basic types of constants. This section describes
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2284 them all and their syntax.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2285
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2286<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2287<h3>
2288 <a name="simpleconstants">Simple Constants</a>
2289</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2290
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2291<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2292
2293<dl>
2294 <dt><b>Boolean constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2295 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]2296 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2297
2298 <dt><b>Integer constants</b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2299 <dd>Standard integers (such as '4') are constants of
2300 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2301 with integer types.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2302
2303 <dt><b>Floating point constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2304 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2305 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2306 notation (see below). The assembler requires the exact decimal value of a
2307 floating-point constant. For example, the assembler accepts 1.25 but
2308 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2309 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2310
2311 <dt><b>Null pointer constants</b></dt>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2312 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2313 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2314</dl>
2315
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2316<p>The one non-intuitive notation for constants is the hexadecimal form of
2317 floating point constants. For example, the form '<tt>double
2318 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2319 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2320 constants are required (and the only time that they are generated by the
2321 disassembler) is when a floating point constant must be emitted but it cannot
2322 be represented as a decimal floating point number in a reasonable number of
2323 digits. For example, NaN's, infinities, and other special values are
2324 represented in their IEEE hexadecimal format so that assembly and disassembly
2325 do not cause any bits to change in the constants.</p>
2326
Dan Gohmance163392011-12-17 00:04:22 +0000[diff] [blame]2327<p>When using the hexadecimal form, constants of types half, float, and double are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2328 represented using the 16-digit form shown above (which matches the IEEE754
Dan Gohmance163392011-12-17 00:04:22 +0000[diff] [blame]2329 representation for double); half and float values must, however, be exactly
2330 representable as IEE754 half and single precision, respectively.
2331 Hexadecimal format is always used
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2332 for long double, and there are three forms of long double. The 80-bit format
2333 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2334 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2335 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2336 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2337 currently supported target uses this format. Long doubles will only work if
Tobias Grosser057beb82012-05-24 15:59:06 +0000[diff] [blame]2338 they match the long double format on your target. The IEEE 16-bit format
2339 (half precision) is represented by <tt>0xH</tt> followed by 4 hexadecimal
2340 digits. All hexadecimal formats are big-endian (sign bit at the left).</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2341
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]2342<p>There are no constants of type x86mmx.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2343</div>
2344
2345<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2346<h3>
Bill Wendlingd9fe2982009-07-20 02:32:41 +0000[diff] [blame]2347<a name="aggregateconstants"></a> <!-- old anchor -->
2348<a name="complexconstants">Complex Constants</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2349</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2350
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2351<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2352
Chris Lattner70882792009-02-28 18:32:25 +0000[diff] [blame]2353<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2354 constants and smaller complex constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2355
2356<dl>
2357 <dt><b>Structure constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2358 <dd>Structure constants are represented with notation similar to structure
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2359 type definitions (a comma separated list of elements, surrounded by braces
2360 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2361 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2362 Structure constants must have <a href="#t_struct">structure type</a>, and
2363 the number and types of elements must match those specified by the
2364 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2365
2366 <dt><b>Array constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2367 <dd>Array constants are represented with notation similar to array type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2368 definitions (a comma separated list of elements, surrounded by square
2369 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2370 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2371 the number and types of elements must match those specified by the
2372 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2373
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2374 <dt><b>Vector constants</b></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2375 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2376 definitions (a comma separated list of elements, surrounded by
2377 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2378 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2379 have <a href="#t_vector">vector type</a>, and the number and types of
2380 elements must match those specified by the type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2381
2382 <dt><b>Zero initialization</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2383 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]2384 value to zero of <em>any</em> type, including scalar and
2385 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2386 This is often used to avoid having to print large zero initializers
2387 (e.g. for large arrays) and is always exactly equivalent to using explicit
2388 zero initializers.</dd>
Nick Lewycky21cc4462009-04-04 07:22:01 +0000[diff] [blame]2389
2390 <dt><b>Metadata node</b></dt>
Nick Lewycky1e8c7a62009-05-30 16:08:30 +0000[diff] [blame]2391 <dd>A metadata node is a structure-like constant with
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2392 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2393 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2394 be interpreted as part of the instruction stream, metadata is a place to
2395 attach additional information such as debug info.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2396</dl>
2397
2398</div>
2399
2400<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2401<h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2402 <a name="globalconstants">Global Variable and Function Addresses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2403</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2404
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2405<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2406
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2407<p>The addresses of <a href="#globalvars">global variables</a>
2408 and <a href="#functionstructure">functions</a> are always implicitly valid
2409 (link-time) constants. These constants are explicitly referenced when
2410 the <a href="#identifiers">identifier for the global</a> is used and always
2411 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2412 legal LLVM file:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2413
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2414<pre class="doc_code">
Chris Lattnera18a4242007-06-06 18:28:13 +0000[diff] [blame]2415@X = global i32 17
2416@Y = global i32 42
2417@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2418</pre>
2419
2420</div>
2421
2422<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2423<h3>
2424 <a name="undefvalues">Undefined Values</a>
2425</h3>
2426
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2427<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2428
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2429<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]2430 indicates that the user of the value may receive an unspecified bit-pattern.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2431 Undefined values may be of any type (other than '<tt>label</tt>'
2432 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2433
Chris Lattnerc608cb12009-09-11 01:49:31 +0000[diff] [blame]2434<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2435 program is well defined no matter what value is used. This gives the
2436 compiler more freedom to optimize. Here are some examples of (potentially
2437 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2438
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2439
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2440<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2441 %A = add %X, undef
2442 %B = sub %X, undef
2443 %C = xor %X, undef
2444Safe:
2445 %A = undef
2446 %B = undef
2447 %C = undef
2448</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2449
2450<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]2451 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]2452
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2453<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2454 %A = or %X, undef
2455 %B = and %X, undef
2456Safe:
2457 %A = -1
2458 %B = 0
2459Unsafe:
2460 %A = undef
2461 %B = undef
2462</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2463
2464<p>These logical operations have bits that are not always affected by the input.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2465 For example, if <tt>%X</tt> has a zero bit, then the output of the
2466 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2467 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2468 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2469 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2470 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2471 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2472 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2473
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2474<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2475 %A = select undef, %X, %Y
2476 %B = select undef, 42, %Y
2477 %C = select %X, %Y, undef
2478Safe:
2479 %A = %X (or %Y)
2480 %B = 42 (or %Y)
2481 %C = %Y
2482Unsafe:
2483 %A = undef
2484 %B = undef
2485 %C = undef
2486</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2487
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2488<p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2489 branch) conditions can go <em>either way</em>, but they have to come from one
2490 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2491 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2492 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2493 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2494 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2495 eliminated.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2496
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2497<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2498 %A = xor undef, undef
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2499
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2500 %B = undef
2501 %C = xor %B, %B
2502
2503 %D = undef
2504 %E = icmp lt %D, 4
2505 %F = icmp gte %D, 4
2506
2507Safe:
2508 %A = undef
2509 %B = undef
2510 %C = undef
2511 %D = undef
2512 %E = undef
2513 %F = undef
2514</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2515
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2516<p>This example points out that two '<tt>undef</tt>' operands are not
2517 necessarily the same. This can be surprising to people (and also matches C
2518 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2519 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2520 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2521 its value over its "live range". This is true because the variable doesn't
2522 actually <em>have a live range</em>. Instead, the value is logically read
2523 from arbitrary registers that happen to be around when needed, so the value
2524 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2525 need to have the same semantics or the core LLVM "replace all uses with"
2526 concept would not hold.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2527
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2528<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2529 %A = fdiv undef, %X
2530 %B = fdiv %X, undef
2531Safe:
2532 %A = undef
2533b: unreachable
2534</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2535
2536<p>These examples show the crucial difference between an <em>undefined
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2537 value</em> and <em>undefined behavior</em>. An undefined value (like
2538 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2539 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2540 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2541 defined on SNaN's. However, in the second example, we can make a more
2542 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2543 arbitrary value, we are allowed to assume that it could be zero. Since a
2544 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2545 the operation does not execute at all. This allows us to delete the divide and
2546 all code after it. Because the undefined operation "can't happen", the
2547 optimizer can assume that it occurs in dead code.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2548
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2549<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2550a: store undef -> %X
2551b: store %X -> undef
2552Safe:
2553a: &lt;deleted&gt;
2554b: unreachable
2555</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2556
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2557<p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2558 undefined value can be assumed to not have any effect; we can assume that the
2559 value is overwritten with bits that happen to match what was already there.
2560 However, a store <em>to</em> an undefined location could clobber arbitrary
2561 memory, therefore, it has undefined behavior.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2562
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2563</div>
2564
2565<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2566<h3>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2567 <a name="poisonvalues">Poison Values</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2568</h3>
2569
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2570<div>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2571
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2572<p>Poison values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2573 they also represent the fact that an instruction or constant expression which
2574 cannot evoke side effects has nevertheless detected a condition which results
2575 in undefined behavior.</p>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2576
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2577<p>There is currently no way of representing a poison value in the IR; they
Dan Gohman855abed2010-05-03 14:51:43 +0000[diff] [blame]2578 only exist when produced by operations such as
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2579 <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]2580
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2581<p>Poison value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2582
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2583<ul>
2584<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2585 their operands.</li>
2586
2587<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2588 to their dynamic predecessor basic block.</li>
2589
2590<li>Function arguments depend on the corresponding actual argument values in
2591 the dynamic callers of their functions.</li>
2592
2593<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2594 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2595 control back to them.</li>
2596
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2597<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
Bill Wendling7b9e5392012-02-06 21:57:33 +0000[diff] [blame]2598 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_resume"><tt>resume</tt></a>,
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2599 or exception-throwing call instructions that dynamically transfer control
2600 back to them.</li>
2601
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2602<li>Non-volatile loads and stores depend on the most recent stores to all of the
2603 referenced memory addresses, following the order in the IR
2604 (including loads and stores implied by intrinsics such as
2605 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2606
Dan Gohman7c24ff12010-05-03 14:59:34 +0000[diff] [blame]2607<!-- TODO: In the case of multiple threads, this only applies if the store
2608 "happens-before" the load or store. -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2609
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2610<!-- TODO: floating-point exception state -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2611
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2612<li>An instruction with externally visible side effects depends on the most
2613 recent preceding instruction with externally visible side effects, following
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]2614 the order in the IR. (This includes
2615 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2616
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2617<li>An instruction <i>control-depends</i> on a
2618 <a href="#terminators">terminator instruction</a>
2619 if the terminator instruction has multiple successors and the instruction
2620 is always executed when control transfers to one of the successors, and
Chris Lattner7a2bdde2011-04-15 05:18:47 +0000[diff] [blame]2621 may not be executed when control is transferred to another.</li>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2622
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2623<li>Additionally, an instruction also <i>control-depends</i> on a terminator
2624 instruction if the set of instructions it otherwise depends on would be
Chris Lattner7a2bdde2011-04-15 05:18:47 +0000[diff] [blame]2625 different if the terminator had transferred control to a different
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2626 successor.</li>
2627
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2628<li>Dependence is transitive.</li>
2629
2630</ul>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2631
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2632<p>Poison Values have the same behavior as <a href="#undefvalues">undef values</a>,
2633 with the additional affect that any instruction which has a <i>dependence</i>
2634 on a poison value has undefined behavior.</p>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2635
2636<p>Here are some examples:</p>
Dan Gohmanc30f6e12010-04-26 20:54:53 +0000[diff] [blame]2637
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2638<pre class="doc_code">
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2639entry:
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2640 %poison = sub nuw i32 0, 1 ; Results in a poison value.
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2641 %still_poison = and i32 %poison, 0 ; 0, but also poison.
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2642 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2643 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2644
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2645 store i32 %poison, i32* @g ; Poison value stored to memory.
2646 %poison2 = load i32* @g ; Poison value loaded back from memory.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2647
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2648 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2649
2650 %narrowaddr = bitcast i32* @g to i16*
2651 %wideaddr = bitcast i32* @g to i64*
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2652 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2653 %poison4 = load i64* %wideaddr ; Returns a poison value.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2654
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2655 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2656 br i1 %cmp, label %true, label %end ; Branch to either destination.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2657
2658true:
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2659 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2660 ; it has undefined behavior.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2661 br label %end
2662
2663end:
2664 %p = phi i32 [ 0, %entry ], [ 1, %true ]
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2665 ; Both edges into this PHI are
2666 ; control-dependent on %cmp, so this
2667 ; always results in a poison value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2668
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2669 store volatile i32 0, i32* @g ; This would depend on the store in %true
2670 ; if %cmp is true, or the store in %entry
2671 ; otherwise, so this is undefined behavior.
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2672
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2673 br i1 %cmp, label %second_true, label %second_end
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2674 ; The same branch again, but this time the
2675 ; true block doesn't have side effects.
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2676
2677second_true:
2678 ; No side effects!
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2679 ret void
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2680
2681second_end:
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2682 store volatile i32 0, i32* @g ; This time, the instruction always depends
2683 ; on the store in %end. Also, it is
2684 ; control-equivalent to %end, so this is
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2685 ; well-defined (ignoring earlier undefined
2686 ; behavior in this example).
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2687</pre>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2688
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2689</div>
2690
2691<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2692<h3>
2693 <a name="blockaddress">Addresses of Basic Blocks</a>
2694</h3>
2695
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2696<div>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2697
Chris Lattnercdfc9402009-11-01 01:27:45 +0000[diff] [blame]2698<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2699
2700<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner2dfdf2a2009-10-27 21:49:40 +0000[diff] [blame]2701 basic block in the specified function, and always has an i8* type. Taking
Chris Lattnercdfc9402009-11-01 01:27:45 +0000[diff] [blame]2702 the address of the entry block is illegal.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2703
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2704<p>This value only has defined behavior when used as an operand to the
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2705 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2706 comparisons against null. Pointer equality tests between labels addresses
2707 results in undefined behavior &mdash; though, again, comparison against null
2708 is ok, and no label is equal to the null pointer. This may be passed around
2709 as an opaque pointer sized value as long as the bits are not inspected. This
2710 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2711 long as the original value is reconstituted before the <tt>indirectbr</tt>
2712 instruction.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2713
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2714<p>Finally, some targets may provide defined semantics when using the value as
2715 the operand to an inline assembly, but that is target specific.</p>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2716
2717</div>
2718
2719
2720<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2721<h3>
2722 <a name="constantexprs">Constant Expressions</a>
2723</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2724
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2725<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2726
2727<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2728 to be used as constants. Constant expressions may be of
2729 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2730 operation that does not have side effects (e.g. load and call are not
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2731 supported). The following is the syntax for constant expressions:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2732
2733<dl>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2734 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2735 <dd>Truncate a constant to another type. The bit size of CST must be larger
2736 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2737
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2738 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2739 <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]2740 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2741
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2742 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2743 <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]2744 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2745
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2746 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2747 <dd>Truncate a floating point constant to another floating point type. The
2748 size of CST must be larger than the size of TYPE. Both types must be
2749 floating point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2750
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2751 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2752 <dd>Floating point extend a constant to another type. The size of CST must be
2753 smaller or equal to the size of TYPE. Both types must be floating
2754 point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2755
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2756 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2757 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2758 constant. TYPE must be a scalar or vector integer type. CST must be of
2759 scalar or vector floating point type. Both CST and TYPE must be scalars,
2760 or vectors of the same number of elements. If the value won't fit in the
2761 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2762
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2763 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2764 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2765 constant. TYPE must be a scalar or vector integer type. CST must be of
2766 scalar or vector floating point type. Both CST and TYPE must be scalars,
2767 or vectors of the same number of elements. If the value won't fit in the
2768 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2769
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2770 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2771 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2772 constant. TYPE must be a scalar or vector floating point type. CST must be
2773 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2774 vectors of the same number of elements. If the value won't fit in the
2775 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2776
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2777 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2778 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2779 constant. TYPE must be a scalar or vector floating point type. CST must be
2780 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2781 vectors of the same number of elements. If the value won't fit in the
2782 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2783
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2784 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2785 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2786 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2787 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2788 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2789
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2790 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Sylvestre Ledru7f7390e2012-07-25 22:01:31 +0000[diff] [blame]2791 <dd>Convert an integer constant to a pointer constant. TYPE must be a pointer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2792 type. CST must be of integer type. The CST value is zero extended,
2793 truncated, or unchanged to make it fit in a pointer size. This one is
2794 <i>really</i> dangerous!</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2795
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2796 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner03bbad62009-02-28 18:27:03 +0000[diff] [blame]2797 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2798 are the same as those for the <a href="#i_bitcast">bitcast
2799 instruction</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2800
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2801 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2802 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2803 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2804 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2805 instruction, the index list may have zero or more indexes, which are
2806 required to make sense for the type of "CSTPTR".</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2807
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2808 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2809 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2810
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2811 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2812 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2813
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2814 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2815 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]2816
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2817 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2818 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2819 constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]2820
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2821 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2822 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2823 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]2824
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2825 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2826 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2827 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]2828
Nick Lewycky9e130ce2010-05-29 06:44:15 +0000[diff] [blame]2829 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2830 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2831 constants. The index list is interpreted in a similar manner as indices in
2832 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2833 index value must be specified.</dd>
2834
2835 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2836 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2837 constants. The index list is interpreted in a similar manner as indices in
2838 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2839 index value must be specified.</dd>
2840
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2841 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2842 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2843 be any of the <a href="#binaryops">binary</a>
2844 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2845 on operands are the same as those for the corresponding instruction
2846 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2847</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2848
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2849</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000[diff] [blame]2850
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2851</div>
2852
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2853<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2854<h2><a name="othervalues">Other Values</a></h2>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2855<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2856<div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2857<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2858<h3>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2859<a name="inlineasm">Inline Assembler Expressions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2860</h3>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2861
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2862<div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2863
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2864<p>LLVM supports inline assembler expressions (as opposed
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2865 to <a href="#moduleasm">Module-Level Inline Assembly</a>) through the use of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2866 a special value. This value represents the inline assembler as a string
2867 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2868 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2869 expression has side effects, and a flag indicating whether the function
2870 containing the asm needs to align its stack conservatively. An example
2871 inline assembler expression is:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2872
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2873<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2874i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2875</pre>
2876
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2877<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
Nuno Lopes0b5f1ac2012-07-17 23:51:33 +0000[diff] [blame]2878 a <a href="#i_call"><tt>call</tt></a> or an
2879 <a href="#i_invoke"><tt>invoke</tt></a> instruction.
2880 Thus, typically we have:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2881
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2882<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2883%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]2884</pre>
2885
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2886<p>Inline asms with side effects not visible in the constraint list must be
2887 marked as having side effects. This is done through the use of the
2888 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2889
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2890<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2891call void asm sideeffect "eieio", ""()
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2892</pre>
2893
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2894<p>In some cases inline asms will contain code that will not work unless the
2895 stack is aligned in some way, such as calls or SSE instructions on x86,
2896 yet will not contain code that does that alignment within the asm.
2897 The compiler should make conservative assumptions about what the asm might
2898 contain and should generate its usual stack alignment code in the prologue
2899 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2900
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2901<pre class="doc_code">
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2902call void asm alignstack "eieio", ""()
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2903</pre>
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2904
2905<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2906 first.</p>
2907
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2908<!--
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2909<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2910 documented here. Constraints on what can be done (e.g. duplication, moving,
2911 etc need to be documented). This is probably best done by reference to
2912 another document that covers inline asm from a holistic perspective.</p>
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2913 -->
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2914
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2915<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2916<h4>
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2917 <a name="inlineasm_md">Inline Asm Metadata</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2918</h4>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2919
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2920<div>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2921
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2922<p>The call instructions that wrap inline asm nodes may have a
2923 "<tt>!srcloc</tt>" MDNode attached to it that contains a list of constant
2924 integers. If present, the code generator will use the integer as the
2925 location cookie value when report errors through the <tt>LLVMContext</tt>
2926 error reporting mechanisms. This allows a front-end to correlate backend
2927 errors that occur with inline asm back to the source code that produced it.
2928 For example:</p>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2929
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2930<pre class="doc_code">
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2931call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2932...
