Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / 20536b5bca2edfd8184ad54523ab9678931ed13a / . / docs / LangRef.html
blob: 90308a45d223a8bd83552d08fbe2b2c471389f2e [file] [log] [blame]
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencer3921c742004-08-26 20:44:00 +0000[diff] [blame]6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]8 <meta name="description"
Reid Spencer3921c742004-08-26 20:44:00 +0000[diff] [blame]9 content="LLVM Assembly Language Reference Manual.">
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]12
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]13<body>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]14
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]15<h1>LLVM Language Reference Manual</h1>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]16<ol>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a>
24 <ol>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]27 <li><a href="#linkage_linker_private_weak">'<tt>linker_private_weak</tt>' Linkage</a></li>
Bill Wendling55ae5152010-08-20 22:05:50 +0000[diff] [blame]28 <li><a href="#linkage_linker_private_weak_def_auto">'<tt>linker_private_weak_def_auto</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]29 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
30 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
31 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
32 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
33 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
34 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
35 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner5a2d8752009-10-10 18:26:06 +0000[diff] [blame]36 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendling987e7eb2009-07-20 02:41:50 +0000[diff] [blame]37 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
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>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]106 <li><a href="#fpaccuracy">'<tt>fpaccuracy</tt>' Metadata</a></li>
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]107 </ol>
108 </li>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]109 </ol>
110 </li>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]111 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
112 <ol>
113 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]114 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
115 Global Variable</a></li>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]116 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
117 Global Variable</a></li>
118 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
119 Global Variable</a></li>
120 </ol>
121 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]122 <li><a href="#instref">Instruction Reference</a>
123 <ol>
124 <li><a href="#terminators">Terminator Instructions</a>
125 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]126 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
127 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]128 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]129 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]130 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]131 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]132 <li><a href="#i_resume">'<tt>resume</tt>' Instruction</a></li>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]133 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]134 </ol>
135 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]136 <li><a href="#binaryops">Binary Operations</a>
137 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]138 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]139 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]140 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]141 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]142 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]143 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]144 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
145 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
146 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]147 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
148 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
149 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]150 </ol>
151 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]152 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
153 <ol>
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]154 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
155 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
156 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]157 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]158 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]159 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]160 </ol>
161 </li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]162 <li><a href="#vectorops">Vector Operations</a>
163 <ol>
164 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
165 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
166 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]167 </ol>
168 </li>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]169 <li><a href="#aggregateops">Aggregate Operations</a>
170 <ol>
171 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
172 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
173 </ol>
174 </li>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]175 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]176 <ol>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]177 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
178 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
179 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
180 <li><a href="#i_fence">'<tt>fence</tt>' Instruction</a></li>
181 <li><a href="#i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a></li>
182 <li><a href="#i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a></li>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]183 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]184 </ol>
185 </li>
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]186 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]187 <ol>
188 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
189 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
190 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
191 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
192 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]193 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
194 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
195 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
196 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]197 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
198 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]199 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]200 </ol>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]201 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]202 <li><a href="#otherops">Other Operations</a>
203 <ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]204 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
205 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]206 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]207 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]208 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]209 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]210 <li><a href="#i_landingpad">'<tt>landingpad</tt>' Instruction</a></li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]211 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]212 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]213 </ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]214 </li>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]215 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]216 <ol>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]217 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
218 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]219 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
220 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
221 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]222 </ol>
223 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]224 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
225 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]226 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
227 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
228 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]229 </ol>
230 </li>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]231 <li><a href="#int_codegen">Code Generator Intrinsics</a>
232 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]233 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
234 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
235 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
236 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
237 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
238 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohman31f1af12010-05-26 21:56:15 +0000[diff] [blame]239 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswell7123e272004-04-09 16:43:20 +0000[diff] [blame]240 </ol>
241 </li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]242 <li><a href="#int_libc">Standard C Library Intrinsics</a>
243 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]244 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
245 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
246 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
247 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
248 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]249 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
250 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
251 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohman08b280b2011-05-27 00:36:31 +0000[diff] [blame]252 <li><a href="#int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a></li>
253 <li><a href="#int_log">'<tt>llvm.log.*</tt>' Intrinsic</a></li>
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]254 <li><a href="#int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a></li>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]255 </ol>
256 </li>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]257 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]258 <ol>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]259 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]260 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
261 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
262 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]263 </ol>
264 </li>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]265 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
266 <ol>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]267 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
268 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
269 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
270 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
271 <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]272 <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]273 </ol>
274 </li>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]275 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
276 <ol>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]277 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
278 <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]279 </ol>
280 </li>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]281 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]282 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]283 <li><a href="#int_trampoline">Trampoline Intrinsics</a>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]284 <ol>
285 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]286 <li><a href="#int_at">'<tt>llvm.adjust.trampoline</tt>' Intrinsic</a></li>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]287 </ol>
288 </li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]289 <li><a href="#int_memorymarkers">Memory Use Markers</a>
290 <ol>
Jakub Staszak8e1b12a2011-12-04 20:44:25 +0000[diff] [blame]291 <li><a href="#int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a></li>
292 <li><a href="#int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a></li>
293 <li><a href="#int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a></li>
294 <li><a href="#int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a></li>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]295 </ol>
296 </li>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame]297 <li><a href="#int_general">General intrinsics</a>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]298 <ol>
Reid Spencer20677642007-07-20 19:59:11 +0000[diff] [blame]299 <li><a href="#int_var_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]300 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]301 <li><a href="#int_annotation">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]302 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]303 <li><a href="#int_trap">
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]304 '<tt>llvm.trap</tt>' Intrinsic</a></li>
305 <li><a href="#int_stackprotector">
306 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]307 <li><a href="#int_objectsize">
308 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]309 <li><a href="#int_expect">
310 '<tt>llvm.expect</tt>' Intrinsic</a></li>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]311 </ol>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]312 </li>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]313 </ol>
314 </li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]315</ol>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]316
317<div class="doc_author">
318 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
319 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]320</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]321
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]322<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]323<h2><a name="abstract">Abstract</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]324<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]325
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]326<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]327
328<p>This document is a reference manual for the LLVM assembly language. LLVM is
329 a Static Single Assignment (SSA) based representation that provides type
330 safety, low-level operations, flexibility, and the capability of representing
331 'all' high-level languages cleanly. It is the common code representation
332 used throughout all phases of the LLVM compilation strategy.</p>
333
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]334</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]335
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]336<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]337<h2><a name="introduction">Introduction</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]338<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]339
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]340<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]341
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]342<p>The LLVM code representation is designed to be used in three different forms:
343 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
344 for fast loading by a Just-In-Time compiler), and as a human readable
345 assembly language representation. This allows LLVM to provide a powerful
346 intermediate representation for efficient compiler transformations and
347 analysis, while providing a natural means to debug and visualize the
348 transformations. The three different forms of LLVM are all equivalent. This
349 document describes the human readable representation and notation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]350
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]351<p>The LLVM representation aims to be light-weight and low-level while being
352 expressive, typed, and extensible at the same time. It aims to be a
353 "universal IR" of sorts, by being at a low enough level that high-level ideas
354 may be cleanly mapped to it (similar to how microprocessors are "universal
355 IR's", allowing many source languages to be mapped to them). By providing
356 type information, LLVM can be used as the target of optimizations: for
357 example, through pointer analysis, it can be proven that a C automatic
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]358 variable is never accessed outside of the current function, allowing it to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]359 be promoted to a simple SSA value instead of a memory location.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]360
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]361<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]362<h4>
363 <a name="wellformed">Well-Formedness</a>
364</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]365
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]366<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]367
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]368<p>It is important to note that this document describes 'well formed' LLVM
369 assembly language. There is a difference between what the parser accepts and
370 what is considered 'well formed'. For example, the following instruction is
371 syntactically okay, but not well formed:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]372
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]373<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]374%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]375</pre>
376
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]377<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
378 LLVM infrastructure provides a verification pass that may be used to verify
379 that an LLVM module is well formed. This pass is automatically run by the
380 parser after parsing input assembly and by the optimizer before it outputs
381 bitcode. The violations pointed out by the verifier pass indicate bugs in
382 transformation passes or input to the parser.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]383
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]384</div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]385
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]386</div>
387
Chris Lattnercc689392007-10-03 17:34:29 +0000[diff] [blame]388<!-- Describe the typesetting conventions here. -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]389
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]390<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]391<h2><a name="identifiers">Identifiers</a></h2>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]392<!-- *********************************************************************** -->
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]393
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]394<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]395
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]396<p>LLVM identifiers come in two basic types: global and local. Global
397 identifiers (functions, global variables) begin with the <tt>'@'</tt>
398 character. Local identifiers (register names, types) begin with
399 the <tt>'%'</tt> character. Additionally, there are three different formats
400 for identifiers, for different purposes:</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]401
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]402<ol>
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]403 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]404 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