2933!42 = !{ i32 1234567 }
2934</pre>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2935
2936<p>It is up to the front-end to make sense of the magic numbers it places in the
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2937 IR. If the MDNode contains multiple constants, the code generator will use
Chris Lattnerce1b9ad2010-11-17 08:20:42 +0000[diff] [blame]2938 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]2939
2940</div>
2941
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2942</div>
2943
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2944<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2945<h3>
2946 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2947</h3>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2948
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2949<div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2950
2951<p>LLVM IR allows metadata to be attached to instructions in the program that
2952 can convey extra information about the code to the optimizers and code
2953 generator. One example application of metadata is source-level debug
2954 information. There are two metadata primitives: strings and nodes. All
2955 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2956 preceding exclamation point ('<tt>!</tt>').</p>
2957
2958<p>A metadata string is a string surrounded by double quotes. It can contain
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2959 any character by escaping non-printable characters with "<tt>\xx</tt>" where
2960 "<tt>xx</tt>" is the two digit hex code. For example:
2961 "<tt>!"test\00"</tt>".</p>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2962
2963<p>Metadata nodes are represented with notation similar to structure constants
2964 (a comma separated list of elements, surrounded by braces and preceded by an
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2965 exclamation point). Metadata nodes can have any values as their operand. For
2966 example:</p>
2967
2968<div class="doc_code">
2969<pre>
2970!{ metadata !"test\00", i32 10}
2971</pre>
2972</div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2973
2974<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2975 metadata nodes, which can be looked up in the module symbol table. For
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2976 example:</p>
2977
2978<div class="doc_code">
2979<pre>
2980!foo = metadata !{!4, !3}
2981</pre>
2982</div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2983
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2984<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2985 function is using two metadata arguments:</p>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2986
Bill Wendling9ff5de92011-03-02 02:17:11 +0000[diff] [blame]2987<div class="doc_code">
2988<pre>
2989call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2990</pre>
2991</div>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2992
2993<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2994 attached to the <tt>add</tt> instruction using the <tt>!dbg</tt>
2995 identifier:</p>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2996
Bill Wendling9ff5de92011-03-02 02:17:11 +0000[diff] [blame]2997<div class="doc_code">
2998<pre>
2999%indvar.next = add i64 %indvar, 1, !dbg !21
3000</pre>
3001</div>
3002
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]3003<p>More information about specific metadata nodes recognized by the optimizers
3004 and code generator is found below.</p>
3005
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]3006<!-- _______________________________________________________________________ -->
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]3007<h4>
3008 <a name="tbaa">'<tt>tbaa</tt>' Metadata</a>
3009</h4>
3010
3011<div>
3012
3013<p>In LLVM IR, memory does not have types, so LLVM's own type system is not
3014 suitable for doing TBAA. Instead, metadata is added to the IR to describe
3015 a type system of a higher level language. This can be used to implement
3016 typical C/C++ TBAA, but it can also be used to implement custom alias
3017 analysis behavior for other languages.</p>
3018
3019<p>The current metadata format is very simple. TBAA metadata nodes have up to
3020 three fields, e.g.:</p>
3021
3022<div class="doc_code">
3023<pre>
3024!0 = metadata !{ metadata !"an example type tree" }
3025!1 = metadata !{ metadata !"int", metadata !0 }
3026!2 = metadata !{ metadata !"float", metadata !0 }
3027!3 = metadata !{ metadata !"const float", metadata !2, i64 1 }
3028</pre>
3029</div>
3030
3031<p>The first field is an identity field. It can be any value, usually
3032 a metadata string, which uniquely identifies the type. The most important
3033 name in the tree is the name of the root node. Two trees with
3034 different root node names are entirely disjoint, even if they
3035 have leaves with common names.</p>
3036
3037<p>The second field identifies the type's parent node in the tree, or
3038 is null or omitted for a root node. A type is considered to alias
3039 all of its descendants and all of its ancestors in the tree. Also,
3040 a type is considered to alias all types in other trees, so that
3041 bitcode produced from multiple front-ends is handled conservatively.</p>
3042
3043<p>If the third field is present, it's an integer which if equal to 1
3044 indicates that the type is "constant" (meaning
3045 <tt>pointsToConstantMemory</tt> should return true; see
3046 <a href="AliasAnalysis.html#OtherItfs">other useful
3047 <tt>AliasAnalysis</tt> methods</a>).</p>
3048
3049</div>
3050
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]3051<!-- _______________________________________________________________________ -->
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3052<h4>
Duncan Sands5e5c5f82012-04-14 12:36:06 +0000[diff] [blame]3053 <a name="fpmath">'<tt>fpmath</tt>' Metadata</a>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3054</h4>
3055
3056<div>
3057
Duncan Sands5e5c5f82012-04-14 12:36:06 +0000[diff] [blame]3058<p><tt>fpmath</tt> metadata may be attached to any instruction of floating point
Duncan Sands8883c432012-04-16 16:28:59 +0000[diff] [blame]3059 type. It can be used to express the maximum acceptable error in the result of
3060 that instruction, in ULPs, thus potentially allowing the compiler to use a
Duncan Sands2867c852012-04-16 19:39:33 +0000[diff] [blame]3061 more efficient but less accurate method of computing it. ULP is defined as
3062 follows:</p>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3063
Bill Wendling0656e252011-11-09 19:33:56 +0000[diff] [blame]3064<blockquote>
3065
3066<p>If <tt>x</tt> is a real number that lies between two finite consecutive
3067 floating-point numbers <tt>a</tt> and <tt>b</tt>, without being equal to one
3068 of them, then <tt>ulp(x) = |b - a|</tt>, otherwise <tt>ulp(x)</tt> is the
3069 distance between the two non-equal finite floating-point numbers nearest
3070 <tt>x</tt>. Moreover, <tt>ulp(NaN)</tt> is <tt>NaN</tt>.</p>
3071
3072</blockquote>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3073
Duncan Sands8883c432012-04-16 16:28:59 +0000[diff] [blame]3074<p>The metadata node shall consist of a single positive floating point number
Duncan Sands2867c852012-04-16 19:39:33 +0000[diff] [blame]3075 representing the maximum relative error, for example:</p>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3076
3077<div class="doc_code">
3078<pre>
Duncan Sands8883c432012-04-16 16:28:59 +0000[diff] [blame]3079!0 = metadata !{ float 2.5 } ; maximum acceptable inaccuracy is 2.5 ULPs
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3080</pre>
3081</div>
3082
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000[diff] [blame]3083</div>
3084
Rafael Espindola39dd3282012-03-24 00:14:51 +0000[diff] [blame]3085<!-- _______________________________________________________________________ -->
3086<h4>
3087 <a name="range">'<tt>range</tt>' Metadata</a>
3088</h4>
3089
3090<div>
3091<p><tt>range</tt> metadata may be attached only to loads of integer types. It
3092 expresses the possible ranges the loaded value is in. The ranges are
3093 represented with a flattened list of integers. The loaded value is known to
3094 be in the union of the ranges defined by each consecutive pair. Each pair
3095 has the following properties:</p>
3096<ul>
3097 <li>The type must match the type loaded by the instruction.</li>
3098 <li>The pair <tt>a,b</tt> represents the range <tt>[a,b)</tt>.</li>
3099 <li>Both <tt>a</tt> and <tt>b</tt> are constants.</li>
3100 <li>The range is allowed to wrap.</li>
3101 <li>The range should not represent the full or empty set. That is,
3102 <tt>a!=b</tt>. </li>
3103</ul>
Rafael Espindolaa1b95f52012-05-31 16:04:26 +0000[diff] [blame]3104<p> In addition, the pairs must be in signed order of the lower bound and
3105 they must be non-contiguous.</p>
Rafael Espindola39dd3282012-03-24 00:14:51 +0000[diff] [blame]3106
3107<p>Examples:</p>
3108<div class="doc_code">
3109<pre>
3110 %a = load i8* %x, align 1, !range !0 ; Can only be 0 or 1
3111 %b = load i8* %y, align 1, !range !1 ; Can only be 255 (-1), 0 or 1
3112 %c = load i8* %z, align 1, !range !2 ; Can only be 0, 1, 3, 4 or 5
Rafael Espindolac49b29e2012-05-31 13:45:46 +0000[diff] [blame]3113 %d = load i8* %z, align 1, !range !3 ; Can only be -2, -1, 3, 4 or 5
Rafael Espindola39dd3282012-03-24 00:14:51 +0000[diff] [blame]3114...
3115!0 = metadata !{ i8 0, i8 2 }
3116!1 = metadata !{ i8 255, i8 2 }
3117!2 = metadata !{ i8 0, i8 2, i8 3, i8 6 }
Rafael Espindolac49b29e2012-05-31 13:45:46 +0000[diff] [blame]3118!3 = metadata !{ i8 -2, i8 0, i8 3, i8 6 }
Rafael Espindola39dd3282012-03-24 00:14:51 +0000[diff] [blame]3119</pre>
3120</div>
3121</div>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3122</div>
3123
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]3124</div>
3125
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3126<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3127<h2>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3128 <a name="module_flags">Module Flags Metadata</a>
3129</h2>
3130<!-- *********************************************************************** -->
3131
3132<div>
3133
3134<p>Information about the module as a whole is difficult to convey to LLVM's
3135 subsystems. The LLVM IR isn't sufficient to transmit this
3136 information. The <tt>llvm.module.flags</tt> named metadata exists in order to
3137 facilitate this. These flags are in the form of key / value pairs &mdash;
3138 much like a dictionary &mdash; making it easy for any subsystem who cares
3139 about a flag to look it up.</p>
3140
3141<p>The <tt>llvm.module.flags</tt> metadata contains a list of metadata
3142 triplets. Each triplet has the following form:</p>
3143
3144<ul>
3145 <li>The first element is a <i>behavior</i> flag, which specifies the behavior
3146 when two (or more) modules are merged together, and it encounters two (or
3147 more) metadata with the same ID. The supported behaviors are described
3148 below.</li>
3149
3150 <li>The second element is a metadata string that is a unique ID for the
3151 metadata. How each ID is interpreted is documented below.</li>
3152
3153 <li>The third element is the value of the flag.</li>
3154</ul>
3155
3156<p>When two (or more) modules are merged together, the resulting
3157 <tt>llvm.module.flags</tt> metadata is the union of the
3158 modules' <tt>llvm.module.flags</tt> metadata. The only exception being a flag
3159 with the <i>Override</i> behavior, which may override another flag's value
3160 (see below).</p>
3161
3162<p>The following behaviors are supported:</p>
3163
3164<table border="1" cellspacing="0" cellpadding="4">
3165 <tbody>
3166 <tr>
3167 <th>Value</th>
3168 <th>Behavior</th>
3169 </tr>
3170 <tr>
3171 <td>1</td>
3172 <td align="left">
Bill Wendlinga0edecf2012-03-06 09:17:04 +0000[diff] [blame]3173 <dl>
3174 <dt><b>Error</b></dt>
3175 <dd>Emits an error if two values disagree. It is an error to have an ID
3176 with both an Error and a Warning behavior.</dd>
3177 </dl>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3178 </td>
3179 </tr>
3180 <tr>
3181 <td>2</td>
3182 <td align="left">
Bill Wendlinga0edecf2012-03-06 09:17:04 +0000[diff] [blame]3183 <dl>
3184 <dt><b>Warning</b></dt>
3185 <dd>Emits a warning if two values disagree.</dd>
3186 </dl>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3187 </td>
3188 </tr>
3189 <tr>
3190 <td>3</td>
3191 <td align="left">
Bill Wendlinga0edecf2012-03-06 09:17:04 +0000[diff] [blame]3192 <dl>
3193 <dt><b>Require</b></dt>
3194 <dd>Emits an error when the specified value is not present or doesn't
3195 have the specified value. It is an error for two (or more)
3196 <tt>llvm.module.flags</tt> with the same ID to have the Require
3197 behavior but different values. There may be multiple Require flags
3198 per ID.</dd>
3199 </dl>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3200 </td>
3201 </tr>
3202 <tr>
3203 <td>4</td>
3204 <td align="left">
Bill Wendlinga0edecf2012-03-06 09:17:04 +0000[diff] [blame]3205 <dl>
3206 <dt><b>Override</b></dt>
3207 <dd>Uses the specified value if the two values disagree. It is an
3208 error for two (or more) <tt>llvm.module.flags</tt> with the same
3209 ID to have the Override behavior but different values.</dd>
3210 </dl>
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3211 </td>
3212 </tr>
3213 </tbody>
3214</table>
3215
3216<p>An example of module flags:</p>
3217
3218<pre class="doc_code">
3219!0 = metadata !{ i32 1, metadata !"foo", i32 1 }
3220!1 = metadata !{ i32 4, metadata !"bar", i32 37 }
3221!2 = metadata !{ i32 2, metadata !"qux", i32 42 }
3222!3 = metadata !{ i32 3, metadata !"qux",
3223 metadata !{
3224 metadata !"foo", i32 1
3225 }
3226}
3227!llvm.module.flags = !{ !0, !1, !2, !3 }
3228</pre>
3229
3230<ul>
3231 <li><p>Metadata <tt>!0</tt> has the ID <tt>!"foo"</tt> and the value '1'. The
3232 behavior if two or more <tt>!"foo"</tt> flags are seen is to emit an
3233 error if their values are not equal.</p></li>
3234
3235 <li><p>Metadata <tt>!1</tt> has the ID <tt>!"bar"</tt> and the value '37'. The
3236 behavior if two or more <tt>!"bar"</tt> flags are seen is to use the
3237 value '37' if their values are not equal.</p></li>
3238
3239 <li><p>Metadata <tt>!2</tt> has the ID <tt>!"qux"</tt> and the value '42'. The
3240 behavior if two or more <tt>!"qux"</tt> flags are seen is to emit a
3241 warning if their values are not equal.</p></li>
3242
3243 <li><p>Metadata <tt>!3</tt> has the ID <tt>!"qux"</tt> and the value:</p>
3244
3245<pre class="doc_code">
3246metadata !{ metadata !"foo", i32 1 }
3247</pre>
Bill Wendlingf7b367c2012-02-16 01:10:50 +0000[diff] [blame]3248
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3249 <p>The behavior is to emit an error if the <tt>llvm.module.flags</tt> does
3250 not contain a flag with the ID <tt>!"foo"</tt> that has the value
3251 '1'. If two or more <tt>!"qux"</tt> flags exist, then they must have
3252 the same value or an error will be issued.</p></li>
3253</ul>
3254
Bill Wendlingf7b367c2012-02-16 01:10:50 +0000[diff] [blame]3255
3256<!-- ======================================================================= -->
3257<h3>
3258<a name="objc_gc_flags">Objective-C Garbage Collection Module Flags Metadata</a>
3259</h3>
3260
3261<div>
3262
3263<p>On the Mach-O platform, Objective-C stores metadata about garbage collection
3264 in a special section called "image info". The metadata consists of a version
3265 number and a bitmask specifying what types of garbage collection are
3266 supported (if any) by the file. If two or more modules are linked together
3267 their garbage collection metadata needs to be merged rather than appended
3268 together.</p>
3269
3270<p>The Objective-C garbage collection module flags metadata consists of the
3271 following key-value pairs:</p>
3272
3273<table border="1" cellspacing="0" cellpadding="4">
Bill Wendlingb3ef2232012-03-06 09:23:25 +0000[diff] [blame]3274 <col width="30%">
Bill Wendlingf7b367c2012-02-16 01:10:50 +0000[diff] [blame]3275 <tbody>
3276 <tr>
Bill Wendlinga0edecf2012-03-06 09:17:04 +0000[diff] [blame]3277 <th>Key</th>
Bill Wendlingf7b367c2012-02-16 01:10:50 +0000[diff] [blame]3278 <th>Value</th>
3279 </tr>
3280 <tr>
3281 <td><tt>Objective-C&nbsp;Version</tt></td>
3282 <td align="left"><b>[Required]</b> &mdash; The Objective-C ABI
3283 version. Valid values are 1 and 2.</td>
3284 </tr>
3285 <tr>
3286 <td><tt>Objective-C&nbsp;Image&nbsp;Info&nbsp;Version</tt></td>
3287 <td align="left"><b>[Required]</b> &mdash; The version of the image info
3288 section. Currently always 0.</td>
3289 </tr>
3290 <tr>
3291 <td><tt>Objective-C&nbsp;Image&nbsp;Info&nbsp;Section</tt></td>
3292 <td align="left"><b>[Required]</b> &mdash; The section to place the
3293 metadata. Valid values are <tt>"__OBJC, __image_info, regular"</tt> for
3294 Objective-C ABI version 1, and <tt>"__DATA,__objc_imageinfo, regular,
3295 no_dead_strip"</tt> for Objective-C ABI version 2.</td>
3296 </tr>
3297 <tr>
3298 <td><tt>Objective-C&nbsp;Garbage&nbsp;Collection</tt></td>
3299 <td align="left"><b>[Required]</b> &mdash; Specifies whether garbage
3300 collection is supported or not. Valid values are 0, for no garbage
3301 collection, and 2, for garbage collection supported.</td>
3302 </tr>
3303 <tr>
3304 <td><tt>Objective-C&nbsp;GC&nbsp;Only</tt></td>
3305 <td align="left"><b>[Optional]</b> &mdash; Specifies that only garbage
3306 collection is supported. If present, its value must be 6. This flag
3307 requires that the <tt>Objective-C Garbage Collection</tt> flag have the
3308 value 2.</td>
3309 </tr>
3310 </tbody>
3311</table>
3312
3313<p>Some important flag interactions:</p>
3314
3315<ul>
3316 <li>If a module with <tt>Objective-C Garbage Collection</tt> set to 0 is
3317 merged with a module with <tt>Objective-C Garbage Collection</tt> set to
3318 2, then the resulting module has the <tt>Objective-C Garbage
3319 Collection</tt> flag set to 0.</li>
3320
3321 <li>A module with <tt>Objective-C Garbage Collection</tt> set to 0 cannot be
3322 merged with a module with <tt>Objective-C GC Only</tt> set to 6.</li>
3323</ul>
3324
3325</div>
3326
Bill Wendlingb9d75a92012-02-11 11:59:36 +0000[diff] [blame]3327</div>
3328
3329<!-- *********************************************************************** -->
3330<h2>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3331 <a name="intrinsic_globals">Intrinsic Global Variables</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3332</h2>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3333<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3334<div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3335<p>LLVM has a number of "magic" global variables that contain data that affect
3336code generation or other IR semantics. These are documented here. All globals
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3337of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
3338section and all globals that start with "<tt>llvm.</tt>" are reserved for use
3339by LLVM.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3340
3341<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3342<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3343<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3344</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3345
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3346<div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3347
3348<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
3349href="#linkage_appending">appending linkage</a>. This array contains a list of
3350pointers to global variables and functions which may optionally have a pointer
3351cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
3352
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3353<div class="doc_code">
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3354<pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3355@X = global i8 4
3356@Y = global i32 123
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3357
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3358@llvm.used = appending global [2 x i8*] [
3359 i8* @X,
3360 i8* bitcast (i32* @Y to i8*)
3361], section "llvm.metadata"
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3362</pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3363</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3364
3365<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3366 compiler, assembler, and linker are required to treat the symbol as if there
3367 is a reference to the global that it cannot see. For example, if a variable
3368 has internal linkage and no references other than that from
3369 the <tt>@llvm.used</tt> list, it cannot be deleted. This is commonly used to
3370 represent references from inline asms and other things the compiler cannot
3371 "see", and corresponds to "<tt>attribute((used))</tt>" in GNU C.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3372
3373<p>On some targets, the code generator must emit a directive to the assembler or
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3374 object file to prevent the assembler and linker from molesting the
3375 symbol.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3376
3377</div>
3378
3379<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3380<h3>
3381 <a name="intg_compiler_used">
3382 The '<tt>llvm.compiler.used</tt>' Global Variable
3383 </a>
3384</h3>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3385
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3386<div>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3387
3388<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3389 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
3390 touching the symbol. On targets that support it, this allows an intelligent
3391 linker to optimize references to the symbol without being impeded as it would
3392 be by <tt>@llvm.used</tt>.</p>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3393
3394<p>This is a rare construct that should only be used in rare circumstances, and
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3395 should not be exposed to source languages.</p>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3396
3397</div>
3398
3399<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3400<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3401<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3402</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3403
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3404<div>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3405
3406<div class="doc_code">
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3407<pre>
3408%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +0000[diff] [blame]3409@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3410</pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3411</div>
3412
3413<p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor
3414 functions and associated priorities. The functions referenced by this array
3415 will be called in ascending order of priority (i.e. lowest first) when the
3416 module is loaded. The order of functions with the same priority is not
3417 defined.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3418
3419</div>
3420
3421<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3422<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3423<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3424</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3425
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3426<div>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3427
3428<div class="doc_code">
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3429<pre>
3430%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +0000[diff] [blame]3431@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3432</pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3433</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3434
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3435<p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions
3436 and associated priorities. The functions referenced by this array will be
3437 called in descending order of priority (i.e. highest first) when the module
3438 is loaded. The order of functions with the same priority is not defined.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3439
3440</div>
3441
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3442</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3443
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]3444<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3445<h2><a name="instref">Instruction Reference</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3446<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3447
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3448<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3449
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3450<p>The LLVM instruction set consists of several different classifications of
3451 instructions: <a href="#terminators">terminator
3452 instructions</a>, <a href="#binaryops">binary instructions</a>,
3453 <a href="#bitwiseops">bitwise binary instructions</a>,
3454 <a href="#memoryops">memory instructions</a>, and
3455 <a href="#otherops">other instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3456
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3457<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3458<h3>
3459 <a name="terminators">Terminator Instructions</a>
3460</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3461
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3462<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3463
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3464<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3465 in a program ends with a "Terminator" instruction, which indicates which
3466 block should be executed after the current block is finished. These
3467 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3468 control flow, not values (the one exception being the
3469 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3470
Chris Lattner6445ecb2011-08-02 20:29:13 +0000[diff] [blame]3471<p>The terminator instructions are:
3472 '<a href="#i_ret"><tt>ret</tt></a>',
3473 '<a href="#i_br"><tt>br</tt></a>',
3474 '<a href="#i_switch"><tt>switch</tt></a>',
3475 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3476 '<a href="#i_invoke"><tt>invoke</tt></a>',
Chris Lattner6445ecb2011-08-02 20:29:13 +0000[diff] [blame]3477 '<a href="#i_resume"><tt>resume</tt></a>', and
3478 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3479
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3480<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3481<h4>
3482 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3483</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3484
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +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<h5>Syntax:</h5>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]3488<pre>
3489 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]3490 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3491</pre>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3492
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3493<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3494<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3495 a value) from a function back to the caller.</p>
3496
3497<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3498 value and then causes control flow, and one that just causes control flow to
3499 occur.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3500
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3501<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3502<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3503 return value. The type of the return value must be a
3504 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]3505
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3506<p>A function is not <a href="#wellformed">well formed</a> if it it has a
3507 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3508 value or a return value with a type that does not match its type, or if it
3509 has a void return type and contains a '<tt>ret</tt>' instruction with a
3510 return value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3511
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3512<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3513<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3514 the calling function's context. If the caller is a
3515 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3516 instruction after the call. If the caller was an
3517 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3518 the beginning of the "normal" destination block. If the instruction returns
3519 a value, that value shall set the call or invoke instruction's return
3520 value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3521
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3522<h5>Example:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3523<pre>
3524 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]3525 ret void <i>; Return from a void function</i>
Bill Wendling0a4bbbf2009-02-28 22:12:54 +0000[diff] [blame]3526 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]3527</pre>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +0000[diff] [blame]3528
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3529</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3530<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3531<h4>
3532 <a name="i_br">'<tt>br</tt>' Instruction</a>
3533</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3534
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3535<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3536
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3537<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3538<pre>
Bill Wendlingb3aa4712011-07-26 10:41:15 +0000[diff] [blame]3539 br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;
3540 br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3541</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3542
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3543<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3544<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3545 different basic block in the current function. There are two forms of this
3546 instruction, corresponding to a conditional branch and an unconditional
3547 branch.</p>
3548
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3549<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3550<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3551 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3552 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3553 target.</p>
3554
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3555<h5>Semantics:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3556<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3557 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3558 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3559 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3560
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3561<h5>Example:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]3562<pre>
3563Test:
3564 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3565 br i1 %cond, label %IfEqual, label %IfUnequal
3566IfEqual:
3567 <a href="#i_ret">ret</a> i32 1
3568IfUnequal:
3569 <a href="#i_ret">ret</a> i32 0
3570</pre>
3571
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3572</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3573
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3574<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3575<h4>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3576 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3577</h4>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3578
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3579<div>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3580
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3581<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3582<pre>
3583 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
3584</pre>
3585
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3586<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3587<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]3588 several different places. It is a generalization of the '<tt>br</tt>'
3589 instruction, allowing a branch to occur to one of many possible
3590 destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3591
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3592<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3593<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3594 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3595 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3596 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3597
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3598<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3599<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3600 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3601 is searched for the given value. If the value is found, control flow is
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]3602 transferred to the corresponding destination; otherwise, control flow is
3603 transferred to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3604
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3605<h5>Implementation:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3606<p>Depending on properties of the target machine and the particular
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3607 <tt>switch</tt> instruction, this instruction may be code generated in
3608 different ways. For example, it could be generated as a series of chained
3609 conditional branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3610
3611<h5>Example:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3612<pre>
3613 <i>; Emulate a conditional br instruction</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3614 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]3615 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3616
3617 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3618 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3619
3620 <i>; Implement a jump table:</i>
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]3621 switch i32 %val, label %otherwise [ i32 0, label %onzero
3622 i32 1, label %onone
3623 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3624</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3625
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3626</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3627
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3628
3629<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3630<h4>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3631 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3632</h4>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3633
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3634<div>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3635
3636<h5>Syntax:</h5>
3637<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3638 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]3639</pre>
3640
3641<h5>Overview:</h5>
3642
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3643<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3644 within the current function, whose address is specified by
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]3645 "<tt>address</tt>". Address must be derived from a <a
3646 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3647
3648<h5>Arguments:</h5>
3649
3650<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3651 rest of the arguments indicate the full set of possible destinations that the
3652 address may point to. Blocks are allowed to occur multiple times in the
3653 destination list, though this isn't particularly useful.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3654
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3655<p>This destination list is required so that dataflow analysis has an accurate
3656 understanding of the CFG.</p>
3657
3658<h5>Semantics:</h5>
3659
3660<p>Control transfers to the block specified in the address argument. All
3661 possible destination blocks must be listed in the label list, otherwise this
3662 instruction has undefined behavior. This implies that jumps to labels
3663 defined in other functions have undefined behavior as well.</p>
3664
3665<h5>Implementation:</h5>
3666
3667<p>This is typically implemented with a jump through a register.</p>
3668
3669<h5>Example:</h5>
3670<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3671 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3672</pre>
3673
3674</div>
3675
3676
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3677<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3678<h4>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3679 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3680</h4>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3681
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3682<div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3683
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3684<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3685<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]3686 &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]3687 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3688</pre>
3689
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3690<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3691<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3692 function, with the possibility of control flow transfer to either the
3693 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3694 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3695 control flow will return to the "normal" label. If the callee (or any
Bill Wendling7b9e5392012-02-06 21:57:33 +0000[diff] [blame]3696 indirect callees) returns via the "<a href="#i_resume"><tt>resume</tt></a>"
3697 instruction or other exception handling mechanism, control is interrupted and
3698 continued at the dynamically nearest "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3699
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3700<p>The '<tt>exception</tt>' label is a
3701 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3702 exception. As such, '<tt>exception</tt>' label is required to have the
3703 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
Chad Rosier85f5a1a2011-12-09 02:00:44 +0000[diff] [blame]3704 the information about the behavior of the program after unwinding
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3705 happens, as its first non-PHI instruction. The restrictions on the
3706 "<tt>landingpad</tt>" instruction's tightly couples it to the
3707 "<tt>invoke</tt>" instruction, so that the important information contained
3708 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3709 code motion.</p>
3710
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3711<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3712<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3713
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3714<ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3715 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3716 convention</a> the call should use. If none is specified, the call
3717 defaults to using C calling conventions.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]3718
3719 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3720 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3721 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]3722
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3723 <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]3724 function value being invoked. In most cases, this is a direct function
3725 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3726 off an arbitrary pointer to function value.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3727
3728 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3729 function to be invoked. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3730
3731 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]3732 signature argument types and parameter attributes. All arguments must be
3733 of <a href="#t_firstclass">first class</a> type. If the function
3734 signature indicates the function accepts a variable number of arguments,
3735 the extra arguments can be specified.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3736
3737 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3738 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3739
Bill Wendling7b9e5392012-02-06 21:57:33 +0000[diff] [blame]3740 <li>'<tt>exception label</tt>': the label reached when a callee returns via
3741 the <a href="#i_resume"><tt>resume</tt></a> instruction or other exception
3742 handling mechanism.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3743
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]3744 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3745 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3746 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3747</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3748
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3749<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3750<p>This instruction is designed to operate as a standard
3751 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3752 primary difference is that it establishes an association with a label, which
3753 is used by the runtime library to unwind the stack.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3754
3755<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3756 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3757 exception. Additionally, this is important for implementation of
3758 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3759
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3760<p>For the purposes of the SSA form, the definition of the value returned by the
3761 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3762 block to the "normal" label. If the callee unwinds then no return value is
3763 available.</p>
Dan Gohmanf96a4992009-05-22 21:47:08 +0000[diff] [blame]3764
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3765<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3766<pre>
Nick Lewyckyd703f652008-03-16 07:18:12 +0000[diff] [blame]3767 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3768 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckyd703f652008-03-16 07:18:12 +0000[diff] [blame]3769 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3770 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3771</pre>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3772
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3773</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3774
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3775 <!-- _______________________________________________________________________ -->
3776
3777<h4>
3778 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3779</h4>
3780
3781<div>
3782
3783<h5>Syntax:</h5>
3784<pre>
3785 resume &lt;type&gt; &lt;value&gt;
3786</pre>
3787
3788<h5>Overview:</h5>
3789<p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3790 successors.</p>
3791
3792<h5>Arguments:</h5>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3793<p>The '<tt>resume</tt>' instruction requires one argument, which must have the
Bill Wendlinge4ad50b2011-08-03 18:37:32 +0000[diff] [blame]3794 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3795 function.</p>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3796
3797<h5>Semantics:</h5>
3798<p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3799 (in-flight) exception whose unwinding was interrupted with
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3800 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3801
3802<h5>Example:</h5>
3803<pre>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3804 resume { i8*, i32 } %exn
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3805</pre>
3806
3807</div>
3808
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3809<!-- _______________________________________________________________________ -->
3810
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3811<h4>
3812 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3813</h4>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3814
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3815<div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3816
3817<h5>Syntax:</h5>
3818<pre>
3819 unreachable
3820</pre>
3821
3822<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3823<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3824 instruction is used to inform the optimizer that a particular portion of the
3825 code is not reachable. This can be used to indicate that the code after a
3826 no-return function cannot be reached, and other facts.</p>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3827
3828<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3829<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3830
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3831</div>
3832
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3833</div>
3834
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3835<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3836<h3>
3837 <a name="binaryops">Binary Operations</a>
3838</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3839
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3840<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3841
3842<p>Binary operators are used to do most of the computation in a program. They
3843 require two operands of the same type, execute an operation on them, and