405 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
406 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
407 other characters in their names can be surrounded with quotes. Special
408 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
409 ASCII code for the character in hexadecimal. In this way, any character
410 can be used in a name value, even quotes themselves.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]411
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]412 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]413 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]414
Reid Spencercc16dc32004-12-09 18:02:53 +0000[diff] [blame]415 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]416 constants</a>, below.</li>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]417</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]418
Reid Spencer2c452282007-08-07 14:34:28 +0000[diff] [blame]419<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]420 don't need to worry about name clashes with reserved words, and the set of
421 reserved words may be expanded in the future without penalty. Additionally,
422 unnamed identifiers allow a compiler to quickly come up with a temporary
423 variable without having to avoid symbol table conflicts.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]424
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]425<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]426 languages. There are keywords for different opcodes
427 ('<tt><a href="#i_add">add</a></tt>',
428 '<tt><a href="#i_bitcast">bitcast</a></tt>',
429 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
430 ('<tt><a href="#t_void">void</a></tt>',
431 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
432 reserved words cannot conflict with variable names, because none of them
433 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]434
435<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]436 '<tt>%X</tt>' by 8:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]437
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]438<p>The easy way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]439
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]440<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]441%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]442</pre>
443
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]444<p>After strength reduction:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]445
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]446<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]447%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]448</pre>
449
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]450<p>And the hard way:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]451
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]452<pre class="doc_code">
Gabor Greifec58f752009-10-28 13:05:07 +0000[diff] [blame]453%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
454%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]455%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]456</pre>
457
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]458<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
459 lexical features of LLVM:</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]460
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]461<ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]462 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]463 line.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]464
465 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]466 assigned to a named value.</li>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]467
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]468 <li>Unnamed temporaries are numbered sequentially</li>
469</ol>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]470
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]471<p>It also shows a convention that we follow in this document. When
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]472 demonstrating instructions, we will follow an instruction with a comment that
473 defines the type and name of value produced. Comments are shown in italic
474 text.</p>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]475
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]476</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]477
478<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]479<h2><a name="highlevel">High Level Structure</a></h2>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]480<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]481<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]482<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]483<h3>
484 <a name="modulestructure">Module Structure</a>
485</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]486
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]487<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]488
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]489<p>LLVM programs are composed of "Module"s, each of which is a translation unit
490 of the input programs. Each module consists of functions, global variables,
491 and symbol table entries. Modules may be combined together with the LLVM
492 linker, which merges function (and global variable) definitions, resolves
493 forward declarations, and merges symbol table entries. Here is an example of
494 the "hello world" module:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]495
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]496<pre class="doc_code">
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]497<i>; Declare the string constant as a global constant.</i>&nbsp;
Nick Lewyckydb9cd762011-01-29 01:09:53 +0000[diff] [blame]498<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a>&nbsp;<a href="#globalvars">constant</a>&nbsp;<a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]499
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]500<i>; External declaration of the puts function</i>&nbsp;
501<a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]502
503<i>; Definition of main function</i>
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]504define i32 @main() { <i>; i32()* </i>&nbsp;
505 <i>; Convert [13 x i8]* to i8 *...</i>&nbsp;
506 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>&nbsp;
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]507
Chris Lattner63e4ccb2010-08-17 17:13:42 +0000[diff] [blame]508 <i>; Call puts function to write out the string to stdout.</i>&nbsp;
509 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>&nbsp;
510 <a href="#i_ret">ret</a> i32 0&nbsp;
511}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]512
513<i>; Named metadata</i>
514!1 = metadata !{i32 41}
515!foo = !{!1, null}
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]516</pre>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]517
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]518<p>This example is made up of a <a href="#globalvars">global variable</a> named
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]519 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]520 a <a href="#functionstructure">function definition</a> for
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]521 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
522 "<tt>foo"</tt>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]523
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]524<p>In general, a module is made up of a list of global values, where both
525 functions and global variables are global values. Global values are
526 represented by a pointer to a memory location (in this case, a pointer to an
527 array of char, and a pointer to a function), and have one of the
528 following <a href="#linkage">linkage types</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]529
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]530</div>
531
532<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]533<h3>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]534 <a name="linkage">Linkage Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]535</h3>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]536
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]537<div>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]538
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]539<p>All Global Variables and Functions have one of the following types of
540 linkage:</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]541
542<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]543 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]544 <dd>Global values with "<tt>private</tt>" linkage are only directly accessible
545 by objects in the current module. In particular, linking code into a
546 module with an private global value may cause the private to be renamed as
547 necessary to avoid collisions. Because the symbol is private to the
548 module, all references can be updated. This doesn't show up in any symbol
549 table in the object file.</dd>
Rafael Espindolabb46f522009-01-15 20:18:42 +0000[diff] [blame]550
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]551 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]552 <dd>Similar to <tt>private</tt>, but the symbol is passed through the
553 assembler and evaluated by the linker. Unlike normal strong symbols, they
554 are removed by the linker from the final linked image (executable or
555 dynamic library).</dd>
556
557 <dt><tt><b><a name="linkage_linker_private_weak">linker_private_weak</a></b></tt></dt>
558 <dd>Similar to "<tt>linker_private</tt>", but the symbol is weak. Note that
559 <tt>linker_private_weak</tt> symbols are subject to coalescing by the
560 linker. The symbols are removed by the linker from the final linked image
561 (executable or dynamic library).</dd>
Bill Wendling3d10a5a2009-07-20 01:03:30 +0000[diff] [blame]562
Bill Wendling55ae5152010-08-20 22:05:50 +0000[diff] [blame]563 <dt><tt><b><a name="linkage_linker_private_weak_def_auto">linker_private_weak_def_auto</a></b></tt></dt>
564 <dd>Similar to "<tt>linker_private_weak</tt>", but it's known that the address
565 of the object is not taken. For instance, functions that had an inline
566 definition, but the compiler decided not to inline it. Note,
567 unlike <tt>linker_private</tt> and <tt>linker_private_weak</tt>,
568 <tt>linker_private_weak_def_auto</tt> may have only <tt>default</tt>
569 visibility. The symbols are removed by the linker from the final linked
570 image (executable or dynamic library).</dd>
571
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]572 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendling07d31772010-06-29 22:34:52 +0000[diff] [blame]573 <dd>Similar to private, but the value shows as a local symbol
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]574 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
575 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]576
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]577 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000[diff] [blame]578 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]579 into the object file corresponding to the LLVM module. They exist to
580 allow inlining and other optimizations to take place given knowledge of
581 the definition of the global, which is known to be somewhere outside the
582 module. Globals with <tt>available_externally</tt> linkage are allowed to
583 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
584 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner266c7bb2009-04-13 05:44:34 +0000[diff] [blame]585
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]586 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Chris Lattner4887bd82007-01-14 06:51:48 +0000[diff] [blame]587 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattner873187c2010-01-09 19:15:14 +0000[diff] [blame]588 the same name when linkage occurs. This can be used to implement
589 some forms of inline functions, templates, or other code which must be
590 generated in each translation unit that uses it, but where the body may
591 be overridden with a more definitive definition later. Unreferenced
592 <tt>linkonce</tt> globals are allowed to be discarded. Note that
593 <tt>linkonce</tt> linkage does not actually allow the optimizer to
594 inline the body of this function into callers because it doesn't know if
595 this definition of the function is the definitive definition within the
596 program or whether it will be overridden by a stronger definition.
597 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
598 linkage.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]599
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]600 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]601 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
602 <tt>linkonce</tt> linkage, except that unreferenced globals with
603 <tt>weak</tt> linkage may not be discarded. This is used for globals that
604 are declared "weak" in C source code.</dd>
605
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]606 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]607 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
608 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
609 global scope.
610 Symbols with "<tt>common</tt>" linkage are merged in the same way as
611 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000[diff] [blame]612 <tt>common</tt> symbols may not have an explicit section,
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]613 must have a zero initializer, and may not be marked '<a
Chris Lattnercd81f5d2009-08-05 05:41:44 +0000[diff] [blame]614 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
615 have common linkage.</dd>
Chris Lattner26d054d2009-08-05 05:21:07 +0000[diff] [blame]616
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]617
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]618 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]619 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]620 pointer to array type. When two global variables with appending linkage
621 are linked together, the two global arrays are appended together. This is
622 the LLVM, typesafe, equivalent of having the system linker append together
623 "sections" with identical names when .o files are linked.</dd>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]624
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]625 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]626 <dd>The semantics of this linkage follow the ELF object file model: the symbol
627 is weak until linked, if not linked, the symbol becomes null instead of
628 being an undefined reference.</dd>
Anton Korobeynikov7f705592007-01-12 19:20:47 +0000[diff] [blame]629
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]630 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
631 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]632 <dd>Some languages allow differing globals to be merged, such as two functions
633 with different semantics. Other languages, such as <tt>C++</tt>, ensure
Bill Wendling5e721d72010-07-01 21:55:59 +0000[diff] [blame]634 that only equivalent globals are ever merged (the "one definition rule"
635 &mdash; "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]636 and <tt>weak_odr</tt> linkage types to indicate that the global will only
637 be merged with equivalent globals. These linkage types are otherwise the
638 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]639
Bill Wendling5c3a9f72011-11-04 20:40:41 +0000[diff] [blame]640 <dt><tt><b><a name="linkage_external">external</a></b></tt></dt>
Chris Lattnere5d947b2004-12-09 16:36:40 +0000[diff] [blame]641 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]642 visible, meaning that it participates in linkage and can be used to
643 resolve external symbol references.</dd>
Reid Spencerc8910842007-04-11 23:49:50 +0000[diff] [blame]644</dl>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]645
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]646<p>The next two types of linkage are targeted for Microsoft Windows platform
647 only. They are designed to support importing (exporting) symbols from (to)
648 DLLs (Dynamic Link Libraries).</p>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]649
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]650<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]651 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]652 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]653 or variable via a global pointer to a pointer that is set up by the DLL
654 exporting the symbol. On Microsoft Windows targets, the pointer name is
655 formed by combining <code>__imp_</code> and the function or variable
656 name.</dd>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]657
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]658 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Anton Korobeynikovb74ed072006-09-14 18:23:27 +0000[diff] [blame]659 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]660 pointer to a pointer in a DLL, so that it can be referenced with the
661 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
662 name is formed by combining <code>__imp_</code> and the function or
663 variable name.</dd>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]664</dl>
665
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]666<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
667 another module defined a "<tt>.LC0</tt>" variable and was linked with this
668 one, one of the two would be renamed, preventing a collision. Since
669 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
670 declarations), they are accessible outside of the current module.</p>
671
672<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]673 other than <tt>external</tt>, <tt>dllimport</tt>
674 or <tt>extern_weak</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]675
Duncan Sands667d4b82009-03-07 15:45:40 +0000[diff] [blame]676<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]677 or <tt>weak_odr</tt> linkages.</p>
678
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]679</div>
680
681<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]682<h3>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]683 <a name="callingconv">Calling Conventions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]684</h3>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]685
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]686<div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]687
688<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]689 and <a href="#i_invoke">invokes</a> can all have an optional calling
690 convention specified for the call. The calling convention of any pair of
691 dynamic caller/callee must match, or the behavior of the program is
692 undefined. The following calling conventions are supported by LLVM, and more
693 may be added in the future:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]694
695<dl>
696 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]697 <dd>This calling convention (the default if no other calling convention is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]698 specified) matches the target C calling conventions. This calling
699 convention supports varargs function calls and tolerates some mismatch in
700 the declared prototype and implemented declaration of the function (as
701 does normal C).</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]702
703 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]704 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]705 (e.g. by passing things in registers). This calling convention allows the
706 target to use whatever tricks it wants to produce fast code for the
707 target, without having to conform to an externally specified ABI
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]708 (Application Binary Interface).
709 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattner29689432010-03-11 00:22:57 +0000[diff] [blame]710 when this or the GHC convention is used.</a> This calling convention
711 does not support varargs and requires the prototype of all callees to
712 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]713
714 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]715 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]716 as possible under the assumption that the call is not commonly executed.
717 As such, these calls often preserve all registers so that the call does
718 not break any live ranges in the caller side. This calling convention
719 does not support varargs and requires the prototype of all callees to
720 exactly match the prototype of the function definition.</dd>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]721
Chris Lattner29689432010-03-11 00:22:57 +0000[diff] [blame]722 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
723 <dd>This calling convention has been implemented specifically for use by the
724 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
725 It passes everything in registers, going to extremes to achieve this by
726 disabling callee save registers. This calling convention should not be
727 used lightly but only for specific situations such as an alternative to
728 the <em>register pinning</em> performance technique often used when
729 implementing functional programming languages.At the moment only X86
730 supports this convention and it has the following limitations:
731 <ul>
732 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
733 floating point types are supported.</li>
734 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
735 6 floating point parameters.</li>
736 </ul>
737 This calling convention supports
738 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
739 requires both the caller and callee are using it.
740 </dd>
741
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]742 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]743 <dd>Any calling convention may be specified by number, allowing
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]744 target-specific calling conventions to be used. Target specific calling
745 conventions start at 64.</dd>
Chris Lattnercfe6b372005-05-07 01:46:40 +0000[diff] [blame]746</dl>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]747
748<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]749 support Pascal conventions or any other well-known target-independent
750 convention.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]751
752</div>
753
754<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]755<h3>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]756 <a name="visibility">Visibility Styles</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]757</h3>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]758
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]759<div>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]760
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]761<p>All Global Variables and Functions have one of the following visibility
762 styles:</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]763
764<dl>
765 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]766 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]767 that the declaration is visible to other modules and, in shared libraries,
768 means that the declared entity may be overridden. On Darwin, default
769 visibility means that the declaration is visible to other modules. Default
770 visibility corresponds to "external linkage" in the language.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]771
772 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]773 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]774 object if they are in the same shared object. Usually, hidden visibility
775 indicates that the symbol will not be placed into the dynamic symbol
776 table, so no other module (executable or shared library) can reference it
777 directly.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]778
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]779 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Anton Korobeynikov6f9896f2007-04-29 18:35:00 +0000[diff] [blame]780 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]781 the dynamic symbol table, but that references within the defining module
782 will bind to the local symbol. That is, the symbol cannot be overridden by
783 another module.</dd>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]784</dl>
785
786</div>
787
788<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]789<h3>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]790 <a name="namedtypes">Named Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]791</h3>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]792
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]793<div>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]794
795<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]796 it easier to read the IR and make the IR more condensed (particularly when
797 recursive types are involved). An example of a name specification is:</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]798
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]799<pre class="doc_code">
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]800%mytype = type { %mytype*, i32 }
801</pre>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]802
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]803<p>You may give a name to any <a href="#typesystem">type</a> except
Chris Lattnerdc65f222010-08-17 23:26:04 +0000[diff] [blame]804 "<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]805 is expected with the syntax "%mytype".</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]806
807<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]808 and that you can therefore specify multiple names for the same type. This
809 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
810 uses structural typing, the name is not part of the type. When printing out
811 LLVM IR, the printer will pick <em>one name</em> to render all types of a
812 particular shape. This means that if you have code where two different
813 source types end up having the same LLVM type, that the dumper will sometimes
814 print the "wrong" or unexpected type. This is an important design point and
815 isn't going to change.</p>
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]816
817</div>
818
Chris Lattnere7886e42009-01-11 20:53:49 +0000[diff] [blame]819<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]820<h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]821 <a name="globalvars">Global Variables</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]822</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]823
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]824<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]825
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]826<p>Global variables define regions of memory allocated at compilation time
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]827 instead of run-time. Global variables may optionally be initialized, may
828 have an explicit section to be placed in, and may have an optional explicit
829 alignment specified. A variable may be defined as "thread_local", which
830 means that it will not be shared by threads (each thread will have a
831 separated copy of the variable). A variable may be defined as a global
832 "constant," which indicates that the contents of the variable
833 will <b>never</b> be modified (enabling better optimization, allowing the
834 global data to be placed in the read-only section of an executable, etc).
835 Note that variables that need runtime initialization cannot be marked
836 "constant" as there is a store to the variable.</p>
Chris Lattner3689a342005-02-12 19:30:21 +0000[diff] [blame]837
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]838<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
839 constant, even if the final definition of the global is not. This capability
840 can be used to enable slightly better optimization of the program, but
841 requires the language definition to guarantee that optimizations based on the
842 'constantness' are valid for the translation units that do not include the
843 definition.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]844
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]845<p>As SSA values, global variables define pointer values that are in scope
846 (i.e. they dominate) all basic blocks in the program. Global variables
847 always define a pointer to their "content" type because they describe a
848 region of memory, and all memory objects in LLVM are accessed through
849 pointers.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]850
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]851<p>Global variables can be marked with <tt>unnamed_addr</tt> which indicates
852 that the address is not significant, only the content. Constants marked
Rafael Espindolaa5eaa862011-01-15 08:20:57 +0000[diff] [blame]853 like this can be merged with other constants if they have the same
854 initializer. Note that a constant with significant address <em>can</em>
855 be merged with a <tt>unnamed_addr</tt> constant, the result being a
856 constant whose address is significant.</p>
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]857
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]858<p>A global variable may be declared to reside in a target-specific numbered
859 address space. For targets that support them, address spaces may affect how
860 optimizations are performed and/or what target instructions are used to
861 access the variable. The default address space is zero. The address space
862 qualifier must precede any other attributes.</p>
Christopher Lamb284d9922007-12-11 09:31:00 +0000[diff] [blame]863
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]864<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]865 supports it, it will emit globals to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]866
Chris Lattnerce99fa92010-04-28 00:13:42 +0000[diff] [blame]867<p>An explicit alignment may be specified for a global, which must be a power
868 of 2. If not present, or if the alignment is set to zero, the alignment of
869 the global is set by the target to whatever it feels convenient. If an
870 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner2d4b8ee2010-04-28 00:31:12 +0000[diff] [blame]871 alignment. Targets and optimizers are not allowed to over-align the global
872 if the global has an assigned section. In this case, the extra alignment
873 could be observable: for example, code could assume that the globals are
874 densely packed in their section and try to iterate over them as an array,
875 alignment padding would break this iteration.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]876
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]877<p>For example, the following defines a global in a numbered address space with
878 an initializer, section, and alignment:</p>
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]879
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]880<pre class="doc_code">
Dan Gohman398873c2009-01-11 00:40:00 +0000[diff] [blame]881@G = addrspace(5) constant float 1.0, section "foo", align 4
Chris Lattner68027ea2007-01-14 00:27:09 +0000[diff] [blame]882</pre>
883
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]884</div>
885
886
887<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]888<h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]889 <a name="functionstructure">Functions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]890</h3>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]891
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]892<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]893
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]894<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]895 optional <a href="#linkage">linkage type</a>, an optional
896 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]897 <a href="#callingconv">calling convention</a>,
898 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]899 <a href="#paramattrs">parameter attribute</a> for the return type, a function
900 name, a (possibly empty) argument list (each with optional
901 <a href="#paramattrs">parameter attributes</a>), optional
902 <a href="#fnattrs">function attributes</a>, an optional section, an optional
903 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
904 curly brace, a list of basic blocks, and a closing curly brace.</p>
Anton Korobeynikov8cea37b2007-01-23 12:35:46 +0000[diff] [blame]905
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]906<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
907 optional <a href="#linkage">linkage type</a>, an optional
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]908 <a href="#visibility">visibility style</a>, an optional
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]909 <a href="#callingconv">calling convention</a>,
910 an optional <tt>unnamed_addr</tt> attribute, a return type, an optional
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]911 <a href="#paramattrs">parameter attribute</a> for the return type, a function
912 name, a possibly empty list of arguments, an optional alignment, and an
913 optional <a href="#gc">garbage collector name</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]914
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]915<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]916 (Control Flow Graph) for the function. Each basic block may optionally start
917 with a label (giving the basic block a symbol table entry), contains a list
918 of instructions, and ends with a <a href="#terminators">terminator</a>
919 instruction (such as a branch or function return).</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]920
Chris Lattner4a3c9012007-06-08 16:52:14 +0000[diff] [blame]921<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]922 executed on entrance to the function, and it is not allowed to have
923 predecessor basic blocks (i.e. there can not be any branches to the entry
924 block of a function). Because the block can have no predecessors, it also
925 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]926
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]927<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]928 supports it, it will emit functions to the section specified.</p>
Chris Lattner88f6c462005-11-12 00:45:07 +0000[diff] [blame]929
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]930<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]931 the alignment is set to zero, the alignment of the function is set by the
932 target to whatever it feels convenient. If an explicit alignment is
933 specified, the function is forced to have at least that much alignment. All
934 alignments must be a power of 2.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]935
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]936<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]937 be significant and two identical functions can be merged.</p>
Rafael Espindolabea46262011-01-08 16:42:36 +0000[diff] [blame]938
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]939<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]940<pre class="doc_code">
Chris Lattner50ad45c2008-10-13 16:55:18 +0000[diff] [blame]941define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]942 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
943 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
944 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
945 [<a href="#gc">gc</a>] { ... }
946</pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]947
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]948</div>
949
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]950<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]951<h3>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]952 <a name="aliasstructure">Aliases</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]953</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]954
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]955<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]956
957<p>Aliases act as "second name" for the aliasee value (which can be either
958 function, global variable, another alias or bitcast of global value). Aliases
959 may have an optional <a href="#linkage">linkage type</a>, and an
960 optional <a href="#visibility">visibility style</a>.</p>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]961
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]962<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]963<pre class="doc_code">
Duncan Sands0b23ac12008-09-12 20:48:21 +0000[diff] [blame]964@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]965</pre>
Anton Korobeynikov8b0a8c82007-04-25 14:27:10 +0000[diff] [blame]966
967</div>
968
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]969<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]970<h3>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]971 <a name="namedmetadatastructure">Named Metadata</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]972</h3>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]973
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]974<div>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]975
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]976<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]977 nodes</a> (but not metadata strings) are the only valid operands for
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]978 a named metadata.</p>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]979
980<h5>Syntax:</h5>
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]981<pre class="doc_code">
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]982; Some unnamed metadata nodes, which are referenced by the named metadata.