3844 produce a single value. The operands might represent multiple data, as is
3845 the case with the <a href="#t_vector">vector</a> data type. The result value
3846 has the same type as its operands.</p>
3847
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3848<p>There are several different binary operators:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3849
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3850<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3851<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3852 <a name="i_add">'<tt>add</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3853</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3854
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3855<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3856
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3857<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3858<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3859 &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]3860 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3861 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3862 &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]3863</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3864
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3865<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3866<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3867
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3868<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3869<p>The two arguments to the '<tt>add</tt>' instruction must
3870 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3871 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3872
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3873<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3874<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3875
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3876<p>If the sum has unsigned overflow, the result returned is the mathematical
3877 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]3878
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3879<p>Because LLVM integers use a two's complement representation, this instruction
3880 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3881
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3882<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3883 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3884 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3885 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3886 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3887
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3888<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3889<pre>
3890 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3891</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3892
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3893</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3894
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3895<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3896<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3897 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3898</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3899
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3900<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3901
3902<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3903<pre>
3904 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3905</pre>
3906
3907<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3908<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3909
3910<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3911<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3912 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3913 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3914
3915<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3916<p>The value produced is the floating point sum of the two operands.</p>
3917
3918<h5>Example:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3919<pre>
3920 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3921</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3922
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3923</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3924
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3925<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3926<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3927 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3928</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3929
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3930<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3931
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3932<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3933<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3934 &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]3935 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3936 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3937 &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]3938</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3939
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3940<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3941<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3942 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3943
3944<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3945 '<tt>neg</tt>' instruction present in most other intermediate
3946 representations.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3947
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3948<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3949<p>The two arguments to the '<tt>sub</tt>' instruction must
3950 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3951 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3952
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3953<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3954<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3955
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3956<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3957 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3958 result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3959
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3960<p>Because LLVM integers use a two's complement representation, this instruction
3961 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3962
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3963<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3964 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3965 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3966 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3967 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3968
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3969<h5>Example:</h5>
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]3970<pre>
3971 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3972 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3973</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3974
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3975</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3976
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3977<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3978<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3979 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3980</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3981
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3982<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3983
3984<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3985<pre>
3986 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3987</pre>
3988
3989<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3990<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3991 operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3992
3993<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3994 '<tt>fneg</tt>' instruction present in most other intermediate
3995 representations.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3996
3997<h5>Arguments:</h5>
Bill Wendlingd9fe2982009-07-20 02:32:41 +0000[diff] [blame]3998<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3999 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4000 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4001
4002<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4003<p>The value produced is the floating point difference of the two operands.</p>
4004
4005<h5>Example:</h5>
4006<pre>
4007 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
4008 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
4009</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4010
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4011</div>
4012
4013<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4014<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4015 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4016</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4017
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4018<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4019
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4020<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4021<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]4022 &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]4023 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4024 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4025 &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]4026</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4027
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4028<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4029<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4030
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4031<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4032<p>The two arguments to the '<tt>mul</tt>' instruction must
4033 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
4034 integer values. Both arguments must have identical types.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4035
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4036<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4037<p>The value produced is the integer product of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4038
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4039<p>If the result of the multiplication has unsigned overflow, the result
4040 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
4041 width of the result.</p>
4042
4043<p>Because LLVM integers use a two's complement representation, and the result
4044 is the same width as the operands, this instruction returns the correct
4045 result for both signed and unsigned integers. If a full product
4046 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
4047 be sign-extended or zero-extended as appropriate to the width of the full
4048 product.</p>
4049
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]4050<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
4051 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
4052 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4053 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]4054 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]4055
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4056<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4057<pre>
4058 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4059</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4060
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4061</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4062
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4063<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4064<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4065 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4066</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4067
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4068<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4069
4070<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4071<pre>
4072 &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]4073</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4074
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4075<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4076<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4077
4078<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4079<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4080 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4081 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4082
4083<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4084<p>The value produced is the floating point product of the two operands.</p>
4085
4086<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4087<pre>
4088 &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]4089</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4090
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]4091</div>
4092
4093<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4094<h4>
4095 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
4096</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4097
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4098<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4099
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4100<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4101<pre>
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]4102 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4103 &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]4104</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4105
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4106<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4107<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4108
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4109<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4110<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4111 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4112 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4113
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4114<h5>Semantics:</h5>
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4115<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4116
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4117<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4118 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
4119
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4120<p>Division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4121
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]4122<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4123 <tt>udiv</tt> is a <a href="#poisonvalues">poison value</a> if %op1 is not a
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]4124 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
4125
4126
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4127<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4128<pre>
4129 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4130</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4131
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4132</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4133
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4134<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4135<h4>
4136 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
4137</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4138
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4139<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4140
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4141<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4142<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]4143 &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]4144 &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]4145</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4146
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4147<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4148<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4149
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4150<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4151<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4152 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4153 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4154
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4155<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4156<p>The value produced is the signed integer quotient of the two operands rounded
4157 towards zero.</p>
4158
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4159<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4160 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
4161
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4162<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4163 undefined behavior; this is a rare case, but can occur, for example, by doing
4164 a 32-bit division of -2147483648 by -1.</p>
4165
Dan Gohman9c5beed2009-07-22 00:04:19 +0000[diff] [blame]4166<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4167 <tt>sdiv</tt> is a <a href="#poisonvalues">poison value</a> if the result would
Dan Gohman38da9272010-07-11 00:08:34 +0000[diff] [blame]4168 be rounded.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]4169
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4170<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4171<pre>
4172 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4173</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4174
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4175</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4176
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4177<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4178<h4>
4179 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
4180</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4181
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4182<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4183
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4184<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4185<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]4186 &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]4187</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4188
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4189<h5>Overview:</h5>
4190<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4191
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4192<h5>Arguments:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]4193<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4194 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4195 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4196
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4197<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]4198<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4199
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4200<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4201<pre>
4202 &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]4203</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4204
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4205</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4206
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4207<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4208<h4>
4209 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
4210</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4211
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4212<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4213
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4214<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4215<pre>
4216 &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]4217</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4218
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4219<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4220<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
4221 division of its two arguments.</p>
4222
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4223<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4224<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4225 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4226 values. Both arguments must have identical types.</p>
4227
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4228<h5>Semantics:</h5>
4229<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4230 This instruction always performs an unsigned division to get the
4231 remainder.</p>
4232
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4233<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4234 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
4235
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4236<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4237
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4238<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4239<pre>
4240 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4241</pre>
4242
4243</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4244
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4245<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4246<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4247 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4248</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4249
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4250<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4251
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4252<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4253<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]4254 &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]4255</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4256
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4257<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4258<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
4259 division of its two operands. This instruction can also take
4260 <a href="#t_vector">vector</a> versions of the values in which case the
4261 elements must be integers.</p>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +0000[diff] [blame]4262
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4263<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4264<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4265 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4266 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4267
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4268<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4269<p>This instruction returns the <i>remainder</i> of a division (where the result
Duncan Sandsdea3a5e2011-03-07 09:12:24 +0000[diff] [blame]4270 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
4271 <i>modulo</i> operator (where the result is either zero or has the same sign
4272 as the divisor, <tt>op2</tt>) of a value.
4273 For more information about the difference,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4274 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
4275 Math Forum</a>. For a table of how this is implemented in various languages,
4276 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
4277 Wikipedia: modulo operation</a>.</p>
4278
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4279<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4280 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
4281
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4282<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4283 Overflow also leads to undefined behavior; this is a rare case, but can
4284 occur, for example, by taking the remainder of a 32-bit division of
4285 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
4286 lets srem be implemented using instructions that return both the result of
4287 the division and the remainder.)</p>
4288
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4289<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4290<pre>
4291 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4292</pre>
4293
4294</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4295
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4296<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4297<h4>
4298 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
4299</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4300
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4301<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4302
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4303<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4304<pre>
4305 &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]4306</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4307
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4308<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4309<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
4310 its two operands.</p>
4311
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4312<h5>Arguments:</h5>
4313<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4314 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4315 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4316
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4317<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4318<p>This instruction returns the <i>remainder</i> of a division. The remainder
4319 has the same sign as the dividend.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4320
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4321<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4322<pre>
4323 &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]4324</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4325
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4326</div>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]4327
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4328</div>
4329
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]4330<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4331<h3>
4332 <a name="bitwiseops">Bitwise Binary Operations</a>
4333</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4334
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4335<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4336
4337<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
4338 program. They are generally very efficient instructions and can commonly be
4339 strength reduced from other instructions. They require two operands of the
4340 same type, execute an operation on them, and produce a single value. The
4341 resulting value is the same type as its operands.</p>
4342
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4343<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4344<h4>
4345 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
4346</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4347
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4348<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4349
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4350<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4351<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4352 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4353 &lt;result&gt; = shl nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4354 &lt;result&gt; = shl nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4355 &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]4356</pre>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4357
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4358<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4359<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
4360 a specified number of bits.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4361
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4362<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4363<p>Both arguments to the '<tt>shl</tt>' instruction must be the
4364 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
4365 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4366
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4367<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4368<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
4369 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
4370 is (statically or dynamically) negative or equal to or larger than the number
4371 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4372 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4373 shift amount in <tt>op2</tt>.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4374
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4375<p>If the <tt>nuw</tt> keyword is present, then the shift produces a
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4376 <a href="#poisonvalues">poison value</a> if it shifts out any non-zero bits. If
Chris Lattner66298c12011-02-09 16:44:44 +0000[diff] [blame]4377 the <tt>nsw</tt> keyword is present, then the shift produces a
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4378 <a href="#poisonvalues">poison value</a> if it shifts out any bits that disagree
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4379 with the resultant sign bit. As such, NUW/NSW have the same semantics as
4380 they would if the shift were expressed as a mul instruction with the same
4381 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
4382
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4383<h5>Example:</h5>
4384<pre>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4385 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
4386 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
4387 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4388 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4389 &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]4390</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4391
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4392</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4393
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4394<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4395<h4>
4396 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
4397</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4398
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4399<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4400
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4401<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4402<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4403 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4404 &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]4405</pre>
4406
4407<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4408<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4409 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4410
4411<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4412<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4413 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4414 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4415
4416<h5>Semantics:</h5>
4417<p>This instruction always performs a logical shift right operation. The most
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4418 significant bits of the result will be filled with zero bits after the shift.
4419 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4420 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4421 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4422 shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4423
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4424<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4425 <tt>lshr</tt> is a <a href="#poisonvalues">poison value</a> if any of the bits
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4426 shifted out are non-zero.</p>
4427
4428
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4429<h5>Example:</h5>
4430<pre>
4431 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4432 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4433 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4434 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4435 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4436 &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]4437</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4438
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4439</div>
4440
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]4441<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4442<h4>
4443 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4444</h4>
4445
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4446<div>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4447
4448<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4449<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4450 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4451 &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]4452</pre>
4453
4454<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4455<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4456 operand shifted to the right a specified number of bits with sign
4457 extension.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4458
4459<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4460<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4461 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4462 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4463
4464<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4465<p>This instruction always performs an arithmetic shift right operation, The
4466 most significant bits of the result will be filled with the sign bit
4467 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4468 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4469 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4470 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4471
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4472<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4473 <tt>ashr</tt> is a <a href="#poisonvalues">poison value</a> if any of the bits
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4474 shifted out are non-zero.</p>
4475
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4476<h5>Example:</h5>
4477<pre>
4478 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4479 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4480 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4481 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4482 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4483 &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]4484</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4485
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4486</div>
4487
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4488<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4489<h4>
4490 <a name="i_and">'<tt>and</tt>' Instruction</a>
4491</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4492
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4493<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4494
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4495<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4496<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]4497 &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]4498</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4499
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4500<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4501<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4502 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4503
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4504<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4505<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4506 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4507 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4508
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4509<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4510<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4511
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4512<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4513 <tbody>
4514 <tr>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4515 <th>In0</th>
4516 <th>In1</th>
4517 <th>Out</th>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4518 </tr>
4519 <tr>
4520 <td>0</td>
4521 <td>0</td>
4522 <td>0</td>
4523 </tr>
4524 <tr>
4525 <td>0</td>
4526 <td>1</td>
4527 <td>0</td>
4528 </tr>
4529 <tr>
4530 <td>1</td>
4531 <td>0</td>
4532 <td>0</td>
4533 </tr>
4534 <tr>
4535 <td>1</td>
4536 <td>1</td>
4537 <td>1</td>
4538 </tr>
4539 </tbody>
4540</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4541
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4542<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4543<pre>
4544 &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]4545 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4546 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4547</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4548</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4549<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4550<h4>
4551 <a name="i_or">'<tt>or</tt>' Instruction</a>
4552</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4553
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4554<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4555
4556<h5>Syntax:</h5>
4557<pre>
4558 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4559</pre>
4560
4561<h5>Overview:</h5>
4562<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4563 two operands.</p>
4564
4565<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4566<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4567 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4568 values. Both arguments must have identical types.</p>
4569
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4570<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4571<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4572
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4573<table border="1" cellspacing="0" cellpadding="4">
4574 <tbody>
4575 <tr>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4576 <th>In0</th>
4577 <th>In1</th>
4578 <th>Out</th>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4579 </tr>
4580 <tr>
4581 <td>0</td>
4582 <td>0</td>
4583 <td>0</td>
4584 </tr>
4585 <tr>
4586 <td>0</td>
4587 <td>1</td>
4588 <td>1</td>
4589 </tr>
4590 <tr>
4591 <td>1</td>
4592 <td>0</td>
4593 <td>1</td>
4594 </tr>
4595 <tr>
4596 <td>1</td>
4597 <td>1</td>
4598 <td>1</td>
4599 </tr>
4600 </tbody>
4601</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4602
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4603<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4604<pre>
4605 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4606 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4607 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4608</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4609
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4610</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4611
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4612<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4613<h4>
4614 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4615</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4616
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4617<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4618
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4619<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4620<pre>
4621 &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]4622</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4623
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4624<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4625<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4626 its two operands. The <tt>xor</tt> is used to implement the "one's
4627 complement" operation, which is the "~" operator in C.</p>
4628
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4629<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4630<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4631 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4632 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4633
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4634<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4635<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4636
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4637<table border="1" cellspacing="0" cellpadding="4">
4638 <tbody>
4639 <tr>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4640 <th>In0</th>
4641 <th>In1</th>
4642 <th>Out</th>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4643 </tr>
4644 <tr>
4645 <td>0</td>
4646 <td>0</td>
4647 <td>0</td>
4648 </tr>
4649 <tr>
4650 <td>0</td>
4651 <td>1</td>
4652 <td>1</td>
4653 </tr>
4654 <tr>
4655 <td>1</td>
4656 <td>0</td>
4657 <td>1</td>
4658 </tr>
4659 <tr>
4660 <td>1</td>
4661 <td>1</td>
4662 <td>0</td>
4663 </tr>
4664 </tbody>
4665</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4666
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4667<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4668<pre>
4669 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4670 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4671 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4672 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4673</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4674
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4675</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4676
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4677</div>
4678
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4679<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4680<h3>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4681 <a name="vectorops">Vector Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4682</h3>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4683
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4684<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4685
4686<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4687 target-independent manner. These instructions cover the element-access and
4688 vector-specific operations needed to process vectors effectively. While LLVM
4689 does directly support these vector operations, many sophisticated algorithms
4690 will want to use target-specific intrinsics to take full advantage of a
4691 specific target.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4692
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4693<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4694<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4695 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4696</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4697
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4698<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4699
4700<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4701<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4702 &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]4703</pre>
4704
4705<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4706<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4707 from a vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4708
4709
4710<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4711<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4712 of <a href="#t_vector">vector</a> type. The second operand is an index
4713 indicating the position from which to extract the element. The index may be
4714 a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4715
4716<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4717<p>The result is a scalar of the same type as the element type of
4718 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4719 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4720 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4721
4722<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4723<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4724 &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]4725</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4726
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4727</div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4728
4729<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4730<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4731 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4732</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4733
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4734<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4735
4736<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4737<pre>
Dan Gohmanf3480b92008-05-12 23:38:42 +0000[diff] [blame]4738 &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]4739</pre>
4740
4741<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4742<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4743 vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4744
4745<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4746<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4747 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4748 whose type must equal the element type of the first operand. The third
4749 operand is an index indicating the position at which to insert the value.
4750 The index may be a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4751
4752<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4753<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4754 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4755 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4756 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4757
4758<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4759<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4760 &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]4761</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4762
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4763</div>
4764
4765<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4766<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4767 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4768</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4769
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4770<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4771
4772<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4773<pre>
Mon P Wangaeb06d22008-11-10 04:46:22 +0000[diff] [blame]4774 &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]4775</pre>
4776
4777<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4778<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4779 from two input vectors, returning a vector with the same element type as the
4780 input and length that is the same as the shuffle mask.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4781
4782<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4783<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
Duncan Sandsb5a1bf42012-06-14 14:58:28 +0000[diff] [blame]4784 with the same type. The third argument is a shuffle mask whose
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4785 element type is always 'i32'. The result of the instruction is a vector
4786 whose length is the same as the shuffle mask and whose element type is the
4787 same as the element type of the first two operands.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4788
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4789<p>The shuffle mask operand is required to be a constant vector with either
4790 constant integer or undef values.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4791
4792<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4793<p>The elements of the two input vectors are numbered from left to right across
4794 both of the vectors. The shuffle mask operand specifies, for each element of
4795 the result vector, which element of the two input vectors the result element
4796 gets. The element selector may be undef (meaning "don't care") and the
4797 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4798
4799<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4800<pre>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4801 &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]4802 &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]4803 &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]4804 &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]4805 &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]4806 &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]4807 &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]4808 &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]4809</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4810
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4811</div>
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]4812
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4813</div>
4814
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4815<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4816<h3>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4817 <a name="aggregateops">Aggregate Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4818</h3>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4819
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4820<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4821
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4822<p>LLVM supports several instructions for working with
4823 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4824
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4825<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4826<h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4827 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4828</h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4829
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4830<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4831
4832<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4833<pre>
4834 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4835</pre>
4836
4837<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4838<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4839 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4840
4841<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4842<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4843 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4844 <a href="#t_array">array</a> type. The operands are constant indices to
4845 specify which value to extract in a similar manner as indices in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4846 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Frits van Bommel13242892010-12-05 20:54:38 +0000[diff] [blame]4847 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4848 <ul>
4849 <li>Since the value being indexed is not a pointer, the first index is
4850 omitted and assumed to be zero.</li>
4851 <li>At least one index must be specified.</li>
4852 <li>Not only struct indices but also array indices must be in
4853 bounds.</li>
4854 </ul>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4855
4856<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4857<p>The result is the value at the position in the aggregate specified by the
4858 index operands.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4859
4860<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4861<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4862 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4863</pre>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4864
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4865</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4866
4867<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4868<h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4869 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4870</h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4871
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4872<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4873
4874<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4875<pre>
Bill Wendling194229e2011-07-26 20:42:28 +0000[diff] [blame]4876 &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]4877</pre>
4878
4879<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4880<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4881 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4882
4883<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4884<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4885 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4886 <a href="#t_array">array</a> type. The second operand is a first-class
4887 value to insert. The following operands are constant indices indicating
4888 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]4889 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4890 value to insert must have the same type as the value identified by the
4891 indices.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4892
4893<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4894<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4895 that of <tt>val</tt> except that the value at the position specified by the
4896 indices is that of <tt>elt</tt>.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4897
4898<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4899<pre>
Chris Lattner8645d1a2011-05-22 07:18:08 +0000[diff] [blame]4900 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4901 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4902 %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]4903</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4904
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4905</div>
4906
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4907</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4908
4909<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4910<h3>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]4911 <a name="memoryops">Memory Access and Addressing Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4912</h3>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4913
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4914<div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4915
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4916<p>A key design point of an SSA-based representation is how it represents
4917 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]4918 very simple. This section describes how to read, write, and allocate
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4919 memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4920
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4921<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4922<h4>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4923 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4924</h4>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4925
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4926<div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4927
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4928<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4929<pre>
Dan Gohmanf75a7d32010-05-28 01:14:11 +0000[diff] [blame]4930 &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]4931</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4932
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4933<h5>Overview:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]4934<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4935 currently executing function, to be automatically released when this function
4936 returns to its caller. The object is always allocated in the generic address
4937 space (address space zero).</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4938
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4939<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4940<p>The '<tt>alloca</tt>' instruction
4941 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4942 runtime stack, returning a pointer of the appropriate type to the program.
4943 If "NumElements" is specified, it is the number of elements allocated,
4944 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4945 specified, the value result of the allocation is guaranteed to be aligned to
4946 at least that boundary. If not specified, or if zero, the target can choose
4947 to align the allocation on any convenient boundary compatible with the
4948 type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4949
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4950<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4951
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4952<h5>Semantics:</h5>
Bill Wendling871eb0a2009-05-08 20:49:29 +0000[diff] [blame]4953<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4954 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4955 memory is automatically released when the function returns. The
4956 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4957 variables that must have an address available. When the function returns
4958 (either with the <tt><a href="#i_ret">ret</a></tt>
Bill Wendling7b9e5392012-02-06 21:57:33 +0000[diff] [blame]4959 or <tt><a href="#i_resume">resume</a></tt> instructions), the memory is
Nick Lewycky84a1d232012-02-29 08:26:44 +0000[diff] [blame]4960 reclaimed. Allocating zero bytes is legal, but the result is undefined.
4961 The order in which memory is allocated (ie., which way the stack grows) is
Nick Lewycky75d05e62012-03-18 09:35:50 +0000[diff] [blame]4962 not specified.</p>
Nick Lewycky84a1d232012-02-29 08:26:44 +0000[diff] [blame]4963
4964<p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4965
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4966<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4967<pre>
Dan Gohman81e21672009-01-04 23:49:44 +0000[diff] [blame]4968 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4969 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4970 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4971 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4972</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4973
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4974</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4975
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4976<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4977<h4>
4978 <a name="i_load">'<tt>load</tt>' Instruction</a>
4979</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4980
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4981<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4982
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4983<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4984<pre>
Pete Cooperf95acc62012-02-10 18:13:54 +0000[diff] [blame]4985 &lt;result&gt; = load [volatile] &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;][, !invariant.load !&lt;index&gt;]
Eli Friedmanf03bb262011-08-12 22:50:01 +0000[diff] [blame]4986 &lt;result&gt; = load atomic [volatile] &lt;ty&gt;* &lt;pointer&gt; [singlethread] &lt;ordering&gt;, align &lt;alignment&gt;
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4987 !&lt;index&gt; = !{ i32 1 }
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4988</pre>
4989
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4990<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4991<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4992
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4993<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4994<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4995 from which to load. The pointer must point to
4996 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4997 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]4998 number or order of execution of this <tt>load</tt> with other <a
4999 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5000
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]5001<p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
5002 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
5003 argument. The <code>release</code> and <code>acq_rel</code> orderings are
5004 not valid on <code>load</code> instructions. Atomic loads produce <a
5005 href="#memorymodel">defined</a> results when they may see multiple atomic
5006 stores. The type of the pointee must be an integer type whose bit width
5007 is a power of two greater than or equal to eight and less than or equal
5008 to a target-specific size limit. <code>align</code> must be explicitly
5009 specified on atomic loads, and the load has undefined behavior if the
5010 alignment is not set to a value which is at least the size in bytes of
5011 the pointee. <code>!nontemporal</code> does not have any defined semantics
5012 for atomic loads.</p>
5013
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]5014<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5015 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]5016 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5017 alignment for the target. It is the responsibility of the code emitter to
5018 ensure that the alignment information is correct. Overestimating the
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]5019 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5020 produce less efficient code. An alignment of 1 is always safe.</p>
5021
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]5022<p>The optional <tt>!nontemporal</tt> metadata must reference a single
5023 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]5024 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]5025 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
5026 and code generator that this load is not expected to be reused in the cache.
5027 The code generator may select special instructions to save cache bandwidth,
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]5028 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]5029
Pete Cooperf95acc62012-02-10 18:13:54 +0000[diff] [blame]5030<p>The optional <tt>!invariant.load</tt> metadata must reference a single
5031 metatadata name &lt;index&gt; corresponding to a metadata node with no
5032 entries. The existence of the <tt>!invariant.load</tt> metatadata on the
5033 instruction tells the optimizer and code generator that this load address
5034 points to memory which does not change value during program execution.
5035 The optimizer may then move this load around, for example, by hoisting it
5036 out of loops using loop invariant code motion.</p>
5037
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5038<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5039<p>The location of memory pointed to is loaded. If the value being loaded is of
5040 scalar type then the number of bytes read does not exceed the minimum number
5041 of bytes needed to hold all bits of the type. For example, loading an
5042 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
5043 <tt>i20</tt> with a size that is not an integral number of bytes, the result
5044 is undefined if the value was not originally written using a store of the
5045 same type.</p>
5046
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5047<h5>Examples:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5048<pre>
5049 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
5050 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5051 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5052</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5053
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5054</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5055
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5056<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5057<h4>
5058 <a name="i_store">'<tt>store</tt>' Instruction</a>
5059</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5060
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5061<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5062
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5063<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5064<pre>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]5065 store [volatile] &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;] <i>; yields {void}</i>
5066 store atomic [volatile] &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt; [singlethread] &lt;ordering&gt;, align &lt;alignment&gt; <i>; yields {void}</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5067</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5068
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5069<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5070<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5071
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5072<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5073<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
5074 and an address at which to store it. The type of the
5075 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
5076 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]5077 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
5078 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
5079 order of execution of this <tt>store</tt> with other <a
5080 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5081
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]5082<p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
5083 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
5084 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
5085 valid on <code>store</code> instructions. Atomic loads produce <a
5086 href="#memorymodel">defined</a> results when they may see multiple atomic
5087 stores. The type of the pointee must be an integer type whose bit width
5088 is a power of two greater than or equal to eight and less than or equal
5089 to a target-specific size limit. <code>align</code> must be explicitly
5090 specified on atomic stores, and the store has undefined behavior if the
5091 alignment is not set to a value which is at least the size in bytes of
5092 the pointee. <code>!nontemporal</code> does not have any defined semantics
5093 for atomic stores.</p>
5094
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5095<p>The optional constant "align" argument specifies the alignment of the
5096 operation (that is, the alignment of the memory address). A value of 0 or an
5097 omitted "align" argument means that the operation has the preferential
5098 alignment for the target. It is the responsibility of the code emitter to
5099 ensure that the alignment information is correct. Overestimating the
5100 alignment results in an undefined behavior. Underestimating the alignment may
5101 produce less efficient code. An alignment of 1 is always safe.</p>
5102
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]5103<p>The optional !nontemporal metadata must reference a single metatadata
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]5104 name &lt;index&gt; corresponding to a metadata node with one i32 entry of
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]5105 value 1. The existence of the !nontemporal metatadata on the
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]5106 instruction tells the optimizer and code generator that this load is
5107 not expected to be reused in the cache. The code generator may
5108 select special instructions to save cache bandwidth, such as the
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]5109 MOVNT instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]5110
5111
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]5112<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5113<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
5114 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
5115 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
5116 does not exceed the minimum number of bytes needed to hold all bits of the
5117 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
5118 writing a value of a type like <tt>i20</tt> with a size that is not an
5119 integral number of bytes, it is unspecified what happens to the extra bits
5120 that do not belong to the type, but they will typically be overwritten.</p>
5121
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5122<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5123<pre>
5124 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8c6c72d2007-10-22 05:10:05 +0000[diff] [blame]5125 store i32 3, i32* %ptr <i>; yields {void}</i>
5126 %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]5127</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5128
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]5129</div>
5130
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]5131<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]5132<h4>
5133<a name="i_fence">'<tt>fence</tt>' Instruction</a>
5134</h4>
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]5135
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]5136<div>
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]5137
5138<h5>Syntax:</h5>
5139<pre>
5140 fence [singlethread] &lt;ordering&gt; <i>; yields {void}</i>
5141</pre>
5142
5143<h5>Overview:</h5>
5144<p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
5145between operations.</p>
5146
5147<h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
5148href="#ordering">ordering</a> argument which defines what
5149<i>synchronizes-with</i> edges they add. They can only be given
5150<code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
5151<code>seq_cst</code> orderings.</p>
5152
5153<h5>Semantics:</h5>
5154<p>A fence <var>A</var> which has (at least) <code>release</code> ordering
5155semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
5156<code>acquire</code> ordering semantics if and only if there exist atomic
5157operations <var>X</var> and <var>Y</var>, both operating on some atomic object
5158<var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
5159<var>X</var> modifies <var>M</var> (either directly or through some side effect
5160of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
5161<var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
5162<i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
5163than an explicit <code>fence</code>, one (but not both) of the atomic operations
5164<var>X</var> or <var>Y</var> might provide a <code>release</code> or
5165<code>acquire</code> (resp.) ordering constraint and still
5166<i>synchronize-with</i> the explicit <code>fence</code> and establish the
5167<i>happens-before</i> edge.</p>
5168
5169<p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
5170having both <code>acquire</code> and <code>release</code> semantics specified
5171above, participates in the global program order of other <code>seq_cst</code>
5172operations and/or fences.</p>
5173
5174<p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
5175specifies that the fence only synchronizes with other fences in the same
5176thread. (This is useful for interacting with signal handlers.)</p>
5177
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]5178<h5>Example:</h5>
5179<pre>
5180 fence acquire <i>; yields {void}</i>
5181 fence singlethread seq_cst <i>; yields {void}</i>
5182</pre>
5183
5184</div>
5185
5186<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]5187<h4>
5188<a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
5189</h4>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5190
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]5191<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5192
5193<h5>Syntax:</h5>
5194<pre>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]5195 cmpxchg [volatile] &lt;ty&gt;* &lt;pointer&gt;, &lt;ty&gt; &lt;cmp&gt;, &lt;ty&gt; &lt;new&gt; [singlethread] &lt;ordering&gt; <i>; yields {ty}</i>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5196</pre>
5197
5198<h5>Overview:</h5>
5199<p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
5200It loads a value in memory and compares it to a given value. If they are
5201equal, it stores a new value into the memory.</p>
5202
5203<h5>Arguments:</h5>
5204<p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
5205address to operate on, a value to compare to the value currently be at that
5206address, and a new value to place at that address if the compared values are
5207equal. The type of '<var>&lt;cmp&gt;</var>' must be an integer type whose
5208bit width is a power of two greater than or equal to eight and less than
5209or equal to a target-specific size limit. '<var>&lt;cmp&gt;</var>' and
5210'<var>&lt;new&gt;</var>' must have the same type, and the type of
5211'<var>&lt;pointer&gt;</var>' must be a pointer to that type. If the
5212<code>cmpxchg</code> is marked as <code>volatile</code>, then the
5213optimizer is not allowed to modify the number or order of execution
5214of this <code>cmpxchg</code> with other <a href="#volatile">volatile
5215operations</a>.</p>
5216
5217<!-- FIXME: Extend allowed types. -->
5218
5219<p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
5220<code>cmpxchg</code> synchronizes with other atomic operations.</p>
5221
5222<p>The optional "<code>singlethread</code>" argument declares that the
5223<code>cmpxchg</code> is only atomic with respect to code (usually signal
5224handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
5225cmpxchg is atomic with respect to all other code in the system.</p>
5226
5227<p>The pointer passed into cmpxchg must have alignment greater than or equal to
5228the size in memory of the operand.
5229
5230<h5>Semantics:</h5>
5231<p>The contents of memory at the location specified by the
5232'<tt>&lt;pointer&gt;</tt>' operand is read and compared to
5233'<tt>&lt;cmp&gt;</tt>'; if the read value is the equal,
5234'<tt>&lt;new&gt;</tt>' is written. The original value at the location
5235is returned.
5236
5237<p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
5238purpose of identifying <a href="#release_sequence">release sequences</a>. A
5239failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
5240parameter determined by dropping any <code>release</code> part of the
5241<code>cmpxchg</code>'s ordering.</p>
5242
5243<!--
5244FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
5245optimization work on ARM.)
5246
5247FIXME: Is a weaker ordering constraint on failure helpful in practice?
5248-->
5249
5250<h5>Example:</h5>
5251<pre>
5252entry:
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]5253 %orig = atomic <a href="#i_load">load</a> i32* %ptr unordered <i>; yields {i32}</i>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5254 <a href="#i_br">br</a> label %loop
5255
5256loop:
5257 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
5258 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]5259 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5260 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
5261 <a href="#i_br">br</a> i1 %success, label %done, label %loop
5262
5263done:
5264 ...
5265</pre>
5266
5267</div>
5268
5269<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]5270<h4>
5271<a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
5272</h4>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5273
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]5274<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5275
5276<h5>Syntax:</h5>
5277<pre>
Eli Friedmanf03bb262011-08-12 22:50:01 +0000[diff] [blame]5278 atomicrmw [volatile] &lt;operation&gt; &lt;ty&gt;* &lt;pointer&gt;, &lt;ty&gt; &lt;value&gt; [singlethread] &lt;ordering&gt; <i>; yields {ty}</i>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]5279</pre>
5280
5281<h5>Overview:</h5>
5282<p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
5283
5284<h5>Arguments:</h5>
5285<p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
5286operation to apply, an address whose value to modify, an argument to the
5287operation. The operation must be one of the following keywords:</p>
5288<ul>
5289 <li>xchg</li>
5290 <li>add</li>
5291 <li>sub</li>
5292 <li>and</li>
5293 <li>nand</li>
5294 <li>or</li>
5295 <li>xor</li>
5296 <li>max</li>
5297 <li>min</li>
5298 <li>umax</li>
5299 <li>umin</li>
5300</ul>
5301
5302<p>The type of '<var>&lt;value&gt;</var>' must be an integer type whose
5303bit width is a power of two greater than or equal to eight and less than
5304or equal to a target-specific size limit. The type of the
5305'<code>&lt;pointer&gt;</code>' operand must be a pointer to that type.