983!0 = metadata !{metadata !"zero"}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]984!1 = metadata !{metadata !"one"}
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]985!2 = metadata !{metadata !"two"}
Dan Gohman1005bc52010-07-13 19:48:13 +0000[diff] [blame]986; A named metadata.
Dan Gohman872814a2010-07-21 18:54:18 +0000[diff] [blame]987!name = !{!0, !1, !2}
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]988</pre>
Devang Patelcd1fd252010-01-11 19:35:55 +0000[diff] [blame]989
990</div>
991
992<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]993<h3>
994 <a name="paramattrs">Parameter Attributes</a>
995</h3>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]996
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]997<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]998
999<p>The return type and each parameter of a function type may have a set of
1000 <i>parameter attributes</i> associated with them. Parameter attributes are
1001 used to communicate additional information about the result or parameters of
1002 a function. Parameter attributes are considered to be part of the function,
1003 not of the function type, so functions with different parameter attributes
1004 can have the same function type.</p>
1005
1006<p>Parameter attributes are simple keywords that follow the type specified. If
1007 multiple parameter attributes are needed, they are space separated. For
1008 example:</p>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1009
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1010<pre class="doc_code">
Nick Lewyckyb6a7d252009-02-15 23:06:14 +0000[diff] [blame]1011declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattner66d922c2008-10-04 18:33:34 +0000[diff] [blame]1012declare i32 @atoi(i8 zeroext)
1013declare signext i8 @returns_signed_char()
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1014</pre>
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1015
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1016<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1017 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1018
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1019<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1020
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1021<dl>
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1022 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1023 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarichebe81732011-03-16 22:20:18 +0000[diff] [blame]1024 should be zero-extended to the extent required by the target's ABI (which
1025 is usually 32-bits, but is 8-bits for a i1 on x86-64) by the caller (for a
1026 parameter) or the callee (for a return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1027
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1028 <dt><tt><b>signext</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1029 <dd>This indicates to the code generator that the parameter or return value
Cameron Zwarich9e69ff92011-03-17 14:21:58 +0000[diff] [blame]1030 should be sign-extended to the extent required by the target's ABI (which
1031 is usually 32-bits) by the caller (for a parameter) or the callee (for a
1032 return value).</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1033
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1034 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1035 <dd>This indicates that this parameter or return value should be treated in a
1036 special target-dependent fashion during while emitting code for a function
1037 call or return (usually, by putting it in a register as opposed to memory,
1038 though some targets use it to distinguish between two different kinds of
1039 registers). Use of this attribute is target-specific.</dd>
Chris Lattner47507de2008-01-11 06:20:47 +0000[diff] [blame]1040
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1041 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Chris Lattnera6fd81d2010-11-20 23:49:06 +0000[diff] [blame]1042 <dd><p>This indicates that the pointer parameter should really be passed by
1043 value to the function. The attribute implies that a hidden copy of the
1044 pointee
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1045 is made between the caller and the callee, so the callee is unable to
1046 modify the value in the callee. This attribute is only valid on LLVM
1047 pointer arguments. It is generally used to pass structs and arrays by
1048 value, but is also valid on pointers to scalars. The copy is considered
1049 to belong to the caller not the callee (for example,
1050 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1051 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnera6fd81d2010-11-20 23:49:06 +0000[diff] [blame]1052 values.</p>
1053
1054 <p>The byval attribute also supports specifying an alignment with
1055 the align attribute. It indicates the alignment of the stack slot to
1056 form and the known alignment of the pointer specified to the call site. If
1057 the alignment is not specified, then the code generator makes a
1058 target-specific assumption.</p></dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1059
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1060 <dt><tt><b><a name="sret">sret</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1061 <dd>This indicates that the pointer parameter specifies the address of a
1062 structure that is the return value of the function in the source program.
1063 This pointer must be guaranteed by the caller to be valid: loads and
1064 stores to the structure may be assumed by the callee to not to trap. This
1065 may only be applied to the first parameter. This is not a valid attribute
1066 for return values. </dd>
1067
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1068 <dt><tt><b><a name="noalias">noalias</a></b></tt></dt>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1069 <dd>This indicates that pointer values
1070 <a href="#pointeraliasing"><i>based</i></a> on the argument or return
Dan Gohmanefca7f92010-07-02 23:46:54 +0000[diff] [blame]1071 value do not alias pointer values which are not <i>based</i> on it,
1072 ignoring certain "irrelevant" dependencies.
1073 For a call to the parent function, dependencies between memory
1074 references from before or after the call and from those during the call
1075 are "irrelevant" to the <tt>noalias</tt> keyword for the arguments and
1076 return value used in that call.
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1077 The caller shares the responsibility with the callee for ensuring that
1078 these requirements are met.
1079 For further details, please see the discussion of the NoAlias response in
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]1080 <a href="AliasAnalysis.html#MustMayNo">alias analysis</a>.<br>
1081<br>
John McCall191d4ee2010-07-06 21:07:14 +0000[diff] [blame]1082 Note that this definition of <tt>noalias</tt> is intentionally
1083 similar to the definition of <tt>restrict</tt> in C99 for function
Chris Lattner211244a2010-07-06 20:51:35 +0000[diff] [blame]1084 arguments, though it is slightly weaker.
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]1085<br>
1086 For function return values, C99's <tt>restrict</tt> is not meaningful,
1087 while LLVM's <tt>noalias</tt> is.
1088 </dd>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1089
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1090 <dt><tt><b><a name="nocapture">nocapture</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1091 <dd>This indicates that the callee does not make any copies of the pointer
1092 that outlive the callee itself. This is not a valid attribute for return
1093 values.</dd>
1094
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]1095 <dt><tt><b><a name="nest">nest</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1096 <dd>This indicates that the pointer parameter can be excised using the
1097 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1098 attribute for return values.</dd>
1099</dl>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1100
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1101</div>
1102
1103<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1104<h3>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1105 <a name="gc">Garbage Collector Names</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1106</h3>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1107
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1108<div>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1109
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1110<p>Each function may specify a garbage collector name, which is simply a
1111 string:</p>
1112
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1113<pre class="doc_code">
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1114define void @f() gc "name" { ... }
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1115</pre>
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1116
1117<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1118 collector which will cause the compiler to alter its output in order to
1119 support the named garbage collection algorithm.</p>
1120
Gordon Henriksen80a75bf2007-12-10 03:18:06 +0000[diff] [blame]1121</div>
1122
1123<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1124<h3>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1125 <a name="fnattrs">Function Attributes</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1126</h3>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1127
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1128<div>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1129
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1130<p>Function attributes are set to communicate additional information about a
1131 function. Function attributes are considered to be part of the function, not
1132 of the function type, so functions with different parameter attributes can
1133 have the same function type.</p>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1134
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1135<p>Function attributes are simple keywords that follow the type specified. If
1136 multiple attributes are needed, they are space separated. For example:</p>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1137
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1138<pre class="doc_code">
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1139define void @f() noinline { ... }
1140define void @f() alwaysinline { ... }
1141define void @f() alwaysinline optsize { ... }
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1142define void @f() optsize { ... }
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1143</pre>
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1144
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1145<dl>
Charles Davis1e063d12010-02-12 00:31:15 +0000[diff] [blame]1146 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1147 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1148 the backend should forcibly align the stack pointer. Specify the
1149 desired alignment, which must be a power of two, in parentheses.