5306If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
5307optimizer is not allowed to modify the number or order of execution of this
5308<code>atomicrmw</code> with other <a href="#volatile">volatile
5309 operations</a>.</p>
5310
5311<!-- FIXME: Extend allowed types. -->
5312
5313<h5>Semantics:</h5>
5314<p>The contents of memory at the location specified by the
5315'<tt>&lt;pointer&gt;</tt>' operand are atomically read, modified, and written
5316back. The original value at the location is returned. The modification is
5317specified by the <var>operation</var> argument:</p>
5318
5319<ul>
5320 <li>xchg: <code>*ptr = val</code></li>
5321 <li>add: <code>*ptr = *ptr + val</code></li>
5322 <li>sub: <code>*ptr = *ptr - val</code></li>
5323 <li>and: <code>*ptr = *ptr &amp; val</code></li>
5324 <li>nand: <code>*ptr = ~(*ptr &amp; val)</code></li>
5325 <li>or: <code>*ptr = *ptr | val</code></li>
5326 <li>xor: <code>*ptr = *ptr ^ val</code></li>
5327 <li>max: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using a signed comparison)</li>
5328 <li>min: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using a signed comparison)</li>
5329 <li>umax: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using an unsigned comparison)</li>
5330 <li>umin: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using an unsigned comparison)</li>
5331</ul>
5332
5333<h5>Example:</h5>
5334<pre>
5335 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
5336</pre>
5337
5338</div>
5339
5340<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5341<h4>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5342 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5343</h4>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5344
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5345<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5346
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5347<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5348<pre>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5349 &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]5350 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5351 &lt;result&gt; = getelementptr &lt;ptr vector&gt; ptrval, &lt;vector index type&gt; idx
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5352</pre>
5353
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5354<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5355<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5356 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
5357 It performs address calculation only and does not access memory.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5358
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5359<h5>Arguments:</h5>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5360<p>The first argument is always a pointer or a vector of pointers,
5361 and forms the basis of the
Chris Lattnerc8eef442009-07-29 06:44:13 +0000[diff] [blame]5362 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5363 elements of the aggregate object are indexed. The interpretation of each
5364 index is dependent on the type being indexed into. The first index always
5365 indexes the pointer value given as the first argument, the second index
5366 indexes a value of the type pointed to (not necessarily the value directly
5367 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5368 indexed into must be a pointer value, subsequent types can be arrays,
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]5369 vectors, and structs. Note that subsequent types being indexed into
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5370 can never be pointers, since that would require loading the pointer before
5371 continuing calculation.</p>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5372
5373<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]5374 When indexing into a (optionally packed) structure, only <tt>i32</tt>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5375 integer <b>constants</b> are allowed. When indexing into an array, pointer
5376 or vector, integers of any width are allowed, and they are not required to be
Eli Friedman266246c2011-08-12 23:37:55 +0000[diff] [blame]5377 constant. These integers are treated as signed values where relevant.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5378
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5379<p>For example, let's consider a C code fragment and how it gets compiled to
5380 LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5381
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]5382<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5383struct RT {
5384 char A;
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]5385 int B[10][20];
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5386 char C;
5387};
5388struct ST {
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]5389 int X;
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5390 double Y;
5391 struct RT Z;
5392};
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5393
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]5394int *foo(struct ST *s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5395 return &amp;s[1].Z.B[5][13];
5396}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5397</pre>
5398
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5399<p>The LLVM code generated by Clang is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5400
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]5401<pre class="doc_code">
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5402%struct.RT = <a href="#namedtypes">type</a> { i8, [10 x [20 x i32]], i8 }
5403%struct.ST = <a href="#namedtypes">type</a> { i32, double, %struct.RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5404
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5405define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5406entry:
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5407 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5408 ret i32* %arrayidx
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5409}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5410</pre>
5411
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5412<h5>Semantics:</h5>
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5413<p>In the example above, the first index is indexing into the
5414 '<tt>%struct.ST*</tt>' type, which is a pointer, yielding a
5415 '<tt>%struct.ST</tt>' = '<tt>{ i32, double, %struct.RT }</tt>' type, a
5416 structure. The second index indexes into the third element of the structure,
5417 yielding a '<tt>%struct.RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]], i8 }</tt>'
5418 type, another structure. The third index indexes into the second element of
5419 the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an array. The
5420 two dimensions of the array are subscripted into, yielding an '<tt>i32</tt>'
5421 type. The '<tt>getelementptr</tt>' instruction returns a pointer to this
5422 element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5423
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5424<p>Note that it is perfectly legal to index partially through a structure,
5425 returning a pointer to an inner element. Because of this, the LLVM code for
5426 the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5427
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5428<pre class="doc_code">
5429define i32* @foo(%struct.ST* %s) {
5430 %t1 = getelementptr %struct.ST* %s, i32 1 <i>; yields %struct.ST*:%t1</i>
5431 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 <i>; yields %struct.RT*:%t2</i>
5432 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
5433 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5434 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5435 ret i32* %t5
5436}
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5437</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]5438
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5439<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]5440 <tt>getelementptr</tt> is a <a href="#poisonvalues">poison value</a> if the
Dan Gohman27ef9972010-04-23 15:23:32 +0000[diff] [blame]5441 base pointer is not an <i>in bounds</i> address of an allocated object,
5442 or if any of the addresses that would be formed by successive addition of
5443 the offsets implied by the indices to the base address with infinitely
Eli Friedman266246c2011-08-12 23:37:55 +0000[diff] [blame]5444 precise signed arithmetic are not an <i>in bounds</i> address of that
5445 allocated object. The <i>in bounds</i> addresses for an allocated object
5446 are all the addresses that point into the object, plus the address one
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5447 byte past the end.
5448 In cases where the base is a vector of pointers the <tt>inbounds</tt> keyword
5449 applies to each of the computations element-wise. </p>
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5450
5451<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
Eli Friedman266246c2011-08-12 23:37:55 +0000[diff] [blame]5452 the base address with silently-wrapping two's complement arithmetic. If the
5453 offsets have a different width from the pointer, they are sign-extended or
5454 truncated to the width of the pointer. The result value of the
5455 <tt>getelementptr</tt> may be outside the object pointed to by the base
5456 pointer. The result value may not necessarily be used to access memory
5457 though, even if it happens to point into allocated storage. See the
5458 <a href="#pointeraliasing">Pointer Aliasing Rules</a> section for more
5459 information.</p>
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5460
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5461<p>The getelementptr instruction is often confusing. For some more insight into
5462 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]5463
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5464<h5>Example:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5465<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5466 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5467 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5468 <i>; yields i8*:vptr</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5469 %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]5470 <i>; yields i8*:eptr</i>
5471 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta9f805c22009-04-25 07:27:44 +0000[diff] [blame]5472 <i>; yields i32*:iptr</i>
Sanjiv Gupta16ffa802009-04-24 16:38:13 +0000[diff] [blame]5473 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5474</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5475
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5476<p>In cases where the pointer argument is a vector of pointers, only a
5477 single index may be used, and the number of vector elements has to be
5478 the same. For example: </p>
5479<pre class="doc_code">
5480 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
5481</pre>
5482
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5483</div>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]5484
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5485</div>
5486
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5487<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5488<h3>
5489 <a name="convertops">Conversion Operations</a>
5490</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5491
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5492<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5493
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5494<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5495 which all take a single operand and a type. They perform various bit
5496 conversions on the operand.</p>
5497
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5498<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5499<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5500 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5501</h4>
5502
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5503<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5504
5505<h5>Syntax:</h5>
5506<pre>
5507 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5508</pre>
5509
5510<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5511<p>The '<tt>trunc</tt>' instruction truncates its operand to the
5512 type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5513
5514<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5515<p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5516 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5517 of the same number of integers.
5518 The bit size of the <tt>value</tt> must be larger than
5519 the bit size of the destination type, <tt>ty2</tt>.
5520 Equal sized types are not allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5521
5522<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5523<p>The '<tt>trunc</tt>' instruction truncates the high order bits
5524 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5525 source size must be larger than the destination size, <tt>trunc</tt> cannot
5526 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5527
5528<h5>Example:</h5>
5529<pre>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5530 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5531 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5532 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5533 %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]5534</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5535
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5536</div>
5537
5538<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5539<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5540 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5541</h4>
5542
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5543<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5544
5545<h5>Syntax:</h5>
5546<pre>
5547 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5548</pre>
5549
5550<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5551<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5552 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5553
5554
5555<h5>Arguments:</h5>
Nadav Rotemed9b9342011-02-20 12:37:50 +0000[diff] [blame]5556<p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5557 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5558 of the same number of integers.
5559 The bit size of the <tt>value</tt> must be smaller than
5560 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5561 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5562
5563<h5>Semantics:</h5>
5564<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]5565 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5566
Reid Spencerb5929522007-01-12 15:46:11 +0000[diff] [blame]5567<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]5568
5569<h5>Example:</h5>
5570<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5571 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5572 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Nadav Rotemed9b9342011-02-20 12:37:50 +0000[diff] [blame]5573 %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]5574</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5575
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5576</div>
5577
5578<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5579<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5580 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5581</h4>
5582
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5583<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5584
5585<h5>Syntax:</h5>
5586<pre>
5587 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5588</pre>
5589
5590<h5>Overview:</h5>
5591<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5592
5593<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5594<p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5595 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5596 of the same number of integers.
5597 The bit size of the <tt>value</tt> must be smaller than
5598 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5599 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5600
5601<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5602<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5603 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5604 of the type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5605
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5606<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]5607
5608<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5609<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5610 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5611 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5612 %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]5613</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5614
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5615</div>
5616
5617<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5618<h4>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5619 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5620</h4>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5621
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5622<div>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5623
5624<h5>Syntax:</h5>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5625<pre>
5626 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5627</pre>
5628
5629<h5>Overview:</h5>
5630<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5631 <tt>ty2</tt>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5632
5633<h5>Arguments:</h5>
5634<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5635 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5636 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]5637 <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]5638 <i>no-op cast</i>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5639
5640<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5641<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5642 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5643 <a href="#t_floating">floating point</a> type. If the value cannot fit
5644 within the destination type, <tt>ty2</tt>, then the results are
5645 undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5646
5647<h5>Example:</h5>
5648<pre>
5649 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5650 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5651</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5652
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5653</div>
5654
5655<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5656<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5657 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5658</h4>
5659
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5660<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5661
5662<h5>Syntax:</h5>
5663<pre>
5664 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5665</pre>
5666
5667<h5>Overview:</h5>
5668<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]5669 floating point value.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5670
5671<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5672<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5673 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5674 a <a href="#t_floating">floating point</a> type to cast it to. The source
5675 type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5676
5677<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5678<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5679 <a href="#t_floating">floating point</a> type to a larger
5680 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5681 used to make a <i>no-op cast</i> because it always changes bits. Use
5682 <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]5683
5684<h5>Example:</h5>
5685<pre>
Nick Lewycky5bb3ece2011-03-31 18:20:19 +0000[diff] [blame]5686 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5687 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5688</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5689
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5690</div>
5691
5692<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5693<h4>
Reid Spencer24d6da52007-01-21 00:29:26 +0000[diff] [blame]5694 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5695</h4>
5696
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5697<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5698
5699<h5>Syntax:</h5>
5700<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5701 &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]5702</pre>
5703
5704<h5>Overview:</h5>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5705<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5706 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5707
5708<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5709<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5710 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5711 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5712 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5713 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5714
5715<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5716<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5717 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5718 towards zero) unsigned integer value. If the value cannot fit
5719 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5720
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5721<h5>Example:</h5>
5722<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5723 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner88519042007-09-22 03:17:52 +0000[diff] [blame]5724 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5725 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5726</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5727
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5728</div>
5729
5730<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5731<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5732 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5733</h4>
5734
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5735<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5736
5737<h5>Syntax:</h5>
5738<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5739 &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]5740</pre>
5741
5742<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5743<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5744 <a href="#t_floating">floating point</a> <tt>value</tt> to
5745 type <tt>ty2</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5746
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5747<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5748<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5749 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5750 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5751 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5752 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5753
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5754<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5755<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5756 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5757 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5758 the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5759
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]5760<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5761<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5762 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner88519042007-09-22 03:17:52 +0000[diff] [blame]5763 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5764 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5765</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5766
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5767</div>
5768
5769<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5770<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5771 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5772</h4>
5773
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5774<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5775
5776<h5>Syntax:</h5>
5777<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5778 &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]5779</pre>
5780
5781<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5782<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5783 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5784
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5785<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +0000[diff] [blame]5786<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]5787 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5788 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5789 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5790 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5791
5792<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5793<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5794 integer quantity and converts it to the corresponding floating point
5795 value. If the value cannot fit in the floating point value, the results are
5796 undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5797
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5798<h5>Example:</h5>
5799<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5800 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5801 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5802</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5803
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5804</div>
5805
5806<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5807<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5808 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5809</h4>
5810
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5811<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5812
5813<h5>Syntax:</h5>
5814<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5815 &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]5816</pre>
5817
5818<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5819<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5820 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5821
5822<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +0000[diff] [blame]5823<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]5824 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5825 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5826 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5827 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5828
5829<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5830<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5831 quantity and converts it to the corresponding floating point value. If the
5832 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5833
5834<h5>Example:</h5>
5835<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5836 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5837 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5838</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5839
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5840</div>
5841
5842<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5843<h4>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5844 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5845</h4>
5846
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5847<div>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5848
5849<h5>Syntax:</h5>
5850<pre>
5851 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5852</pre>
5853
5854<h5>Overview:</h5>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5855<p>The '<tt>ptrtoint</tt>' instruction converts the pointer or a vector of
5856 pointers <tt>value</tt> to
5857 the integer (or vector of integers) type <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5858
5859<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5860<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5861 must be a a value of type <a href="#t_pointer">pointer</a> or a vector of
5862 pointers, and a type to cast it to
5863 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> or a vector
5864 of integers type.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5865
5866<h5>Semantics:</h5>
5867<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5868 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5869 truncating or zero extending that value to the size of the integer type. If
5870 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5871 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5872 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5873 change.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5874
5875<h5>Example:</h5>
5876<pre>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5877 %X = ptrtoint i32* %P to i8 <i>; yields truncation on 32-bit architecture</i>
5878 %Y = ptrtoint i32* %P to i64 <i>; yields zero extension on 32-bit architecture</i>
5879 %Z = ptrtoint &lt;4 x i32*&gt; %P to &lt;4 x i64&gt;<i>; yields vector zero extension for a vector of addresses on 32-bit architecture</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5880</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5881
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5882</div>
5883
5884<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5885<h4>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5886 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5887</h4>
5888
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5889<div>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5890
5891<h5>Syntax:</h5>
5892<pre>
5893 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5894</pre>
5895
5896<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5897<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5898 pointer type, <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5899
5900<h5>Arguments:</h5>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]5901<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5902 value to cast, and a type to cast it to, which must be a
5903 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5904
5905<h5>Semantics:</h5>
5906<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5907 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5908 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5909 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5910 than the size of a pointer then a zero extension is done. If they are the
5911 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5912
5913<h5>Example:</h5>
5914<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5915 %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]5916 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5917 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5918 %Z = inttoptr &lt;4 x i32&gt; %G to &lt;4 x i8*&gt;<i>; yields truncation of vector G to four pointers</i>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5919</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5920
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5921</div>
5922
5923<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5924<h4>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5925 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5926</h4>
5927
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5928<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5929
5930<h5>Syntax:</h5>
5931<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5932 &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]5933</pre>
5934
5935<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5936<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5937 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5938
5939<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5940<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5941 non-aggregate first class value, and a type to cast it to, which must also be
5942 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5943 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5944 identical. If the source type is a pointer, the destination type must also be
5945 a pointer. This instruction supports bitwise conversion of vectors to
5946 integers and to vectors of other types (as long as they have the same
5947 size).</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5948
5949<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5950<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5951 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5952 this conversion. The conversion is done as if the <tt>value</tt> had been
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5953 stored to memory and read back as type <tt>ty2</tt>.
5954 Pointer (or vector of pointers) types may only be converted to other pointer
5955 (or vector of pointers) types with this instruction. To convert
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5956 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5957 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5958
5959<h5>Example:</h5>
5960<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5961 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5962 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5963 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
5964 %Z = bitcast &lt;2 x i32*&gt; %V to &lt;2 x i64*&gt; <i>; yields &lt;2 x i64*&gt;</i>
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]5965</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5966
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5967</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5968
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5969</div>
5970
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5971<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5972<h3>
5973 <a name="otherops">Other Operations</a>
5974</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5975
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5976<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5977
5978<p>The instructions in this category are the "miscellaneous" instructions, which
5979 defy better classification.</p>
5980
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5981<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5982<h4>
5983 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5984</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5985
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5986<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5987
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5988<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5989<pre>
5990 &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]5991</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5992
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5993<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5994<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5995 boolean values based on comparison of its two integer, integer vector,
5996 pointer, or pointer vector operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5997
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5998<h5>Arguments:</h5>
5999<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6000 the condition code indicating the kind of comparison to perform. It is not a
6001 value, just a keyword. The possible condition code are:</p>
6002
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6003<ol>
6004 <li><tt>eq</tt>: equal</li>
6005 <li><tt>ne</tt>: not equal </li>
6006 <li><tt>ugt</tt>: unsigned greater than</li>
6007 <li><tt>uge</tt>: unsigned greater or equal</li>
6008 <li><tt>ult</tt>: unsigned less than</li>
6009 <li><tt>ule</tt>: unsigned less or equal</li>
6010 <li><tt>sgt</tt>: signed greater than</li>
6011 <li><tt>sge</tt>: signed greater or equal</li>
6012 <li><tt>slt</tt>: signed less than</li>
6013 <li><tt>sle</tt>: signed less or equal</li>
6014</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6015
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]6016<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6017 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
6018 typed. They must also be identical types.</p>
6019
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6020<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6021<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
6022 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]6023 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]6024 result, as follows:</p>
6025
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6026<ol>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6027 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6028 <tt>false</tt> otherwise. No sign interpretation is necessary or
6029 performed.</li>
6030
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6031 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6032 <tt>false</tt> otherwise. No sign interpretation is necessary or
6033 performed.</li>
6034
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6035 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6036 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
6037
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6038 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6039 <tt>true</tt> if <tt>op1</tt> is greater than or equal
6040 to <tt>op2</tt>.</li>
6041
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6042 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6043 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
6044
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6045 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6046 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
6047
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6048 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6049 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
6050
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6051 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6052 <tt>true</tt> if <tt>op1</tt> is greater than or equal
6053 to <tt>op2</tt>.</li>
6054
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6055 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6056 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
6057
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6058 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6059 <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]6060</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6061
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6062<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6063 values are compared as if they were integers.</p>
6064
6065<p>If the operands are integer vectors, then they are compared element by
6066 element. The result is an <tt>i1</tt> vector with the same number of elements
6067 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6068
6069<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6070<pre>
6071 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]6072 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
6073 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
6074 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
6075 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
6076 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6077</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]6078
6079<p>Note that the code generator does not yet support vector types with
6080 the <tt>icmp</tt> instruction.</p>
6081
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6082</div>
6083
6084<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6085<h4>
6086 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
6087</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6088
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6089<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6090
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6091<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6092<pre>
6093 &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]6094</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6095
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6096<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6097<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
6098 values based on comparison of its operands.</p>
6099
6100<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]6101(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6102
6103<p>If the operands are floating point vectors, then the result type is a vector
6104 of boolean with the same number of elements as the operands being
6105 compared.</p>
6106
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6107<h5>Arguments:</h5>
6108<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6109 the condition code indicating the kind of comparison to perform. It is not a
6110 value, just a keyword. The possible condition code are:</p>
6111
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6112<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]6113 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6114 <li><tt>oeq</tt>: ordered and equal</li>
6115 <li><tt>ogt</tt>: ordered and greater than </li>
6116 <li><tt>oge</tt>: ordered and greater than or equal</li>
6117 <li><tt>olt</tt>: ordered and less than </li>
6118 <li><tt>ole</tt>: ordered and less than or equal</li>
6119 <li><tt>one</tt>: ordered and not equal</li>
6120 <li><tt>ord</tt>: ordered (no nans)</li>
6121 <li><tt>ueq</tt>: unordered or equal</li>
6122 <li><tt>ugt</tt>: unordered or greater than </li>
6123 <li><tt>uge</tt>: unordered or greater than or equal</li>
6124 <li><tt>ult</tt>: unordered or less than </li>
6125 <li><tt>ule</tt>: unordered or less than or equal</li>
6126 <li><tt>une</tt>: unordered or not equal</li>
6127 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]6128 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6129</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6130
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6131<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6132 <i>unordered</i> means that either operand may be a QNAN.</p>
6133
6134<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
6135 a <a href="#t_floating">floating point</a> type or
6136 a <a href="#t_vector">vector</a> of floating point type. They must have
6137 identical types.</p>
6138
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6139<h5>Semantics:</h5>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]6140<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6141 according to the condition code given as <tt>cond</tt>. If the operands are
6142 vectors, then the vectors are compared element by element. Each comparison
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]6143 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6144 follows:</p>
6145
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6146<ol>
6147 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6148
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6149 <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]6150 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
6151
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]6152 <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]6153 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6154
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6155 <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]6156 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
6157
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6158 <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]6159 <tt>op1</tt> is less than <tt>op2</tt>.</li>
6160
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6161 <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]6162 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
6163
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6164 <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]6165 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
6166
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]6167 <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]6168
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6169 <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]6170 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
6171
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6172 <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]6173 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
6174
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6175 <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]6176 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
6177
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6178 <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]6179 <tt>op1</tt> is less than <tt>op2</tt>.</li>
6180
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6181 <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]6182 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
6183
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6184 <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]6185 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
6186
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]6187 <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]6188
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6189 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
6190</ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6191
6192<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6193<pre>
6194 &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]6195 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
6196 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
6197 &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]6198</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]6199
6200<p>Note that the code generator does not yet support vector types with
6201 the <tt>fcmp</tt> instruction.</p>
6202
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]6203</div>
6204
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6205<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6206<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6207 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6208</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6209
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6210<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6211
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6212<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6213<pre>
6214 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
6215</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6216
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6217<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6218<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
6219 SSA graph representing the function.</p>
6220
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6221<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6222<p>The type of the incoming values is specified with the first type field. After
6223 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
6224 one pair for each predecessor basic block of the current block. Only values
6225 of <a href="#t_firstclass">first class</a> type may be used as the value
6226 arguments to the PHI node. Only labels may be used as the label
6227 arguments.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6228
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6229<p>There must be no non-phi instructions between the start of a basic block and
6230 the PHI instructions: i.e. PHI instructions must be first in a basic
6231 block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6232
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6233<p>For the purposes of the SSA form, the use of each incoming value is deemed to
6234 occur on the edge from the corresponding predecessor block to the current
6235 block (but after any definition of an '<tt>invoke</tt>' instruction's return
6236 value on the same edge).</p>
Jay Foadd2449092009-06-03 10:20:10 +0000[diff] [blame]6237
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6238<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6239<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6240 specified by the pair corresponding to the predecessor basic block that
6241 executed just prior to the current block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6242
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6243<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]6244<pre>
6245Loop: ; Infinite loop that counts from 0 on up...