1150
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1151 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1152 <dd>This attribute indicates that the inliner should attempt to inline this
1153 function into callers whenever possible, ignoring any active inlining size
1154 threshold for this caller.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1155
Dan Gohman129bd562011-06-16 16:03:13 +0000[diff] [blame]1156 <dt><tt><b>nonlazybind</b></tt></dt>
1157 <dd>This attribute suppresses lazy symbol binding for the function. This
1158 may make calls to the function faster, at the cost of extra program
1159 startup time if the function is not called during program startup.</dd>
1160
Jakob Stoklund Olesen570a4a52010-02-06 01:16:28 +0000[diff] [blame]1161 <dt><tt><b>inlinehint</b></tt></dt>
1162 <dd>This attribute indicates that the source code contained a hint that inlining
1163 this function is desirable (such as the "inline" keyword in C/C++). It
1164 is just a hint; it imposes no requirements on the inliner.</dd>
1165
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1166 <dt><tt><b>naked</b></tt></dt>
1167 <dd>This attribute disables prologue / epilogue emission for the function.
1168 This can have very system-specific consequences.</dd>
1169
1170 <dt><tt><b>noimplicitfloat</b></tt></dt>
1171 <dd>This attributes disables implicit floating point instructions.</dd>
1172
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1173 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1174 <dd>This attribute indicates that the inliner should never inline this
1175 function in any situation. This attribute may not be used together with
1176 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1177
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1178 <dt><tt><b>noredzone</b></tt></dt>
1179 <dd>This attribute indicates that the code generator should not use a red
1180 zone, even if the target-specific ABI normally permits it.</dd>
Devang Patel2c9c3e72008-09-26 23:51:19 +0000[diff] [blame]1181
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1182 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1183 <dd>This function attribute indicates that the function never returns
1184 normally. This produces undefined behavior at runtime if the function
1185 ever does dynamically return.</dd>
Bill Wendling31359ba2008-11-13 01:02:51 +0000[diff] [blame]1186
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1187 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1188 <dd>This function attribute indicates that the function never returns with an
1189 unwind or exceptional control flow. If the function does unwind, its
1190 runtime behavior is undefined.</dd>
Bill Wendlingfbaa7ed2008-11-26 19:07:40 +0000[diff] [blame]1191
Nick Lewycky76ec37a2010-07-06 18:24:09 +0000[diff] [blame]1192 <dt><tt><b>optsize</b></tt></dt>
1193 <dd>This attribute suggests that optimization passes and code generator passes
1194 make choices that keep the code size of this function low, and otherwise
1195 do optimizations specifically to reduce code size.</dd>
1196
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1197 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1198 <dd>This attribute indicates that the function computes its result (or decides
1199 to unwind an exception) based strictly on its arguments, without
1200 dereferencing any pointer arguments or otherwise accessing any mutable
1201 state (e.g. memory, control registers, etc) visible to caller functions.
1202 It does not write through any pointer arguments
1203 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1204 changes any state visible to callers. This means that it cannot unwind
1205 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1206 could use the <tt>unwind</tt> instruction.</dd>
Devang Patel5d96fda2009-06-12 19:45:19 +0000[diff] [blame]1207
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1208 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1209 <dd>This attribute indicates that the function does not write through any
1210 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1211 arguments) or otherwise modify any state (e.g. memory, control registers,
1212 etc) visible to caller functions. It may dereference pointer arguments
1213 and read state that may be set in the caller. A readonly function always
1214 returns the same value (or unwinds an exception identically) when called
1215 with the same set of arguments and global state. It cannot unwind an
1216 exception by calling the <tt>C++</tt> exception throwing methods, but may
1217 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovc5ec8a72009-07-17 18:07:26 +0000[diff] [blame]1218
Bill Wendling9bd5d042011-12-05 21:27:54 +0000[diff] [blame]1219 <dt><tt><b><a name="returns_twice">returns_twice</a></b></tt></dt>
1220 <dd>This attribute indicates that this function can return twice. The
1221 C <code>setjmp</code> is an example of such a function. The compiler
1222 disables some optimizations (like tail calls) in the caller of these
1223 functions.</dd>
1224
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1225 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1226 <dd>This attribute indicates that the function should emit a stack smashing
1227 protector. It is in the form of a "canary"&mdash;a random value placed on
1228 the stack before the local variables that's checked upon return from the
1229 function to see if it has been overwritten. A heuristic is used to
1230 determine if a function needs stack protectors or not.<br>
1231<br>
1232 If a function that has an <tt>ssp</tt> attribute is inlined into a
1233 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1234 function will have an <tt>ssp</tt> attribute.</dd>
1235
Bill Wendlingf82d40a2009-11-02 00:24:16 +0000[diff] [blame]1236 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1237 <dd>This attribute indicates that the function should <em>always</em> emit a
1238 stack smashing protector. This overrides
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1239 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1240<br>
1241 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1242 function that doesn't have an <tt>sspreq</tt> attribute or which has
1243 an <tt>ssp</tt> attribute, then the resulting function will have
1244 an <tt>sspreq</tt> attribute.</dd>
Rafael Espindolafbff0ec2011-07-25 15:27:59 +0000[diff] [blame]1245
1246 <dt><tt><b><a name="uwtable">uwtable</a></b></tt></dt>
1247 <dd>This attribute indicates that the ABI being targeted requires that
1248 an unwind table entry be produce for this function even if we can
1249 show that no exceptions passes by it. This is normally the case for
1250 the ELF x86-64 abi, but it can be disabled for some compilation
1251 units.</dd>
Bill Wendlinge36dccc2008-09-07 10:26:33 +0000[diff] [blame]1252</dl>
1253
Devang Patelf8b94812008-09-04 23:05:13 +0000[diff] [blame]1254</div>
1255
1256<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1257<h3>
Chris Lattner1eeeb0c2006-04-08 04:40:53 +0000[diff] [blame]1258 <a name="moduleasm">Module-Level Inline Assembly</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1259</h3>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1260
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1261<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1262
1263<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1264 the GCC "file scope inline asm" blocks. These blocks are internally
1265 concatenated by LLVM and treated as a single unit, but may be separated in
1266 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1267
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1268<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]1269module asm "inline asm code goes here"
1270module asm "more can go here"
1271</pre>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1272
1273<p>The strings can contain any character by escaping non-printable characters.
1274 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]1275 for the number.</p>
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1276
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1277<p>The inline asm code is simply printed to the machine code .s file when
1278 assembly code is generated.</p>
1279
Chris Lattner4e9aba72006-01-23 23:23:47 +0000[diff] [blame]1280</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1281
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1282<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1283<h3>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1284 <a name="datalayout">Data Layout</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1285</h3>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1286
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1287<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1288
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1289<p>A module may specify a target specific data layout string that specifies how
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1290 data is to be laid out in memory. The syntax for the data layout is
1291 simply:</p>
1292
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]1293<pre class="doc_code">
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1294target datalayout = "<i>layout specification</i>"
1295</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1296
1297<p>The <i>layout specification</i> consists of a list of specifications
1298 separated by the minus sign character ('-'). Each specification starts with
1299 a letter and may include other information after the letter to define some
1300 aspect of the data layout. The specifications accepted are as follows:</p>
1301
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1302<dl>
1303 <dt><tt>E</tt></dt>
1304 <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]1305 bits with the most significance have the lowest address location.</dd>
1306
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1307 <dt><tt>e</tt></dt>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]1308 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1309 the bits with the least significance have the lowest address
1310 location.</dd>
1311
Lang Hamesbb5b3f32011-10-10 23:42:08 +0000[diff] [blame]1312 <dt><tt>S<i>size</i></tt></dt>
1313 <dd>Specifies the natural alignment of the stack in bits. Alignment promotion
1314 of stack variables is limited to the natural stack alignment to avoid
1315 dynamic stack realignment. The stack alignment must be a multiple of
Lang Hames5f119a62011-10-11 17:50:14 +0000[diff] [blame]1316 8-bits. If omitted, the natural stack alignment defaults to "unspecified",
1317 which does not prevent any alignment promotions.</dd>
Lang Hamesbb5b3f32011-10-10 23:42:08 +0000[diff] [blame]1318
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1319 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1320 <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]1321 <i>preferred</i> alignments. All sizes are in bits. Specifying
1322 the <i>pref</i> alignment is optional. If omitted, the
1323 preceding <tt>:</tt> should be omitted too.</dd>
1324
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1325 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1326 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1327 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1328
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1329 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1330 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1331 <i>size</i>.</dd>
1332
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1333 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]1334 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesen9d8d2212010-05-28 18:54:47 +0000[diff] [blame]1335 <i>size</i>. Only values of <i>size</i> that are supported by the target
1336 will work. 32 (float) and 64 (double) are supported on all targets;
1337 80 or 128 (different flavors of long double) are also supported on some
1338 targets.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1339
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1340 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1341 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1342 <i>size</i>.</dd>
1343
Daniel Dunbar87bde0b2009-06-08 22:17:53 +0000[diff] [blame]1344 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1345 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1346 <i>size</i>.</dd>
Chris Lattnere82bdc42009-11-07 09:35:34 +0000[diff] [blame]1347
1348 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1349 <dd>This specifies a set of native integer widths for the target CPU
1350 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1351 "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]1352 this set are considered to support most general arithmetic
Chris Lattnere82bdc42009-11-07 09:35:34 +0000[diff] [blame]1353 operations efficiently.</dd>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1354</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1355
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1356<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohman1c70c002010-04-28 00:36:01 +0000[diff] [blame]1357 default set of specifications which are then (possibly) overridden by the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1358 specifications in the <tt>datalayout</tt> keyword. The default specifications
1359 are given in this list:</p>
1360
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1361<ul>
1362 <li><tt>E</tt> - big endian</li>
Dan Gohmanfdf2e8c2010-02-23 02:44:03 +0000[diff] [blame]1363 <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]1364 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1365 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1366 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1367 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]1368 <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]1369 alignment of 64-bits</li>
1370 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1371 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1372 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1373 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1374 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbar87bde0b2009-06-08 22:17:53 +0000[diff] [blame]1375 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1376</ul>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1377
1378<p>When LLVM is determining the alignment for a given type, it uses the
1379 following rules:</p>
1380
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1381<ol>
1382 <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]1383 specification is used.</li>
1384
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1385 <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]1386 smallest integer type that is larger than the bitwidth of the sought type
1387 is used. If none of the specifications are larger than the bitwidth then
1388 the the largest integer type is used. For example, given the default
1389 specifications above, the i7 type will use the alignment of i8 (next
1390 largest) while both i65 and i256 will use the alignment of i64 (largest
1391 specified).</li>
1392
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1393 <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]1394 largest vector type that is smaller than the sought vector type will be
1395 used as a fall back. This happens because &lt;128 x double&gt; can be
1396 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1397</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1398
Chris Lattner6509f502011-10-11 23:01:39 +0000[diff] [blame]1399<p>The function of the data layout string may not be what you expect. Notably,
1400 this is not a specification from the frontend of what alignment the code
1401 generator should use.</p>
1402
1403<p>Instead, if specified, the target data layout is required to match what the
1404 ultimate <em>code generator</em> expects. This string is used by the
1405 mid-level optimizers to
1406 improve code, and this only works if it matches what the ultimate code
1407 generator uses. If you would like to generate IR that does not embed this
1408 target-specific detail into the IR, then you don't have to specify the
1409 string. This will disable some optimizations that require precise layout
1410 information, but this also prevents those optimizations from introducing
1411 target specificity into the IR.</p>
1412
1413
1414
Reid Spencerde151942007-02-19 23:54:10 +0000[diff] [blame]1415</div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1416
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1417<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1418<h3>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1419 <a name="pointeraliasing">Pointer Aliasing Rules</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1420</h3>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1421
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1422<div>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1423
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1424<p>Any memory access must be done through a pointer value associated
Andreas Bolka99a82052009-07-27 20:37:10 +0000[diff] [blame]1425with an address range of the memory access, otherwise the behavior
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1426is undefined. Pointer values are associated with address ranges
1427according to the following rules:</p>
1428
1429<ul>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1430 <li>A pointer value is associated with the addresses associated with
1431 any value it is <i>based</i> on.
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1432 <li>An address of a global variable is associated with the address
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1433 range of the variable's storage.</li>
1434 <li>The result value of an allocation instruction is associated with
1435 the address range of the allocated storage.</li>
1436 <li>A null pointer in the default address-space is associated with
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1437 no address.</li>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1438 <li>An integer constant other than zero or a pointer value returned
1439 from a function not defined within LLVM may be associated with address
1440 ranges allocated through mechanisms other than those provided by
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1441 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1442 allocated by mechanisms provided by LLVM.</li>
Dan Gohman1e109622010-07-02 18:41:32 +0000[diff] [blame]1443</ul>
1444
1445<p>A pointer value is <i>based</i> on another pointer value according
1446 to the following rules:</p>
1447
1448<ul>
1449 <li>A pointer value formed from a
1450 <tt><a href="#i_getelementptr">getelementptr</a></tt> operation
1451 is <i>based</i> on the first operand of the <tt>getelementptr</tt>.</li>
1452 <li>The result value of a
1453 <tt><a href="#i_bitcast">bitcast</a></tt> is <i>based</i> on the operand
1454 of the <tt>bitcast</tt>.</li>
1455 <li>A pointer value formed by an
1456 <tt><a href="#i_inttoptr">inttoptr</a></tt> is <i>based</i> on all
1457 pointer values that contribute (directly or indirectly) to the
1458 computation of the pointer's value.</li>
1459 <li>The "<i>based</i> on" relationship is transitive.</li>
1460</ul>
1461
1462<p>Note that this definition of <i>"based"</i> is intentionally
1463 similar to the definition of <i>"based"</i> in C99, though it is
1464 slightly weaker.</p>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1465
1466<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1467<tt><a href="#i_load">load</a></tt> merely indicates the size and
1468alignment of the memory from which to load, as well as the
Dan Gohmanc22c0f32010-06-17 19:23:50 +0000[diff] [blame]1469interpretation of the value. The first operand type of a
Andreas Bolka55e459a2009-07-29 00:02:05 +0000[diff] [blame]1470<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1471and alignment of the store.</p>
Dan Gohman556ca272009-07-27 18:07:55 +0000[diff] [blame]1472
1473<p>Consequently, type-based alias analysis, aka TBAA, aka
1474<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1475LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1476additional information which specialized optimization passes may use
1477to implement type-based alias analysis.</p>
1478
1479</div>
1480
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1481<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1482<h3>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1483 <a name="volatile">Volatile Memory Accesses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1484</h3>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1485
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1486<div>
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]1487
1488<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1489href="#i_store"><tt>store</tt></a>s, and <a
1490href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1491The optimizers must not change the number of volatile operations or change their
1492order of execution relative to other volatile operations. The optimizers
1493<i>may</i> change the order of volatile operations relative to non-volatile
1494operations. This is not Java's "volatile" and has no cross-thread
1495synchronization behavior.</p>
1496
1497</div>
1498
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1499<!-- ======================================================================= -->
1500<h3>
1501 <a name="memmodel">Memory Model for Concurrent Operations</a>
1502</h3>
1503
1504<div>
1505
1506<p>The LLVM IR does not define any way to start parallel threads of execution
1507or to register signal handlers. Nonetheless, there are platform-specific
1508ways to create them, and we define LLVM IR's behavior in their presence. This
1509model is inspired by the C++0x memory model.</p>
1510
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1511<p>For a more informal introduction to this model, see the
1512<a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.
1513
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1514<p>We define a <i>happens-before</i> partial order as the least partial order
1515that</p>
1516<ul>
1517 <li>Is a superset of single-thread program order, and</li>
1518 <li>When a <i>synchronizes-with</i> <tt>b</tt>, includes an edge from
1519 <tt>a</tt> to <tt>b</tt>. <i>Synchronizes-with</i> pairs are introduced
1520 by platform-specific techniques, like pthread locks, thread
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1521 creation, thread joining, etc., and by atomic instructions.
1522 (See also <a href="#ordering">Atomic Memory Ordering Constraints</a>).
1523 </li>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1524</ul>
1525
1526<p>Note that program order does not introduce <i>happens-before</i> edges
1527between a thread and signals executing inside that thread.</p>
1528
1529<p>Every (defined) read operation (load instructions, memcpy, atomic
1530loads/read-modify-writes, etc.) <var>R</var> reads a series of bytes written by
1531(defined) write operations (store instructions, atomic
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1532stores/read-modify-writes, memcpy, etc.). For the purposes of this section,
1533initialized globals are considered to have a write of the initializer which is
1534atomic and happens before any other read or write of the memory in question.
1535For each byte of a read <var>R</var>, <var>R<sub>byte</sub></var> may see
1536any write to the same byte, except:</p>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1537
1538<ul>
1539 <li>If <var>write<sub>1</sub></var> happens before
1540 <var>write<sub>2</sub></var>, and <var>write<sub>2</sub></var> happens
1541 before <var>R<sub>byte</sub></var>, then <var>R<sub>byte</sub></var>
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1542 does not see <var>write<sub>1</sub></var>.
Bill Wendling0246bb72011-07-31 06:45:03 +0000[diff] [blame]1543 <li>If <var>R<sub>byte</sub></var> happens before
1544 <var>write<sub>3</sub></var>, then <var>R<sub>byte</sub></var> does not
1545 see <var>write<sub>3</sub></var>.
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1546</ul>
1547
1548<p>Given that definition, <var>R<sub>byte</sub></var> is defined as follows:
1549<ul>
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1550 <li>If <var>R</var> is volatile, the result is target-dependent. (Volatile
1551 is supposed to give guarantees which can support
1552 <code>sig_atomic_t</code> in C/C++, and may be used for accesses to
1553 addresses which do not behave like normal memory. It does not generally
1554 provide cross-thread synchronization.)
1555 <li>Otherwise, if there is no write to the same byte that happens before
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1556 <var>R<sub>byte</sub></var>, <var>R<sub>byte</sub></var> returns
1557 <tt>undef</tt> for that byte.
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1558 <li>Otherwise, if <var>R<sub>byte</sub></var> may see exactly one write,
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1559 <var>R<sub>byte</sub></var> returns the value written by that
1560 write.</li>
Eli Friedman118973a2011-07-22 03:04:45 +0000[diff] [blame]1561 <li>Otherwise, if <var>R</var> is atomic, and all the writes
1562 <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]1563 values written. See the <a href="#ordering">Atomic Memory Ordering
1564 Constraints</a> section for additional constraints on how the choice
1565 is made.
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1566 <li>Otherwise <var>R<sub>byte</sub></var> returns <tt>undef</tt>.</li>
1567</ul>
1568
1569<p><var>R</var> returns the value composed of the series of bytes it read.