6246 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
6247 %nextindvar = add i32 %indvar, 1
6248 br label %Loop
6249</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6250
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]6251</div>
6252
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6253<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6254<h4>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6255 <a name="i_select">'<tt>select</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6256</h4>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6257
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6258<div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6259
6260<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6261<pre>
Dan Gohmanf72fb672008-09-09 01:02:47 +0000[diff] [blame]6262 &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>
6263
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6264 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6265</pre>
6266
6267<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6268<p>The '<tt>select</tt>' instruction is used to choose one value based on a
6269 condition, without branching.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6270
6271
6272<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6273<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
6274 values indicating the condition, and two values of the
6275 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
6276 vectors and the condition is a scalar, then entire vectors are selected, not
6277 individual elements.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6278
6279<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6280<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
6281 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6282
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6283<p>If the condition is a vector of i1, then the value arguments must be vectors
6284 of the same size, and the selection is done element by element.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6285
6286<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6287<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]6288 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6289</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]6290
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6291</div>
6292
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]6293<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6294<h4>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6295 <a name="i_call">'<tt>call</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6296</h4>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6297
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6298<div>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6299
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6300<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6301<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]6302 &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]6303</pre>
6304
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6305<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6306<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6307
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6308<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6309<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6310
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]6311<ol>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6312 <li>The optional "tail" marker indicates that the callee function does not
6313 access any allocas or varargs in the caller. Note that calls may be
6314 marked "tail" even if they do not occur before
6315 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
6316 present, the function call is eligible for tail call optimization,
6317 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Chengdc444e92010-03-08 21:05:02 +0000[diff] [blame]6318 optimized into a jump</a>. The code generator may optimize calls marked
6319 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
6320 sibling call optimization</a> when the caller and callee have
6321 matching signatures, or 2) forced tail call optimization when the
6322 following extra requirements are met:
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6323 <ul>
6324 <li>Caller and callee both have the calling
6325 convention <tt>fastcc</tt>.</li>
6326 <li>The call is in tail position (ret immediately follows call and ret
6327 uses value of call or is void).</li>
6328 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohmanfbbee8d2010-03-02 01:08:11 +0000[diff] [blame]6329 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6330 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
6331 constraints are met.</a></li>
6332 </ul>
6333 </li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]6334
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6335 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
6336 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6337 defaults to using C calling conventions. The calling convention of the
6338 call must match the calling convention of the target function, or else the
6339 behavior is undefined.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]6340
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6341 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
6342 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
6343 '<tt>inreg</tt>' attributes are valid here.</li>
6344
6345 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
6346 type of the return value. Functions that return no value are marked
6347 <tt><a href="#t_void">void</a></tt>.</li>
6348
6349 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
6350 being invoked. The argument types must match the types implied by this
6351 signature. This type can be omitted if the function is not varargs and if
6352 the function type does not return a pointer to a function.</li>
6353
6354 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
6355 be invoked. In most cases, this is a direct function invocation, but
6356 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
6357 to function value.</li>
6358
6359 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]6360 signature argument types and parameter attributes. All arguments must be
6361 of <a href="#t_firstclass">first class</a> type. If the function
6362 signature indicates the function accepts a variable number of arguments,
6363 the extra arguments can be specified.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6364
6365 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
6366 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
6367 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]6368</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6369
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6370<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6371<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
6372 a specified function, with its incoming arguments bound to the specified
6373 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
6374 function, control flow continues with the instruction after the function
6375 call, and the return value of the function is bound to the result
6376 argument.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6377
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6378<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6379<pre>
Nick Lewyckydb7e3c92007-09-08 13:57:50 +0000[diff] [blame]6380 %retval = call i32 @test(i32 %argc)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6381 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattner772fccf2008-03-21 17:24:17 +0000[diff] [blame]6382 %X = tail call i32 @foo() <i>; yields i32</i>
6383 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
6384 call void %foo(i8 97 signext)
Devang Patelc3fc6df2008-03-10 20:49:15 +0000[diff] [blame]6385
6386 %struct.A = type { i32, i8 }
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]6387 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]6388 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
6389 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner85a350f2008-10-08 06:26:11 +0000[diff] [blame]6390 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmancb73d192008-10-07 10:03:45 +0000[diff] [blame]6391 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6392</pre>
6393
Dale Johannesen07de8d12009-09-24 18:38:21 +0000[diff] [blame]6394<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen9f8380b2009-09-25 17:04:42 +0000[diff] [blame]6395standard C99 library as being the C99 library functions, and may perform
6396optimizations or generate code for them under that assumption. This is
6397something we'd like to change in the future to provide better support for
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6398freestanding environments and non-C-based languages.</p>
Dale Johannesen07de8d12009-09-24 18:38:21 +0000[diff] [blame]6399
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6400</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6401
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6402<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6403<h4>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]6404 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6405</h4>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6406
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6407<div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6408
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6409<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6410<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6411 &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]6412</pre>
6413
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6414<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6415<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6416 the "variable argument" area of a function call. It is used to implement the
6417 <tt>va_arg</tt> macro in C.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6418
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6419<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6420<p>This instruction takes a <tt>va_list*</tt> value and the type of the
6421 argument. It returns a value of the specified argument type and increments
6422 the <tt>va_list</tt> to point to the next argument. The actual type
6423 of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6424
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6425<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6426<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
6427 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
6428 to the next argument. For more information, see the variable argument
6429 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6430
6431<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]6432 take a variable number of arguments, for example, the <tt>vfprintf</tt>
6433 function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6434
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6435<p><tt>va_arg</tt> is an LLVM instruction instead of
6436 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
6437 argument.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6438
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6439<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6440<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6441
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6442<p>Note that the code generator does not yet fully support va_arg on many
6443 targets. Also, it does not currently support va_arg with aggregate types on
6444 any target.</p>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +0000[diff] [blame]6445
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6446</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6447
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6448<!-- _______________________________________________________________________ -->
6449<h4>
6450 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6451</h4>
6452
6453<div>
6454
6455<h5>Syntax:</h5>
6456<pre>
Duncan Sands8d6796b2012-01-13 19:59:16 +0000[diff] [blame]6457 &lt;resultval&gt; = landingpad &lt;resultty&gt; personality &lt;type&gt; &lt;pers_fn&gt; &lt;clause&gt;+
6458 &lt;resultval&gt; = landingpad &lt;resultty&gt; personality &lt;type&gt; &lt;pers_fn&gt; cleanup &lt;clause&gt;*
Bill Wendlingbf13ee12011-08-08 08:06:05 +0000[diff] [blame]6459
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6460 &lt;clause&gt; := catch &lt;type&gt; &lt;value&gt;
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6461 &lt;clause&gt; := filter &lt;array constant type&gt; &lt;array constant&gt;
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6462</pre>
6463
6464<h5>Overview:</h5>
6465<p>The '<tt>landingpad</tt>' instruction is used by
6466 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6467 system</a> to specify that a basic block is a landing pad &mdash; one where
6468 the exception lands, and corresponds to the code found in the
6469 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6470 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6471 re-entry to the function. The <tt>resultval</tt> has the
Duncan Sands8d6796b2012-01-13 19:59:16 +0000[diff] [blame]6472 type <tt>resultty</tt>.</p>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6473
6474<h5>Arguments:</h5>
6475<p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6476 function associated with the unwinding mechanism. The optional
6477 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6478
6479<p>A <tt>clause</tt> begins with the clause type &mdash; <tt>catch</tt>
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6480 or <tt>filter</tt> &mdash; and contains the global variable representing the
6481 "type" that may be caught or filtered respectively. Unlike the
6482 <tt>catch</tt> clause, the <tt>filter</tt> clause takes an array constant as
6483 its argument. Use "<tt>[0 x i8**] undef</tt>" for a filter which cannot
6484 throw. The '<tt>landingpad</tt>' instruction must contain <em>at least</em>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6485 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6486
6487<h5>Semantics:</h5>
6488<p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6489 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6490 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6491 calling conventions, how the personality function results are represented in
6492 LLVM IR is target specific.</p>
6493
Bill Wendlingb7a01352011-08-03 17:17:06 +0000[diff] [blame]6494<p>The clauses are applied in order from top to bottom. If two
6495 <tt>landingpad</tt> instructions are merged together through inlining, the
Duncan Sands8d6796b2012-01-13 19:59:16 +0000[diff] [blame]6496 clauses from the calling function are appended to the list of clauses.
6497 When the call stack is being unwound due to an exception being thrown, the
6498 exception is compared against each <tt>clause</tt> in turn. If it doesn't
6499 match any of the clauses, and the <tt>cleanup</tt> flag is not set, then
6500 unwinding continues further up the call stack.</p>
Bill Wendlingb7a01352011-08-03 17:17:06 +0000[diff] [blame]6501
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6502<p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6503
6504<ul>
6505 <li>A landing pad block is a basic block which is the unwind destination of an
6506 '<tt>invoke</tt>' instruction.</li>
6507 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6508 first non-PHI instruction.</li>
6509 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6510 pad block.</li>
6511 <li>A basic block that is not a landing pad block may not include a
6512 '<tt>landingpad</tt>' instruction.</li>
6513 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6514 personality function.</li>
6515</ul>
6516
6517<h5>Example:</h5>
6518<pre>
6519 ;; A landing pad which can catch an integer.
6520 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6521 catch i8** @_ZTIi
6522 ;; A landing pad that is a cleanup.
6523 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6524 cleanup
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6525 ;; A landing pad which can catch an integer and can only throw a double.
6526 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6527 catch i8** @_ZTIi
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6528 filter [1 x i8**] [@_ZTId]
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6529</pre>
6530
6531</div>
6532
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6533</div>
6534
6535</div>
6536
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6537<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6538<h2><a name="intrinsics">Intrinsic Functions</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]6539<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6540
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6541<div>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6542
6543<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6544 well known names and semantics and are required to follow certain
6545 restrictions. Overall, these intrinsics represent an extension mechanism for
6546 the LLVM language that does not require changing all of the transformations
6547 in LLVM when adding to the language (or the bitcode reader/writer, the
6548 parser, etc...).</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6549
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]6550<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6551 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6552 begin with this prefix. Intrinsic functions must always be external
6553 functions: you cannot define the body of intrinsic functions. Intrinsic
6554 functions may only be used in call or invoke instructions: it is illegal to
6555 take the address of an intrinsic function. Additionally, because intrinsic
6556 functions are part of the LLVM language, it is required if any are added that
6557 they be documented here.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6558
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6559<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6560 family of functions that perform the same operation but on different data
6561 types. Because LLVM can represent over 8 million different integer types,
6562 overloading is used commonly to allow an intrinsic function to operate on any
6563 integer type. One or more of the argument types or the result type can be
6564 overloaded to accept any integer type. Argument types may also be defined as
6565 exactly matching a previous argument's type or the result type. This allows
6566 an intrinsic function which accepts multiple arguments, but needs all of them
6567 to be of the same type, to only be overloaded with respect to a single
6568 argument or the result.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6569
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6570<p>Overloaded intrinsics will have the names of its overloaded argument types
6571 encoded into its function name, each preceded by a period. Only those types
6572 which are overloaded result in a name suffix. Arguments whose type is matched
6573 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6574 can take an integer of any width and returns an integer of exactly the same
6575 integer width. This leads to a family of functions such as
6576 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6577 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6578 suffix is required. Because the argument's type is matched against the return
6579 type, it does not require its own name suffix.</p>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]6580
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6581<p>To learn how to add an intrinsic function, please see the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6582 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6583
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6584<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6585<h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6586 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6587</h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6588
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6589<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6590
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6591<p>Variable argument support is defined in LLVM with
6592 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6593 intrinsic functions. These functions are related to the similarly named
6594 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6595
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6596<p>All of these functions operate on arguments that use a target-specific value
6597 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6598 not define what this type is, so all transformations should be prepared to
6599 handle these functions regardless of the type used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6600
Chris Lattner374ab302006-05-15 17:26:46 +0000[diff] [blame]6601<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]6602 instruction and the variable argument handling intrinsic functions are
6603 used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6604
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]6605<pre class="doc_code">
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6606define i32 @test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6607 ; Initialize variable argument processing
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6608 %ap = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6609 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6610 call void @llvm.va_start(i8* %ap2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6611
6612 ; Read a single integer argument
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6613 %tmp = va_arg i8** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6614
6615 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6616 %aq = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6617 %aq2 = bitcast i8** %aq to i8*
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6618 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6619 call void @llvm.va_end(i8* %aq2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6620
6621 ; Stop processing of arguments.
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6622 call void @llvm.va_end(i8* %ap2)
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]6623 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6624}
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6625
6626declare void @llvm.va_start(i8*)
6627declare void @llvm.va_copy(i8*, i8*)
6628declare void @llvm.va_end(i8*)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6629</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6630
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6631<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6632<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6633 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6634</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6635
6636
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6637<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6638
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6639<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6640<pre>
6641 declare void %llvm.va_start(i8* &lt;arglist&gt;)
6642</pre>
6643
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6644<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6645<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
6646 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]6647
6648<h5>Arguments:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6649<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]6650
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6651<h5>Semantics:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6652<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]6653 macro available in C. In a target-dependent way, it initializes
6654 the <tt>va_list</tt> element to which the argument points, so that the next
6655 call to <tt>va_arg</tt> will produce the first variable argument passed to
6656 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6657 need to know the last argument of the function as the compiler can figure
6658 that out.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6659
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6660</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6661
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6662<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6663<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6664 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6665</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6666
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6667<div>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6668
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6669<h5>Syntax:</h5>
6670<pre>
6671 declare void @llvm.va_end(i8* &lt;arglist&gt;)
6672</pre>
6673
6674<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6675<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6676 which has been initialized previously
6677 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6678 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6679
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6680<h5>Arguments:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6681<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]6682
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6683<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6684<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]6685 macro available in C. In a target-dependent way, it destroys
6686 the <tt>va_list</tt> element to which the argument points. Calls
6687 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6688 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6689 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6690
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6691</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6692
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6693<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6694<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6695 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6696</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6697
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6698<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6699
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6700<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6701<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]6702 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6703</pre>
6704
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6705<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6706<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6707 from the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6708
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6709<h5>Arguments:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6710<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]6711 The second argument is a pointer to a <tt>va_list</tt> element to copy
6712 from.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6713
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6714<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6715<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]6716 macro available in C. In a target-dependent way, it copies the
6717 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6718 element. This intrinsic is necessary because
6719 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6720 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6721
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6722</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6723
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6724</div>
6725
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6726<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6727<h3>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6728 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6729</h3>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6730
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6731<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6732
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6733<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]6734Collection</a> (GC) requires the implementation and generation of these
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6735intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6736roots on the stack</a>, as well as garbage collector implementations that
6737require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6738barriers. Front-ends for type-safe garbage collected languages should generate
6739these intrinsics to make use of the LLVM garbage collectors. For more details,
6740see <a href="GarbageCollection.html">Accurate Garbage Collection with
6741LLVM</a>.</p>
Christopher Lamb303dae92007-12-17 01:00:21 +0000[diff] [blame]6742
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6743<p>The garbage collection intrinsics only operate on objects in the generic
6744 address space (address space zero).</p>
Christopher Lamb303dae92007-12-17 01:00:21 +0000[diff] [blame]6745
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6746<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6747<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6748 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6749</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6750
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6751<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6752
6753<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6754<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6755 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6756</pre>
6757
6758<h5>Overview:</h5>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]6759<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]6760 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6761
6762<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6763<p>The first argument specifies the address of a stack object that contains the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6764 root pointer. The second pointer (which must be either a constant or a
6765 global value address) contains the meta-data to be associated with the
6766 root.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6767
6768<h5>Semantics:</h5>
Chris Lattner05d67092008-04-24 05:59:56 +0000[diff] [blame]6769<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]6770 location. At compile-time, the code generator generates information to allow
6771 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6772 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6773 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6774
6775</div>
6776
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6777<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6778<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6779 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6780</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6781
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6782<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6783
6784<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6785<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6786 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6787</pre>
6788
6789<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6790<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6791 locations, allowing garbage collector implementations that require read
6792 barriers.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6793
6794<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]6795<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]6796 allocated from the garbage collector. The first object is a pointer to the
6797 start of the referenced object, if needed by the language runtime (otherwise
6798 null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6799
6800<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6801<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6802 instruction, but may be replaced with substantially more complex code by the
6803 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6804 may only be used in a function which <a href="#gc">specifies a GC
6805 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6806
6807</div>
6808
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6809<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6810<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6811 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6812</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6813
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6814<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6815
6816<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6817<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6818 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6819</pre>
6820
6821<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6822<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6823 locations, allowing garbage collector implementations that require write
6824 barriers (such as generational or reference counting collectors).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6825
6826<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]6827<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]6828 object to store it to, and the third is the address of the field of Obj to
6829 store to. If the runtime does not require a pointer to the object, Obj may
6830 be null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6831
6832<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6833<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6834 instruction, but may be replaced with substantially more complex code by the
6835 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6836 may only be used in a function which <a href="#gc">specifies a GC
6837 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6838
6839</div>
6840
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6841</div>
6842
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6843<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6844<h3>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6845 <a name="int_codegen">Code Generator Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6846</h3>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6847
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6848<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6849
6850<p>These intrinsics are provided by LLVM to expose special features that may
6851 only be implemented with code generator support.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6852
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6853<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6854<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6855 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6856</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6857
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6858<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6859
6860<h5>Syntax:</h5>
6861<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]6862 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6863</pre>
6864
6865<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6866<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6867 target-specific value indicating the return address of the current function
6868 or one of its callers.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6869
6870<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6871<p>The argument to this intrinsic indicates which function to return the address
6872 for. Zero indicates the calling function, one indicates its caller, etc.
6873 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6874
6875<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6876<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6877 indicating the return address of the specified call frame, or zero if it
6878 cannot be identified. The value returned by this intrinsic is likely to be
6879 incorrect or 0 for arguments other than zero, so it should only be used for
6880 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6881
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6882<p>Note that calling this intrinsic does not prevent function inlining or other
6883 aggressive transformations, so the value returned may not be that of the
6884 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6885
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6886</div>
6887
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6888<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6889<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6890 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6891</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6892
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6893<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6894
6895<h5>Syntax:</h5>
6896<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6897 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6898</pre>
6899
6900<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6901<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6902 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6903
6904<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6905<p>The argument to this intrinsic indicates which function to return the frame
6906 pointer for. Zero indicates the calling function, one indicates its caller,
6907 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6908
6909<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6910<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6911 indicating the frame address of the specified call frame, or zero if it
6912 cannot be identified. The value returned by this intrinsic is likely to be
6913 incorrect or 0 for arguments other than zero, so it should only be used for
6914 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6915
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6916<p>Note that calling this intrinsic does not prevent function inlining or other
6917 aggressive transformations, so the value returned may not be that of the
6918 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6919
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6920</div>
6921
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6922<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6923<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6924 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6925</h4>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6926
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6927<div>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6928
6929<h5>Syntax:</h5>
6930<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6931 declare i8* @llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6932</pre>
6933
6934<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6935<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6936 of the function stack, for use
6937 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6938 useful for implementing language features like scoped automatic variable
6939 sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6940
6941<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6942<p>This intrinsic returns a opaque pointer value that can be passed
6943 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6944 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6945 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6946 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6947 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6948 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6949
6950</div>
6951
6952<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6953<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6954 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6955</h4>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6956
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6957<div>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6958
6959<h5>Syntax:</h5>
6960<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6961 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6962</pre>
6963
6964<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6965<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6966 the function stack to the state it was in when the
6967 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6968 executed. This is useful for implementing language features like scoped
6969 automatic variable sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6970
6971<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6972<p>See the description
6973 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6974
6975</div>
6976
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6977<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6978<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6979 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6980</h4>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6981
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6982<div>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6983
6984<h5>Syntax:</h5>
6985<pre>
Bruno Cardoso Lopes9a767332011-06-14 04:58:37 +0000[diff] [blame]6986 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]6987</pre>
6988
6989<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6990<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6991 insert a prefetch instruction if supported; otherwise, it is a noop.
6992 Prefetches have no effect on the behavior of the program but can change its
6993 performance characteristics.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6994
6995<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6996<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6997 specifier determining if the fetch should be for a read (0) or write (1),
6998 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Bruno Cardoso Lopes9a767332011-06-14 04:58:37 +0000[diff] [blame]6999 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
7000 specifies whether the prefetch is performed on the data (1) or instruction (0)
7001 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
7002 must be constant integers.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]7003
7004<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7005<p>This intrinsic does not modify the behavior of the program. In particular,
7006 prefetches cannot trap and do not produce a value. On targets that support
7007 this intrinsic, the prefetch can provide hints to the processor cache for
7008 better performance.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]7009
7010</div>
7011
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7012<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7013<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7014 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7015</h4>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7016
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7017<div>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7018
7019<h5>Syntax:</h5>
7020<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]7021 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7022</pre>
7023
7024<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7025<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
7026 Counter (PC) in a region of code to simulators and other tools. The method
7027 is target specific, but it is expected that the marker will use exported
7028 symbols to transmit the PC of the marker. The marker makes no guarantees
7029 that it will remain with any specific instruction after optimizations. It is
7030 possible that the presence of a marker will inhibit optimizations. The
7031 intended use is to be inserted after optimizations to allow correlations of
7032 simulation runs.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7033
7034<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7035<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7036
7037<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7038<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]7039 not support this intrinsic may ignore it.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]7040
7041</div>
7042
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]7043<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7044<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7045 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7046</h4>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]7047
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7048<div>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]7049
7050<h5>Syntax:</h5>
7051<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7052 declare i64 @llvm.readcyclecounter()
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]7053</pre>
7054
7055<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7056<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
7057 counter register (or similar low latency, high accuracy clocks) on those
7058 targets that support it. On X86, it should map to RDTSC. On Alpha, it
7059 should map to RPCC. As the backing counters overflow quickly (on the order
7060 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]7061
7062<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7063<p>When directly supported, reading the cycle counter should not modify any
7064 memory. Implementations are allowed to either return a application specific
7065 value or a system wide value. On backends without support, this is lowered
7066 to a constant 0.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]7067
7068</div>
7069
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7070</div>
7071
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7072<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7073<h3>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7074 <a name="int_libc">Standard C Library Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7075</h3>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7076
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7077<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7078
7079<p>LLVM provides intrinsics for a few important standard C library functions.
7080 These intrinsics allow source-language front-ends to pass information about
7081 the alignment of the pointer arguments to the code generator, providing
7082 opportunity for more efficient code generation.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7083
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7084<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7085<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7086 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7087</h4>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7088
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7089<div>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7090
7091<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7092<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]7093 integer bit width and for different address spaces. Not all targets support
7094 all bit widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7095
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7096<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]7097 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]7098 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]7099 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]7100 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7101</pre>
7102
7103<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7104<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
7105 source location to the destination location.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7106
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7107<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7108 intrinsics do not return a value, takes extra alignment/isvolatile arguments
7109 and the pointers can be in specified address spaces.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7110
7111<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7112
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7113<p>The first argument is a pointer to the destination, the second is a pointer
7114 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7115 number of bytes to copy, the fourth argument is the alignment of the
7116 source and destination locations, and the fifth is a boolean indicating a
7117 volatile access.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7118
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]7119<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]7120 then the caller guarantees that both the source and destination pointers are
7121 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]7122
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]7123<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7124 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
7125 The detailed access behavior is not very cleanly specified and it is unwise
7126 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7127
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7128<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7129
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7130<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
7131 source location to the destination location, which are not allowed to
7132 overlap. It copies "len" bytes of memory over. If the argument is known to
7133 be aligned to some boundary, this can be specified as the fourth argument,
7134 otherwise it should be set to 0 or 1.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7135
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]7136</div>
7137
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7138<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7139<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7140 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7141</h4>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7142
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7143<div>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7144
7145<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +0000[diff] [blame]7146<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]7147 width and for different address space. Not all targets support all bit
7148 widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7149
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7150<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]7151 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]7152 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]7153 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]7154 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7155</pre>
7156
7157<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7158<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
7159 source location to the destination location. It is similar to the
7160 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
7161 overlap.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7162
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7163<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7164 intrinsics do not return a value, takes extra alignment/isvolatile arguments
7165 and the pointers can be in specified address spaces.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7166
7167<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7168
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7169<p>The first argument is a pointer to the destination, the second is a pointer
7170 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7171 number of bytes to copy, the fourth argument is the alignment of the
7172 source and destination locations, and the fifth is a boolean indicating a
7173 volatile access.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7174
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]7175<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]7176 then the caller guarantees that the source and destination pointers are
7177 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]7178
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]7179<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7180 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
7181 The detailed access behavior is not very cleanly specified and it is unwise
7182 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7183
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7184<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7185
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7186<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
7187 source location to the destination location, which may overlap. It copies
7188 "len" bytes of memory over. If the argument is known to be aligned to some
7189 boundary, this can be specified as the fourth argument, otherwise it should
7190 be set to 0 or 1.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7191
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]7192</div>
7193
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7194<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7195<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7196 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7197</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7198
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7199<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7200
7201<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +0000[diff] [blame]7202<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]7203 width and for different address spaces. However, not all targets support all
7204 bit widths.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7205
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7206<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]7207 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +0000[diff] [blame]7208 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]7209 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +0000[diff] [blame]7210 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7211</pre>
7212
7213<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7214<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
7215 particular byte value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7216
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7217<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]7218 intrinsic does not return a value and takes extra alignment/volatile
7219 arguments. Also, the destination can be in an arbitrary address space.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7220
7221<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7222<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]7223 byte value with which to fill it, the third argument is an integer argument
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7224 specifying the number of bytes to fill, and the fourth argument is the known
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]7225 alignment of the destination location.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7226
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]7227<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]7228 then the caller guarantees that the destination pointer is aligned to that
7229 boundary.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7230
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]7231<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
7232 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
7233 The detailed access behavior is not very cleanly specified and it is unwise
7234 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]7235
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7236<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7237<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
7238 at the destination location. If the argument is known to be aligned to some
7239 boundary, this can be specified as the fourth argument, otherwise it should
7240 be set to 0 or 1.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7241
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]7242</div>
7243
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]7244<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7245<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7246 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7247</h4>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7248
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7249<div>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7250
7251<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7252<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
7253 floating point or vector of floating point type. Not all targets support all
7254 types however.</p>
7255
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7256<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +0000[diff] [blame]7257 declare float @llvm.sqrt.f32(float %Val)
7258 declare double @llvm.sqrt.f64(double %Val)
7259 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
7260 declare fp128 @llvm.sqrt.f128(fp128 %Val)
7261 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7262</pre>
7263
7264<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7265<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
7266 returning the same value as the libm '<tt>sqrt</tt>' functions would.
7267 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
7268 behavior for negative numbers other than -0.0 (which allows for better
7269 optimization, because there is no need to worry about errno being
7270 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]7271
7272<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7273<p>The argument and return value are floating point numbers of the same
7274 type.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7275
7276<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7277<p>This function returns the sqrt of the specified operand if it is a
7278 nonnegative floating point number.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7279
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7280</div>
7281
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7282<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7283<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7284 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7285</h4>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7286
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7287<div>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7288
7289<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7290<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
7291 floating point or vector of floating point type. Not all targets support all
7292 types however.</p>
7293
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7294<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +0000[diff] [blame]7295 declare float @llvm.powi.f32(float %Val, i32 %power)
7296 declare double @llvm.powi.f64(double %Val, i32 %power)
7297 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
7298 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
7299 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7300</pre>
7301
7302<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7303<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
7304 specified (positive or negative) power. The order of evaluation of
7305 multiplications is not defined. When a vector of floating point type is
7306 used, the second argument remains a scalar integer value.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7307
7308<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7309<p>The second argument is an integer power, and the first is a value to raise to
7310 that power.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7311
7312<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7313<p>This function returns the first value raised to the second power with an
7314 unspecified sequence of rounding operations.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7315
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7316</div>
7317
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7318<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7319<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7320 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7321</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7322
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7323<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7324
7325<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7326<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
7327 floating point or vector of floating point type. Not all targets support all
7328 types however.</p>
7329
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7330<pre>
7331 declare float @llvm.sin.f32(float %Val)
7332 declare double @llvm.sin.f64(double %Val)
7333 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7334 declare fp128 @llvm.sin.f128(fp128 %Val)
7335 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7336</pre>
7337
7338<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7339<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7340
7341<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7342<p>The argument and return value are floating point numbers of the same
7343 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7344
7345<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7346<p>This function returns the sine of the specified operand, returning the same
7347 values as the libm <tt>sin</tt> functions would, and handles error conditions
7348 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7349
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7350</div>
7351
7352<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7353<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7354 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7355</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7356
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7357<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7358
7359<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7360<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
7361 floating point or vector of floating point type. Not all targets support all
7362 types however.</p>
7363
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7364<pre>
7365 declare float @llvm.cos.f32(float %Val)
7366 declare double @llvm.cos.f64(double %Val)
7367 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7368 declare fp128 @llvm.cos.f128(fp128 %Val)
7369 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7370</pre>
7371
7372<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7373<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7374
7375<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7376<p>The argument and return value are floating point numbers of the same
7377 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7378
7379<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7380<p>This function returns the cosine of the specified operand, returning the same
7381 values as the libm <tt>cos</tt> functions would, and handles error conditions
7382 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7383
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7384</div>
7385
7386<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7387<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7388 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7389</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7390
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7391<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7392
7393<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7394<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
7395 floating point or vector of floating point type. Not all targets support all
7396 types however.</p>
7397
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7398<pre>
7399 declare float @llvm.pow.f32(float %Val, float %Power)
7400 declare double @llvm.pow.f64(double %Val, double %Power)
7401 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7402 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7403 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7404</pre>
7405
7406<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7407<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
7408 specified (positive or negative) power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7409
7410<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7411<p>The second argument is a floating point power, and the first is a value to
7412 raise to that power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7413
7414<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7415<p>This function returns the first value raised to the second power, returning
7416 the same values as the libm <tt>pow</tt> functions would, and handles error
7417 conditions in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7418
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7419</div>
7420
Dan Gohman4e9011c2011-05-23 21:13:03 +0000[diff] [blame]7421<!-- _______________________________________________________________________ -->
7422<h4>
7423 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
7424</h4>
7425
7426<div>
7427
7428<h5>Syntax:</h5>
7429<p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
7430 floating point or vector of floating point type. Not all targets support all
7431 types however.</p>
7432
7433<pre>
7434 declare float @llvm.exp.f32(float %Val)
7435 declare double @llvm.exp.f64(double %Val)
7436 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7437 declare fp128 @llvm.exp.f128(fp128 %Val)
7438 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7439</pre>
7440
7441<h5>Overview:</h5>
7442<p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7443
7444<h5>Arguments:</h5>
7445<p>The argument and return value are floating point numbers of the same
7446 type.</p>
7447
7448<h5>Semantics:</h5>
7449<p>This function returns the same values as the libm <tt>exp</tt> functions
7450 would, and handles error conditions in the same way.</p>
7451
7452</div>
7453
7454<!-- _______________________________________________________________________ -->
7455<h4>
7456 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7457</h4>
7458
7459<div>
7460
7461<h5>Syntax:</h5>
7462<p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7463 floating point or vector of floating point type. Not all targets support all
7464 types however.</p>
7465
7466<pre>
7467 declare float @llvm.log.f32(float %Val)
7468 declare double @llvm.log.f64(double %Val)
7469 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7470 declare fp128 @llvm.log.f128(fp128 %Val)
7471 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7472</pre>
7473
7474<h5>Overview:</h5>
7475<p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7476
7477<h5>Arguments:</h5>
7478<p>The argument and return value are floating point numbers of the same
7479 type.</p>
7480
7481<h5>Semantics:</h5>
7482<p>This function returns the same values as the libm <tt>log</tt> functions
7483 would, and handles error conditions in the same way.</p>
7484
Nick Lewycky1c929be2011-10-31 01:32:21 +0000[diff] [blame]7485</div>
7486
7487<!-- _______________________________________________________________________ -->
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]7488<h4>
7489 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7490</h4>
7491
7492<div>
7493
7494<h5>Syntax:</h5>
7495<p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7496 floating point or vector of floating point type. Not all targets support all
7497 types however.</p>
7498
7499<pre>
7500 declare float @llvm.fma.f32(float %a, float %b, float %c)
7501 declare double @llvm.fma.f64(double %a, double %b, double %c)
7502 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7503 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7504 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7505</pre>
7506
7507<h5>Overview:</h5>
Cameron Zwarichabc43e62011-07-08 22:13:55 +0000[diff] [blame]7508<p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]7509 operation.</p>
7510
7511<h5>Arguments:</h5>
7512<p>The argument and return value are floating point numbers of the same
7513 type.</p>
7514
7515<h5>Semantics:</h5>
7516<p>This function returns the same values as the libm <tt>fma</tt> functions
7517 would.</p>
7518
Dan Gohman4e9011c2011-05-23 21:13:03 +0000[diff] [blame]7519</div>
7520
Peter Collingbourne168a4c32012-07-03 12:25:40 +0000[diff] [blame]7521<!-- _______________________________________________________________________ -->
7522<h4>
7523 <a name="int_fabs">'<tt>llvm.fabs.*</tt>' Intrinsic</a>
7524</h4>
7525
7526<div>
7527
7528<h5>Syntax:</h5>
7529<p>This is an overloaded intrinsic. You can use <tt>llvm.fabs</tt> on any
7530 floating point or vector of floating point type. Not all targets support all
7531 types however.</p>
7532
7533<pre>
7534 declare float @llvm.fabs.f32(float %Val)
7535 declare double @llvm.fabs.f64(double %Val)
7536 declare x86_fp80 @llvm.fabs.f80(x86_fp80 %Val)
7537 declare fp128 @llvm.fabs.f128(fp128 %Val)
7538 declare ppc_fp128 @llvm.fabs.ppcf128(ppc_fp128 %Val)
7539</pre>
7540
7541<h5>Overview:</h5>
7542<p>The '<tt>llvm.fabs.*</tt>' intrinsics return the absolute value of
7543 the operand.</p>
7544
7545<h5>Arguments:</h5>
7546<p>The argument and return value are floating point numbers of the same
7547 type.</p>
7548
7549<h5>Semantics:</h5>
7550<p>This function returns the same values as the libm <tt>fabs</tt> functions
7551 would, and handles error conditions in the same way.</p>
7552
7553</div>
7554
Dan Gohman27db99f2012-07-26 17:43:27 +0000[diff] [blame]7555<!-- _______________________________________________________________________ -->
7556<h4>
7557 <a name="int_floor">'<tt>llvm.floor.*</tt>' Intrinsic</a>
7558</h4>
7559
7560<div>
7561
7562<h5>Syntax:</h5>
7563<p>This is an overloaded intrinsic. You can use <tt>llvm.floor</tt> on any
7564 floating point or vector of floating point type. Not all targets support all
7565 types however.</p>
7566
7567<pre>
7568 declare float @llvm.floor.f32(float %Val)
7569 declare double @llvm.floor.f64(double %Val)
7570 declare x86_fp80 @llvm.floor.f80(x86_fp80 %Val)
7571 declare fp128 @llvm.floor.f128(fp128 %Val)
7572 declare ppc_fp128 @llvm.floor.ppcf128(ppc_fp128 %Val)
7573</pre>
7574
7575<h5>Overview:</h5>
7576<p>The '<tt>llvm.floor.*</tt>' intrinsics return the floor of
7577 the operand.</p>
7578
7579<h5>Arguments:</h5>
7580<p>The argument and return value are floating point numbers of the same
7581 type.</p>
7582
7583<h5>Semantics:</h5>
7584<p>This function returns the same values as the libm <tt>floor</tt> functions
7585 would, and handles error conditions in the same way.</p>
7586
7587</div>
7588
NAKAMURA Takumi4b2e07a2011-10-31 13:04:26 +0000[diff] [blame]7589</div>
7590
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7591<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7592<h3>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7593 <a name="int_manip">Bit Manipulation Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7594</h3>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7595
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7596<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7597
7598<p>LLVM provides intrinsics for a few important bit manipulation operations.
7599 These allow efficient code generation for some algorithms.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7600
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7601<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7602<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7603 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7604</h4>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7605
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7606<div>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7607
7608<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]7609<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7610 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7611
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7612<pre>
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7613 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
7614 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
7615 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7616</pre>
7617
7618<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7619<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7620 values with an even number of bytes (positive multiple of 16 bits). These
7621 are useful for performing operations on data that is not in the target's
7622 native byte order.</p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7623
7624<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7625<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7626 and low byte of the input i16 swapped. Similarly,
7627 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7628 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7629 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7630 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7631 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7632 more, respectively).</p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7633
7634</div>
7635
7636<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7637<h4>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]7638 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7639</h4>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7640
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7641<div>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7642
7643<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]7644<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]7645 width, or on any vector with integer elements. Not all targets support all
7646 bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7647
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7648<pre>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7649 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7650 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]7651 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7652 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
7653 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7654 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]7655</pre>
7656
7657<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7658<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7659 in a value.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7660
7661<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7662<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]7663 integer type, or a vector with integer elements.
7664 The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7665
7666<h5>Semantics:</h5>
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7667<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7668 element of a vector.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7669
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7670</div>
7671
7672<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7673<h4>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]7674 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7675</h4>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7676
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7677<div>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7678
7679<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7680<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]7681 integer bit width, or any vector whose elements are integers. Not all
7682 targets support all bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7683
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7684<pre>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7685 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7686 declare i16 @llvm.ctlz.i16 (i16 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7687 declare i32 @llvm.ctlz.i32 (i32 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7688 declare i64 @llvm.ctlz.i64 (i64 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7689 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7690 declase &lt;2 x i32&gt; @llvm.ctlz.v2i32(&lt;2 x i32&gt; &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7691</pre>
7692
7693<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7694<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7695 leading zeros in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7696
7697<h5>Arguments:</h5>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7698<p>The first argument is the value to be counted. This argument may be of any
7699 integer type, or a vectory with integer element type. The return type
7700 must match the first argument type.</p>
7701
7702<p>The second argument must be a constant and is a flag to indicate whether the
7703 intrinsic should ensure that a zero as the first argument produces a defined
7704 result. Historically some architectures did not provide a defined result for
7705 zero values as efficiently, and many algorithms are now predicated on
7706 avoiding zero-value inputs.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7707
7708<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7709<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7710 zeros in a variable, or within each element of the vector.
7711 If <tt>src == 0</tt> then the result is the size in bits of the type of
7712 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7713 For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7714
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7715</div>
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]7716
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7717<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7718<h4>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]7719 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7720</h4>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7721
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7722<div>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7723
7724<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7725<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]7726 integer bit width, or any vector of integer elements. Not all targets
7727 support all bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7728
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7729<pre>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7730 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7731 declare i16 @llvm.cttz.i16 (i16 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7732 declare i32 @llvm.cttz.i32 (i32 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7733 declare i64 @llvm.cttz.i64 (i64 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7734 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7735 declase &lt;2 x i32&gt; @llvm.cttz.v2i32(&lt;2 x i32&gt; &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7736</pre>
7737
7738<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7739<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7740 trailing zeros.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7741
7742<h5>Arguments:</h5>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7743<p>The first argument is the value to be counted. This argument may be of any
7744 integer type, or a vectory with integer element type. The return type
7745 must match the first argument type.</p>
7746
7747<p>The second argument must be a constant and is a flag to indicate whether the
7748 intrinsic should ensure that a zero as the first argument produces a defined
7749 result. Historically some architectures did not provide a defined result for
7750 zero values as efficiently, and many algorithms are now predicated on
7751 avoiding zero-value inputs.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7752
7753<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7754<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7755 zeros in a variable, or within each element of a vector.