1570This implies that some bytes within the value may be <tt>undef</tt>
1571<b>without</b> the entire value being <tt>undef</tt>. Note that this only
1572defines the semantics of the operation; it doesn't mean that targets will
1573emit more than one instruction to read the series of bytes.</p>
1574
1575<p>Note that in cases where none of the atomic intrinsics are used, this model
1576places only one restriction on IR transformations on top of what is required
1577for single-threaded execution: introducing a store to a byte which might not
Eli Friedman101c81d2011-08-02 01:15:34 +0000[diff] [blame]1578otherwise be stored is not allowed in general. (Specifically, in the case
1579where another thread might write to and read from an address, introducing a
1580store can change a load that may see exactly one write into a load that may
1581see multiple writes.)</p>
Eli Friedman5b60e1b2011-07-20 21:35:53 +0000[diff] [blame]1582
1583<!-- FIXME: This model assumes all targets where concurrency is relevant have
1584a byte-size store which doesn't affect adjacent bytes. As far as I can tell,
1585none of the backends currently in the tree fall into this category; however,
1586there might be targets which care. If there are, we want a paragraph
1587like the following:
1588
1589Targets may specify that stores narrower than a certain width are not
1590available; on such a target, for the purposes of this model, treat any
1591non-atomic write with an alignment or width less than the minimum width
1592as if it writes to the relevant surrounding bytes.
1593-->
1594
1595</div>
1596
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1597<!-- ======================================================================= -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]1598<h3>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1599 <a name="ordering">Atomic Memory Ordering Constraints</a>
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]1600</h3>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1601
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]1602<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1603
1604<p>Atomic instructions (<a href="#i_cmpxchg"><code>cmpxchg</code></a>,
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]1605<a href="#i_atomicrmw"><code>atomicrmw</code></a>,
1606<a href="#i_fence"><code>fence</code></a>,
1607<a href="#i_load"><code>atomic load</code></a>, and
Eli Friedman8fa281a2011-08-09 23:26:12 +0000[diff] [blame]1608<a href="#i_store"><code>atomic store</code></a>) take an ordering parameter
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1609that determines which other atomic instructions on the same address they
1610<i>synchronize with</i>. These semantics are borrowed from Java and C++0x,
1611but are somewhat more colloquial. If these descriptions aren't precise enough,
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1612check those specs (see spec references in the
1613<a href="Atomic.html#introduction">atomics guide</a>).
1614<a href="#i_fence"><code>fence</code></a> instructions
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1615treat these orderings somewhat differently since they don't take an address.
1616See that instruction's documentation for details.</p>
1617
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1618<p>For a simpler introduction to the ordering constraints, see the
1619<a href="Atomics.html">LLVM Atomic Instructions and Concurrency Guide</a>.</p>
1620
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1621<dl>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1622<dt><code>unordered</code></dt>
1623<dd>The set of values that can be read is governed by the happens-before
1624partial order. A value cannot be read unless some operation wrote it.
1625This is intended to provide a guarantee strong enough to model Java's
1626non-volatile shared variables. This ordering cannot be specified for
1627read-modify-write operations; it is not strong enough to make them atomic
1628in any interesting way.</dd>
1629<dt><code>monotonic</code></dt>
1630<dd>In addition to the guarantees of <code>unordered</code>, there is a single
1631total order for modifications by <code>monotonic</code> operations on each
1632address. All modification orders must be compatible with the happens-before
1633order. There is no guarantee that the modification orders can be combined to
1634a global total order for the whole program (and this often will not be
1635possible). The read in an atomic read-modify-write operation
1636(<a href="#i_cmpxchg"><code>cmpxchg</code></a> and
1637<a href="#i_atomicrmw"><code>atomicrmw</code></a>)
1638reads the value in the modification order immediately before the value it
1639writes. If one atomic read happens before another atomic read of the same
1640address, the later read must see the same value or a later value in the
1641address's modification order. This disallows reordering of
1642<code>monotonic</code> (or stronger) operations on the same address. If an
1643address is written <code>monotonic</code>ally by one thread, and other threads
1644<code>monotonic</code>ally read that address repeatedly, the other threads must
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1645eventually see the write. This corresponds to the C++0x/C1x
1646<code>memory_order_relaxed</code>.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1647<dt><code>acquire</code></dt>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1648<dd>In addition to the guarantees of <code>monotonic</code>,
Eli Friedmanc264b2f2011-08-24 20:28:39 +0000[diff] [blame]1649a <i>synchronizes-with</i> edge may be formed with a <code>release</code>
1650operation. This is intended to model C++'s <code>memory_order_acquire</code>.</dd>
1651<dt><code>release</code></dt>
1652<dd>In addition to the guarantees of <code>monotonic</code>, if this operation
1653writes a value which is subsequently read by an <code>acquire</code> operation,
1654it <i>synchronizes-with</i> that operation. (This isn't a complete
1655description; see the C++0x definition of a release sequence.) This corresponds
1656to the C++0x/C1x <code>memory_order_release</code>.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1657<dt><code>acq_rel</code> (acquire+release)</dt><dd>Acts as both an
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1658<code>acquire</code> and <code>release</code> operation on its address.
1659This corresponds to the C++0x/C1x <code>memory_order_acq_rel</code>.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1660<dt><code>seq_cst</code> (sequentially consistent)</dt><dd>
1661<dd>In addition to the guarantees of <code>acq_rel</code>
1662(<code>acquire</code> for an operation which only reads, <code>release</code>
1663for an operation which only writes), there is a global total order on all
1664sequentially-consistent operations on all addresses, which is consistent with
1665the <i>happens-before</i> partial order and with the modification orders of
1666all the affected addresses. Each sequentially-consistent read sees the last
Eli Friedman234bccd2011-08-22 21:35:27 +0000[diff] [blame]1667preceding write to the same address in this global order. This corresponds
1668to the C++0x/C1x <code>memory_order_seq_cst</code> and Java volatile.</dd>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]1669</dl>
1670
1671<p id="singlethread">If an atomic operation is marked <code>singlethread</code>,
1672it only <i>synchronizes with</i> or participates in modification and seq_cst
1673total orderings with other operations running in the same thread (for example,
1674in signal handlers).</p>
1675
1676</div>
1677
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1678</div>
1679
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1680<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1681<h2><a name="typesystem">Type System</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1682<!-- *********************************************************************** -->
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1683
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1684<div>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1685
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1686<p>The LLVM type system is one of the most important features of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1687 intermediate representation. Being typed enables a number of optimizations
1688 to be performed on the intermediate representation directly, without having
1689 to do extra analyses on the side before the transformation. A strong type
1690 system makes it easier to read the generated code and enables novel analyses
1691 and transformations that are not feasible to perform on normal three address
1692 code representations.</p>
Chris Lattnerfa730212004-12-09 16:11:40 +0000[diff] [blame]1693
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1694<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1695<h3>
1696 <a name="t_classifications">Type Classifications</a>
1697</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1698
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1699<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1700
1701<p>The types fall into a few useful classifications:</p>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1702
1703<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1704 <tbody>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1705 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1706 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1707 <td><a href="#t_integer">integer</a></td>
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1708 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1709 </tr>
1710 <tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1711 <td><a href="#t_floating">floating point</a></td>
1712 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1713 </tr>
1714 <tr>
1715 <td><a name="t_firstclass">first class</a></td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1716 <td><a href="#t_integer">integer</a>,
1717 <a href="#t_floating">floating point</a>,
1718 <a href="#t_pointer">pointer</a>,
Dan Gohman0066db62008-06-18 18:42:13 +0000[diff] [blame]1719 <a href="#t_vector">vector</a>,
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]1720 <a href="#t_struct">structure</a>,
1721 <a href="#t_array">array</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1722 <a href="#t_label">label</a>,
1723 <a href="#t_metadata">metadata</a>.
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]1724 </td>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1725 </tr>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1726 <tr>
1727 <td><a href="#t_primitive">primitive</a></td>
1728 <td><a href="#t_label">label</a>,
1729 <a href="#t_void">void</a>,
Tobias Grosser05387292010-12-28 20:29:31 +0000[diff] [blame]1730 <a href="#t_integer">integer</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1731 <a href="#t_floating">floating point</a>,
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1732 <a href="#t_x86mmx">x86mmx</a>,
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1733 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1734 </tr>
1735 <tr>
1736 <td><a href="#t_derived">derived</a></td>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1737 <td><a href="#t_array">array</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1738 <a href="#t_function">function</a>,
1739 <a href="#t_pointer">pointer</a>,
1740 <a href="#t_struct">structure</a>,
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1741 <a href="#t_vector">vector</a>,
1742 <a href="#t_opaque">opaque</a>.