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7756 If <tt>src == 0</tt> then the result is the size in bits of the type of
7757 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7758 For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7759
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7760</div>
7761
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7762</div>
7763
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7764<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7765<h3>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7766 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7767</h3>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7768
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7769<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7770
7771<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7772
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7773<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7774<h4>
7775 <a name="int_sadd_overflow">
7776 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7777 </a>
7778</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7779
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7780<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7781
7782<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7783<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]7784 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7785
7786<pre>
7787 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7788 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7789 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7790</pre>
7791
7792<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7793<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7794 a signed addition of the two arguments, and indicate whether an overflow
7795 occurred during the signed summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7796
7797<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7798<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]7799 be of integer types of any bit width, but they must have the same bit
7800 width. The second element of the result structure must be of
7801 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7802 undergo signed addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7803
7804<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7805<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7806 a signed addition of the two variables. They return a structure &mdash; the
7807 first element of which is the signed summation, and the second element of
7808 which is a bit specifying if the signed summation resulted in an
7809 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7810
7811<h5>Examples:</h5>
7812<pre>
7813 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7814 %sum = extractvalue {i32, i1} %res, 0
7815 %obit = extractvalue {i32, i1} %res, 1
7816 br i1 %obit, label %overflow, label %normal
7817</pre>
7818
7819</div>
7820
7821<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7822<h4>
7823 <a name="int_uadd_overflow">
7824 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7825 </a>
7826</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7827
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7828<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7829
7830<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7831<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]7832 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7833
7834<pre>
7835 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7836 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7837 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7838</pre>
7839
7840<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7841<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7842 an unsigned addition of the two arguments, and indicate whether a carry
7843 occurred during the unsigned summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7844
7845<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7846<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]7847 be of integer types of any bit width, but they must have the same bit
7848 width. The second element of the result structure must be of
7849 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7850 undergo unsigned addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7851
7852<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7853<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7854 an unsigned addition of the two arguments. They return a structure &mdash;
7855 the first element of which is the sum, and the second element of which is a
7856 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7857
7858<h5>Examples:</h5>
7859<pre>
7860 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7861 %sum = extractvalue {i32, i1} %res, 0
7862 %obit = extractvalue {i32, i1} %res, 1
7863 br i1 %obit, label %carry, label %normal
7864</pre>
7865
7866</div>
7867
7868<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7869<h4>
7870 <a name="int_ssub_overflow">
7871 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7872 </a>
7873</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7874
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7875<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7876
7877<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7878<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]7879 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7880
7881<pre>
7882 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7883 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7884 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7885</pre>
7886
7887<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7888<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7889 a signed subtraction of the two arguments, and indicate whether an overflow
7890 occurred during the signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7891
7892<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7893<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]7894 be of integer types of any bit width, but they must have the same bit
7895 width. The second element of the result structure must be of
7896 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7897 undergo signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7898
7899<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7900<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7901 a signed subtraction of the two arguments. They return a structure &mdash;
7902 the first element of which is the subtraction, and the second element of
7903 which is a bit specifying if the signed subtraction resulted in an
7904 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7905
7906<h5>Examples:</h5>
7907<pre>
7908 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7909 %sum = extractvalue {i32, i1} %res, 0
7910 %obit = extractvalue {i32, i1} %res, 1
7911 br i1 %obit, label %overflow, label %normal
7912</pre>
7913
7914</div>
7915
7916<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7917<h4>
7918 <a name="int_usub_overflow">
7919 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7920 </a>
7921</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7922
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7923<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7924
7925<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7926<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]7927 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7928
7929<pre>
7930 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7931 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7932 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7933</pre>
7934
7935<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7936<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7937 an unsigned subtraction of the two arguments, and indicate whether an
7938 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7939
7940<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7941<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]7942 be of integer types of any bit width, but they must have the same bit
7943 width. The second element of the result structure must be of
7944 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7945 undergo unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7946
7947<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7948<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7949 an unsigned subtraction of the two arguments. They return a structure &mdash;
7950 the first element of which is the subtraction, and the second element of
7951 which is a bit specifying if the unsigned subtraction resulted in an
7952 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7953
7954<h5>Examples:</h5>
7955<pre>
7956 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7957 %sum = extractvalue {i32, i1} %res, 0
7958 %obit = extractvalue {i32, i1} %res, 1
7959 br i1 %obit, label %overflow, label %normal
7960</pre>
7961
7962</div>
7963
7964<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7965<h4>
7966 <a name="int_smul_overflow">
7967 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7968 </a>
7969</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7970
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7971<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7972
7973<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7974<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]7975 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7976
7977<pre>
7978 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7979 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7980 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7981</pre>
7982
7983<h5>Overview:</h5>
7984
7985<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7986 a signed multiplication of the two arguments, and indicate whether an
7987 overflow occurred during the signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7988
7989<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7990<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]7991 be of integer types of any bit width, but they must have the same bit
7992 width. The second element of the result structure must be of
7993 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7994 undergo signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7995
7996<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7997<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7998 a signed multiplication of the two arguments. They return a structure &mdash;
7999 the first element of which is the multiplication, and the second element of
8000 which is a bit specifying if the signed multiplication resulted in an
8001 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]8002
8003<h5>Examples:</h5>
8004<pre>
8005 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
8006 %sum = extractvalue {i32, i1} %res, 0
8007 %obit = extractvalue {i32, i1} %res, 1
8008 br i1 %obit, label %overflow, label %normal
8009</pre>
8010
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]8011</div>
8012
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8013<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8014<h4>
8015 <a name="int_umul_overflow">
8016 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
8017 </a>
8018</h4>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8019
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8020<div>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8021
8022<h5>Syntax:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8023<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]8024 on any integer bit width.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8025
8026<pre>
8027 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
8028 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
8029 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
8030</pre>
8031
8032<h5>Overview:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8033<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8034 a unsigned multiplication of the two arguments, and indicate whether an
8035 overflow occurred during the unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8036
8037<h5>Arguments:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8038<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]8039 be of integer types of any bit width, but they must have the same bit
8040 width. The second element of the result structure must be of
8041 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
8042 undergo unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8043
8044<h5>Semantics:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8045<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8046 an unsigned multiplication of the two arguments. They return a structure
8047 &mdash; the first element of which is the multiplication, and the second
8048 element of which is a bit specifying if the unsigned multiplication resulted
8049 in an overflow.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]8050
8051<h5>Examples:</h5>
8052<pre>
8053 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
8054 %sum = extractvalue {i32, i1} %res, 0
8055 %obit = extractvalue {i32, i1} %res, 1
8056 br i1 %obit, label %overflow, label %normal
8057</pre>
8058
8059</div>
8060
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8061</div>
8062
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]8063<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8064<h3>
Lang Hames5afba6f2012-06-05 19:07:46 +0000[diff] [blame]8065 <a name="spec_arithmetic">Specialised Arithmetic Intrinsics</a>
8066</h3>
8067
8068<!-- _______________________________________________________________________ -->
8069
8070<h4>
8071 <a name="fmuladd">'<tt>llvm.fmuladd.*</tt>' Intrinsic</a>
8072</h4>
8073
8074<div>
8075
8076<h5>Syntax:</h5>
8077<pre>
8078 declare float @llvm.fmuladd.f32(float %a, float %b, float %c)
8079 declare double @llvm.fmuladd.f64(double %a, double %b, double %c)
8080</pre>
8081
8082<h5>Overview:</h5>
8083<p>The '<tt>llvm.fmuladd.*</tt>' intrinsic functions represent multiply-add
8084expressions that can be fused if the code generator determines that the fused
8085expression would be legal and efficient.</p>
8086
8087<h5>Arguments:</h5>
8088<p>The '<tt>llvm.fmuladd.*</tt>' intrinsics each take three arguments: two
8089multiplicands, a and b, and an addend c.</p>
8090
8091<h5>Semantics:</h5>
8092<p>The expression:</p>
8093<pre>
8094 %0 = call float @llvm.fmuladd.f32(%a, %b, %c)
8095</pre>
8096<p>is equivalent to the expression a * b + c, except that rounding will not be
8097performed between the multiplication and addition steps if the code generator
8098fuses the operations. Fusion is not guaranteed, even if the target platform
8099supports it. If a fused multiply-add is required the corresponding llvm.fma.*
8100intrinsic function should be used instead.</p>
8101
8102<h5>Examples:</h5>
8103<pre>
8104 %r2 = call float @llvm.fmuladd.f32(float %a, float %b, float %c) ; yields {float}:r2 = (a * b) + c
8105</pre>
8106
8107</div>
8108
8109<!-- ======================================================================= -->
8110<h3>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8111 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8112</h3>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8113
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8114<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8115
Tobias Grosser057beb82012-05-24 15:59:06 +0000[diff] [blame]8116<p>For most target platforms, half precision floating point is a storage-only
8117 format. This means that it is
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]8118 a dense encoding (in memory) but does not support computation in the
8119 format.</p>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]8120
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]8121<p>This means that code must first load the half-precision floating point
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]8122 value as an i16, then convert it to float with <a
8123 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
8124 Computation can then be performed on the float value (including extending to
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]8125 double etc). To store the value back to memory, it is first converted to
8126 float if needed, then converted to i16 with
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]8127 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
8128 storing as an i16 value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8129
8130<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8131<h4>
8132 <a name="int_convert_to_fp16">
8133 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
8134 </a>
8135</h4>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8136
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8137<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8138
8139<h5>Syntax:</h5>
8140<pre>
8141 declare i16 @llvm.convert.to.fp16(f32 %a)
8142</pre>
8143
8144<h5>Overview:</h5>
8145<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
8146 a conversion from single precision floating point format to half precision
8147 floating point format.</p>
8148
8149<h5>Arguments:</h5>
8150<p>The intrinsic function contains single argument - the value to be
8151 converted.</p>
8152
8153<h5>Semantics:</h5>
8154<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
8155 a conversion from single precision floating point format to half precision
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]8156 floating point format. The return value is an <tt>i16</tt> which
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]8157 contains the converted number.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8158
8159<h5>Examples:</h5>
8160<pre>
8161 %res = call i16 @llvm.convert.to.fp16(f32 %a)
8162 store i16 %res, i16* @x, align 2
8163</pre>
8164
8165</div>
8166
8167<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8168<h4>
8169 <a name="int_convert_from_fp16">
8170 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
8171 </a>
8172</h4>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8173
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8174<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8175
8176<h5>Syntax:</h5>
8177<pre>
8178 declare f32 @llvm.convert.from.fp16(i16 %a)
8179</pre>
8180
8181<h5>Overview:</h5>
8182<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
8183 a conversion from half precision floating point format to single precision
8184 floating point format.</p>
8185
8186<h5>Arguments:</h5>
8187<p>The intrinsic function contains single argument - the value to be
8188 converted.</p>
8189
8190<h5>Semantics:</h5>
8191<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]8192 conversion from half single precision floating point format to single
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]8193 precision floating point format. The input half-float value is represented by
8194 an <tt>i16</tt> value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8195
8196<h5>Examples:</h5>
8197<pre>
8198 %a = load i16* @x, align 2
8199 %res = call f32 @llvm.convert.from.fp16(i16 %a)
8200</pre>
8201
8202</div>
8203
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8204</div>
8205
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]8206<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8207<h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]8208 <a name="int_debugger">Debugger Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8209</h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]8210
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8211<div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]8212
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8213<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
8214 prefix), are described in
8215 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
8216 Level Debugging</a> document.</p>
8217
8218</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]8219
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8220<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8221<h3>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8222 <a name="int_eh">Exception Handling Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8223</h3>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8224
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8225<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8226
8227<p>The LLVM exception handling intrinsics (which all start with
8228 <tt>llvm.eh.</tt> prefix), are described in
8229 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
8230 Handling</a> document.</p>
8231
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8232</div>
8233
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8234<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8235<h3>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]8236 <a name="int_trampoline">Trampoline Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8237</h3>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8238
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8239<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8240
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]8241<p>These intrinsics make it possible to excise one parameter, marked with
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]8242 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
8243 The result is a callable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8244 function pointer lacking the nest parameter - the caller does not need to
8245 provide a value for it. Instead, the value to use is stored in advance in a
8246 "trampoline", a block of memory usually allocated on the stack, which also
8247 contains code to splice the nest value into the argument list. This is used
8248 to implement the GCC nested function address extension.</p>
8249
8250<p>For example, if the function is
8251 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
8252 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
8253 follows:</p>
8254
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]8255<pre class="doc_code">
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]8256 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
8257 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]8258 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
8259 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]8260 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8261</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8262
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8263<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
8264 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8265
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8266<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8267<h4>
8268 <a name="int_it">
8269 '<tt>llvm.init.trampoline</tt>' Intrinsic
8270 </a>
8271</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8272
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8273<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8274
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8275<h5>Syntax:</h5>
8276<pre>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]8277 declare void @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]8278</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8279
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8280<h5>Overview:</h5>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]8281<p>This fills the memory pointed to by <tt>tramp</tt> with executable code,
8282 turning it into a trampoline.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8283
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8284<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8285<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
8286 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
8287 sufficiently aligned block of memory; this memory is written to by the
8288 intrinsic. Note that the size and the alignment are target-specific - LLVM
8289 currently provides no portable way of determining them, so a front-end that
8290 generates this intrinsic needs to have some target-specific knowledge.
8291 The <tt>func</tt> argument must hold a function bitcast to
8292 an <tt>i8*</tt>.</p>
8293
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8294<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8295<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]8296 dependent code, turning it into a function. Then <tt>tramp</tt> needs to be
8297 passed to <a href="#int_at">llvm.adjust.trampoline</a> to get a pointer
8298 which can be <a href="#int_trampoline">bitcast (to a new function) and
8299 called</a>. The new function's signature is the same as that of
8300 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
8301 removed. At most one such <tt>nest</tt> argument is allowed, and it must be of
8302 pointer type. Calling the new function is equivalent to calling <tt>func</tt>
8303 with the same argument list, but with <tt>nval</tt> used for the missing
8304 <tt>nest</tt> argument. If, after calling <tt>llvm.init.trampoline</tt>, the
8305 memory pointed to by <tt>tramp</tt> is modified, then the effect of any later call
8306 to the returned function pointer is undefined.</p>
8307</div>
8308
8309<!-- _______________________________________________________________________ -->
8310<h4>
8311 <a name="int_at">
8312 '<tt>llvm.adjust.trampoline</tt>' Intrinsic
8313 </a>
8314</h4>
8315
8316<div>
8317
8318<h5>Syntax:</h5>
8319<pre>
8320 declare i8* @llvm.adjust.trampoline(i8* &lt;tramp&gt;)
8321</pre>
8322
8323<h5>Overview:</h5>
8324<p>This performs any required machine-specific adjustment to the address of a
8325 trampoline (passed as <tt>tramp</tt>).</p>
8326
8327<h5>Arguments:</h5>
8328<p><tt>tramp</tt> must point to a block of memory which already has trampoline code
8329 filled in by a previous call to <a href="#int_it"><tt>llvm.init.trampoline</tt>
8330 </a>.</p>
8331
8332<h5>Semantics:</h5>
8333<p>On some architectures the address of the code to be executed needs to be
8334 different to the address where the trampoline is actually stored. This
8335 intrinsic returns the executable address corresponding to <tt>tramp</tt>
8336 after performing the required machine specific adjustments.
8337 The pointer returned can then be <a href="#int_trampoline"> bitcast and
8338 executed</a>.
8339</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8340
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8341</div>
8342
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8343</div>
8344
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]8345<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8346<h3>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8347 <a name="int_memorymarkers">Memory Use Markers</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8348</h3>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8349
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8350<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8351
8352<p>This class of intrinsics exists to information about the lifetime of memory
8353 objects and ranges where variables are immutable.</p>
8354
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8355<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8356<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8357 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8358</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8359
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8360<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8361
8362<h5>Syntax:</h5>
8363<pre>
8364 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8365</pre>
8366
8367<h5>Overview:</h5>
8368<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
8369 object's lifetime.</p>
8370
8371<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8372<p>The first argument is a constant integer representing the size of the
8373 object, or -1 if it is variable sized. The second argument is a pointer to
8374 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8375
8376<h5>Semantics:</h5>
8377<p>This intrinsic indicates that before this point in the code, the value of the
8378 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]8379 never be used and has an undefined value. A load from the pointer that
8380 precedes this intrinsic can be replaced with
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8381 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
8382
8383</div>
8384
8385<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8386<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8387 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8388</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8389
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8390<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8391
8392<h5>Syntax:</h5>
8393<pre>
8394 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8395</pre>
8396
8397<h5>Overview:</h5>
8398<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
8399 object's lifetime.</p>
8400
8401<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8402<p>The first argument is a constant integer representing the size of the
8403 object, or -1 if it is variable sized. The second argument is a pointer to
8404 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8405
8406<h5>Semantics:</h5>
8407<p>This intrinsic indicates that after this point in the code, the value of the
8408 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8409 never be used and has an undefined value. Any stores into the memory object
8410 following this intrinsic may be removed as dead.
8411
8412</div>
8413
8414<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8415<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8416 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8417</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8418
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8419<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8420
8421<h5>Syntax:</h5>
8422<pre>
Nick Lewycky29b6cb42010-11-30 04:13:41 +0000[diff] [blame]8423 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8424</pre>
8425
8426<h5>Overview:</h5>
8427<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
8428 a memory object will not change.</p>
8429
8430<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8431<p>The first argument is a constant integer representing the size of the
8432 object, or -1 if it is variable sized. The second argument is a pointer to
8433 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8434
8435<h5>Semantics:</h5>
8436<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
8437 the return value, the referenced memory location is constant and
8438 unchanging.</p>
8439
8440</div>
8441
8442<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8443<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8444 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8445</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8446
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8447<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8448
8449<h5>Syntax:</h5>
8450<pre>
8451 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8452</pre>
8453
8454<h5>Overview:</h5>
8455<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8456 a memory object are mutable.</p>
8457
8458<h5>Arguments:</h5>
8459<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8460 The second argument is a constant integer representing the size of the
8461 object, or -1 if it is variable sized and the third argument is a pointer
8462 to the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8463
8464<h5>Semantics:</h5>
8465<p>This intrinsic indicates that the memory is mutable again.</p>
8466
8467</div>
8468
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8469</div>
8470
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]8471<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8472<h3>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8473 <a name="int_general">General Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8474</h3>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8475
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8476<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8477
8478<p>This class of intrinsics is designed to be generic and has no specific
8479 purpose.</p>
8480
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8481<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8482<h4>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8483 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8484</h4>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8485
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8486<div>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8487
8488<h5>Syntax:</h5>
8489<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8490 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]8491</pre>
8492
8493<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8494<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8495
8496<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8497<p>The first argument is a pointer to a value, the second is a pointer to a
8498 global string, the third is a pointer to a global string which is the source
8499 file name, and the last argument is the line number.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8500
8501<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8502<p>This intrinsic allows annotation of local variables with arbitrary strings.
8503 This can be useful for special purpose optimizations that want to look for
John Criswelle865c032011-08-19 16:57:55 +0000[diff] [blame]8504 these annotations. These have no other defined use; they are ignored by code
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8505 generation and optimization.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8506
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8507</div>
8508
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8509<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8510<h4>
Tanya Lattnere1a8da02007-09-21 23:57:59 +0000[diff] [blame]8511 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8512</h4>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8513
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8514<div>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8515
8516<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8517<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8518 any integer bit width.</p>
8519
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8520<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8521 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8522 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8523 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8524 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8525 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]8526</pre>
8527
8528<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8529<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8530
8531<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8532<p>The first argument is an integer value (result of some expression), the
8533 second is a pointer to a global string, the third is a pointer to a global
8534 string which is the source file name, and the last argument is the line
8535 number. It returns the value of the first argument.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8536
8537<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8538<p>This intrinsic allows annotations to be put on arbitrary expressions with
8539 arbitrary strings. This can be useful for special purpose optimizations that
John Criswelle865c032011-08-19 16:57:55 +0000[diff] [blame]8540 want to look for these annotations. These have no other defined use; they
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8541 are ignored by code generation and optimization.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8542
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8543</div>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8544
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8545<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8546<h4>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8547 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8548</h4>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8549
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8550<div>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8551
8552<h5>Syntax:</h5>
8553<pre>
Chris Lattner86208902012-05-27 23:20:41 +0000[diff] [blame]8554 declare void @llvm.trap() noreturn nounwind
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8555</pre>
8556
8557<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8558<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8559
8560<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8561<p>None.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8562
8563<h5>Semantics:</h5>
John Criswell99534f32012-05-16 00:26:51 +0000[diff] [blame]8564<p>This intrinsic is lowered to the target dependent trap instruction. If the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8565 target does not have a trap instruction, this intrinsic will be lowered to
John Criswell99534f32012-05-16 00:26:51 +0000[diff] [blame]8566 a call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8567
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8568</div>
8569
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8570<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8571<h4>
Dan Gohmana6063c62012-05-14 18:58:10 +0000[diff] [blame]8572 <a name="int_debugtrap">'<tt>llvm.debugtrap</tt>' Intrinsic</a>
Dan Gohmand4347e12012-05-11 00:19:32 +0000[diff] [blame]8573</h4>
8574
8575<div>
8576
8577<h5>Syntax:</h5>
8578<pre>
Chris Lattner86208902012-05-27 23:20:41 +0000[diff] [blame]8579 declare void @llvm.debugtrap() nounwind
Dan Gohmand4347e12012-05-11 00:19:32 +0000[diff] [blame]8580</pre>
8581
8582<h5>Overview:</h5>
Dan Gohmana6063c62012-05-14 18:58:10 +0000[diff] [blame]8583<p>The '<tt>llvm.debugtrap</tt>' intrinsic.</p>
Dan Gohmand4347e12012-05-11 00:19:32 +0000[diff] [blame]8584
8585<h5>Arguments:</h5>
8586<p>None.</p>
8587
8588<h5>Semantics:</h5>
8589<p>This intrinsic is lowered to code which is intended to cause an execution
8590 trap with the intention of requesting the attention of a debugger.</p>
8591
8592</div>
8593
8594<!-- _______________________________________________________________________ -->
8595<h4>
Misha Brukmandccb0252008-11-22 23:55:29 +0000[diff] [blame]8596 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8597</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8598
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8599<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8600
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8601<h5>Syntax:</h5>
8602<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8603 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8604</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8605
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8606<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8607<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8608 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8609 ensure that it is placed on the stack before local variables.</p>
8610
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8611<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8612<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8613 arguments. The first argument is the value loaded from the stack
8614 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8615 that has enough space to hold the value of the guard.</p>
8616
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8617<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8618<p>This intrinsic causes the prologue/epilogue inserter to force the position of
8619 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8620 stack. This is to ensure that if a local variable on the stack is
8621 overwritten, it will destroy the value of the guard. When the function exits,
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8622 the guard on the stack is checked against the original guard. If they are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8623 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8624 function.</p>
8625
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8626</div>
8627
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8628<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8629<h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8630 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8631</h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8632
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8633<div>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8634
8635<h5>Syntax:</h5>
8636<pre>
Nuno Lopes23e75da2012-05-22 15:25:31 +0000[diff] [blame]8637 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;min&gt;)
8638 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;min&gt;)
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8639</pre>
8640
8641<h5>Overview:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8642<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8643 the optimizers to determine at compile time whether a) an operation (like
8644 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8645 runtime check for overflow isn't necessary. An object in this context means
8646 an allocation of a specific class, structure, array, or other object.</p>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8647
8648<h5>Arguments:</h5>
Nuno Lopes23e75da2012-05-22 15:25:31 +0000[diff] [blame]8649<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher8295a0a2009-12-23 00:29:49 +0000[diff] [blame]8650 argument is a pointer to or into the <tt>object</tt>. The second argument
Nuno Lopes23e75da2012-05-22 15:25:31 +0000[diff] [blame]8651 is a boolean and determines whether <tt>llvm.objectsize</tt> returns 0 (if
8652 true) or -1 (if false) when the object size is unknown.
8653 The second argument only accepts constants.</p>
Eric Christopher8295a0a2009-12-23 00:29:49 +0000[diff] [blame]8654
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8655<h5>Semantics:</h5>
Nuno Lopes30759542012-05-09 15:52:43 +0000[diff] [blame]8656<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to a constant representing
8657 the size of the object concerned. If the size cannot be determined at compile
Nuno Lopes23e75da2012-05-22 15:25:31 +0000[diff] [blame]8658 time, <tt>llvm.objectsize</tt> returns <tt>i32/i64 -1 or 0</tt>
8659 (depending on the <tt>min</tt> argument).</p>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8660
8661</div>
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]8662<!-- _______________________________________________________________________ -->
8663<h4>
8664 <a name="int_expect">'<tt>llvm.expect</tt>' Intrinsic</a>
8665</h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8666
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]8667<div>
8668
8669<h5>Syntax:</h5>
8670<pre>
8671 declare i32 @llvm.expect.i32(i32 &lt;val&gt;, i32 &lt;expected_val&gt;)
8672 declare i64 @llvm.expect.i64(i64 &lt;val&gt;, i64 &lt;expected_val&gt;)
8673</pre>
8674
8675<h5>Overview:</h5>
8676<p>The <tt>llvm.expect</tt> intrinsic provides information about expected (the
8677 most probable) value of <tt>val</tt>, which can be used by optimizers.</p>
8678
8679<h5>Arguments:</h5>
8680<p>The <tt>llvm.expect</tt> intrinsic takes two arguments. The first
8681 argument is a value. The second argument is an expected value, this needs to
8682 be a constant value, variables are not allowed.</p>
8683
8684<h5>Semantics:</h5>
8685<p>This intrinsic is lowered to the <tt>val</tt>.</p>
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8686</div>
8687
Nuno Lopesb0c76d72012-07-05 17:37:07 +0000[diff] [blame]8688<!-- _______________________________________________________________________ -->
8689<h4>
8690 <a name="int_donothing">'<tt>llvm.donothing</tt>' Intrinsic</a>
8691</h4>
8692
8693<div>
8694
8695<h5>Syntax:</h5>
8696<pre>
8697 declare void @llvm.donothing() nounwind readnone
8698</pre>
8699
8700<h5>Overview:</h5>
8701<p>The <tt>llvm.donothing</tt> intrinsic doesn't perform any operation. It's the
8702only intrinsic that can be called with an invoke instruction.</p>
8703
8704<h5>Arguments:</h5>
8705<p>None.</p>
8706
8707<h5>Semantics:</h5>
8708<p>This intrinsic does nothing, and it's removed by optimizers and ignored by
8709codegen.</p>
8710</div>
8711
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8712</div>
8713
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]8714</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]8715<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]8716<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8717<address>
8718 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]8719 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8720 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]8721 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]8722
8723 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
NAKAMURA Takumib9a33632011-04-09 02:13:37 +0000[diff] [blame]8724 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8725 Last modified: $Date$
8726</address>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +0000[diff] [blame]8727
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]8728</body>
8729</html>