Dan Gohman01ac1012008-10-14 16:32:04 +0000[diff] [blame]1743 </td>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1744 </tr>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]1745 </tbody>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1746</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]1747
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1748<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1749 important. Values of these types are the only ones which can be produced by
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1750 instructions.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1751
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1752</div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1753
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1754<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1755<h3>
1756 <a name="t_primitive">Primitive Types</a>
1757</h3>
Chris Lattner8f8c7b72008-01-04 04:34:14 +0000[diff] [blame]1758
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1759<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1760
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1761<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1762 system.</p>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1763
1764<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1765<h4>
1766 <a name="t_integer">Integer Type</a>
1767</h4>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1768
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1769<div>
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1770
1771<h5>Overview:</h5>
1772<p>The integer type is a very simple type that simply specifies an arbitrary
1773 bit width for the integer type desired. Any bit width from 1 bit to
1774 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1775
1776<h5>Syntax:</h5>
1777<pre>
1778 iN
1779</pre>
1780
1781<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1782 value.</p>
1783
1784<h5>Examples:</h5>
1785<table class="layout">
1786 <tr class="layout">
1787 <td class="left"><tt>i1</tt></td>
1788 <td class="left">a single-bit integer.</td>
1789 </tr>
1790 <tr class="layout">
1791 <td class="left"><tt>i32</tt></td>
1792 <td class="left">a 32-bit integer.</td>
1793 </tr>
1794 <tr class="layout">
1795 <td class="left"><tt>i1942652</tt></td>
1796 <td class="left">a really big integer of over 1 million bits.</td>
1797 </tr>
1798</table>
1799
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1800</div>
1801
1802<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1803<h4>
1804 <a name="t_floating">Floating Point Types</a>
1805</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1806
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1807<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1808
1809<table>
1810 <tbody>
1811 <tr><th>Type</th><th>Description</th></tr>
1812 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1813 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1814 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1815 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1816 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1817 </tbody>
1818</table>
1819
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1820</div>
1821
1822<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1823<h4>
1824 <a name="t_x86mmx">X86mmx Type</a>
1825</h4>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1826
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1827<div>
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1828
1829<h5>Overview:</h5>
1830<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>
1831
1832<h5>Syntax:</h5>
1833<pre>
Dale Johannesen473a8c82010-10-01 01:07:02 +0000[diff] [blame]1834 x86mmx
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]1835</pre>
1836
1837</div>
1838
1839<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1840<h4>
1841 <a name="t_void">Void Type</a>
1842</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1843
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1844<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1845
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1846<h5>Overview:</h5>
1847<p>The void type does not represent any value and has no size.</p>
1848
1849<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1850<pre>
1851 void
1852</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1853
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1854</div>
1855
1856<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1857<h4>
1858 <a name="t_label">Label Type</a>
1859</h4>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1860
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1861<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1862
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1863<h5>Overview:</h5>
1864<p>The label type represents code labels.</p>
1865
1866<h5>Syntax:</h5>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1867<pre>
1868 label
1869</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1870
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1871</div>
1872
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1873<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1874<h4>
1875 <a name="t_metadata">Metadata Type</a>
1876</h4>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1877
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1878<div>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1879
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1880<h5>Overview:</h5>
Nick Lewyckyc261df92009-09-27 23:27:42 +0000[diff] [blame]1881<p>The metadata type represents embedded metadata. No derived types may be
1882 created from metadata except for <a href="#t_function">function</a>
1883 arguments.
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1884
1885<h5>Syntax:</h5>
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1886<pre>
1887 metadata
1888</pre>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]1889
Nick Lewycky7a0370f2009-05-30 05:06:04 +0000[diff] [blame]1890</div>
1891
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1892</div>
Chris Lattner4f69f462008-01-04 04:32:38 +0000[diff] [blame]1893
1894<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1895<h3>
1896 <a name="t_derived">Derived Types</a>
1897</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1898
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1899<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1900
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1901<p>The real power in LLVM comes from the derived types in the system. This is
1902 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]1903 useful types. Each of these types contain one or more element types which
1904 may be a primitive type, or another derived type. For example, it is
1905 possible to have a two dimensional array, using an array as the element type
1906 of another array.</p>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]1907
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1908<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1909<h4>
1910 <a name="t_aggregate">Aggregate Types</a>
1911</h4>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1912
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1913<div>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1914
1915<p>Aggregate Types are a subset of derived types that can contain multiple
Duncan Sands20536b52011-12-14 15:44:20 +0000[diff] [blame^]1916 member types. <a href="#t_array">Arrays</a> and
1917 <a href="#t_struct">structs</a> are aggregate types.
1918 <a href="#t_vector">Vectors</a> are not considered to be aggregate types.</p>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]1919
1920</div>
1921
Reid Spencer2b916312007-05-16 18:44:01 +0000[diff] [blame]1922<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1923<h4>
1924 <a name="t_array">Array Type</a>
1925</h4>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1926
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1927<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1928
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1929<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1930<p>The array type is a very simple derived type that arranges elements
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1931 sequentially in memory. The array type requires a size (number of elements)
1932 and an underlying data type.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1933
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1934<h5>Syntax:</h5>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1935<pre>
1936 [&lt;# elements&gt; x &lt;elementtype&gt;]
1937</pre>
1938
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1939<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1940 be any type with a size.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]1941
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]1942<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1943<table class="layout">
1944 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]1945 <td class="left"><tt>[40 x i32]</tt></td>
1946 <td class="left">Array of 40 32-bit integer values.</td>
1947 </tr>
1948 <tr class="layout">
1949 <td class="left"><tt>[41 x i32]</tt></td>
1950 <td class="left">Array of 41 32-bit integer values.</td>
1951 </tr>
1952 <tr class="layout">
1953 <td class="left"><tt>[4 x i8]</tt></td>
1954 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1955 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1956</table>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1957<p>Here are some examples of multidimensional arrays:</p>
1958<table class="layout">
1959 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]1960 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1961 <td class="left">3x4 array of 32-bit integer values.</td>
1962 </tr>
1963 <tr class="layout">
1964 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1965 <td class="left">12x10 array of single precision floating point values.</td>
1966 </tr>
1967 <tr class="layout">
1968 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1969 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1970 </tr>
1971</table>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1972
Dan Gohman7657f6b2009-11-09 19:01:53 +0000[diff] [blame]1973<p>There is no restriction on indexing beyond the end of the array implied by
1974 a static type (though there are restrictions on indexing beyond the bounds
1975 of an allocated object in some cases). This means that single-dimension
1976 'variable sized array' addressing can be implemented in LLVM with a zero
1977 length array type. An implementation of 'pascal style arrays' in LLVM could
1978 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]1979
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]1980</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]1981
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1982<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]1983<h4>
1984 <a name="t_function">Function Type</a>
1985</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1986
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]1987<div>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1988
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1989<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]1990<p>The function type can be thought of as a function signature. It consists of
1991 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]1992 function type is a first class type or a void type.</p>
Devang Patelc3fc6df2008-03-10 20:49:15 +0000[diff] [blame]1993
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]1994<h5>Syntax:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1995<pre>
Nick Lewycky51386942009-09-27 07:55:32 +0000[diff] [blame]1996 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]1997</pre>
1998
John Criswell0ec250c2005-10-24 16:17:18 +0000[diff] [blame]1999<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]2000 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
2001 which indicates that the function takes a variable number of arguments.
2002 Variable argument functions can access their arguments with
2003 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]2004 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewyckyc261df92009-09-27 23:27:42 +0000[diff] [blame]2005 <a href="#t_label">label</a>.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]2006
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2007<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2008<table class="layout">
2009 <tr class="layout">
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2010 <td class="left"><tt>i32 (i32)</tt></td>
2011 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2012 </td>
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2013 </tr><tr class="layout">
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]2014 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencerf17a0b72006-12-31 07:20:23 +0000[diff] [blame]2015 </tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2016 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]2017 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
2018 returning <tt>float</tt>.
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2019 </td>
2020 </tr><tr class="layout">
2021 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2022 <td class="left">A vararg function that takes at least one
2023 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
2024 which returns an integer. This is the signature for <tt>printf</tt> in
Reid Spencer92f82302006-12-31 07:18:34 +0000[diff] [blame]2025 LLVM.
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2026 </td>
Devang Patela582f402008-03-24 05:35:41 +0000[diff] [blame]2027 </tr><tr class="layout">
2028 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky51386942009-09-27 07:55:32 +0000[diff] [blame]2029 <td class="left">A function taking an <tt>i32</tt>, returning a
2030 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Patela582f402008-03-24 05:35:41 +0000[diff] [blame]2031 </td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2032 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2033</table>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]2034
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2035</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2036
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2037<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2038<h4>
2039 <a name="t_struct">Structure Type</a>
2040</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2041
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2042<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2043
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2044<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2045<p>The structure type is used to represent a collection of data members together
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2046 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]2047
Jeffrey Yasskin7a088cf2010-01-11 19:19:26 +0000[diff] [blame]2048<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
2049 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
2050 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2051 Structures in registers are accessed using the
2052 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
2053 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2054
2055<p>Structures may optionally be "packed" structures, which indicate that the
2056 alignment of the struct is one byte, and that there is no padding between
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2057 the elements. In non-packed structs, padding between field types is inserted
2058 as defined by the TargetData string in the module, which is required to match
Chris Lattnere4617b02011-10-11 23:02:17 +0000[diff] [blame]2059 what the underlying code generator expects.</p>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2060
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2061<p>Structures can either be "literal" or "identified". A literal structure is
2062 defined inline with other types (e.g. <tt>{i32, i32}*</tt>) whereas identified
2063 types are always defined at the top level with a name. Literal types are
2064 uniqued by their contents and can never be recursive or opaque since there is
Chris Lattneraa175c32011-08-12 18:12:40 +0000[diff] [blame]2065 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]2066 never uniqued.
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2067</p>
2068
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2069<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2070<pre>
Chris Lattner2c38d652011-08-12 17:31:02 +0000[diff] [blame]2071 %T1 = type { &lt;type list&gt; } <i>; Identified normal struct type</i>
2072 %T2 = type &lt;{ &lt;type list&gt; }&gt; <i>; Identified packed struct type</i>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2073</pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2074
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2075<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2076<table class="layout">
2077 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2078 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
2079 <td class="left">A triple of three <tt>i32</tt> values</td>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2080 </tr>
2081 <tr class="layout">
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]2082 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
2083 <td class="left">A pair, where the first element is a <tt>float</tt> and the
2084 second element is a <a href="#t_pointer">pointer</a> to a
2085 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
2086 an <tt>i32</tt>.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2087 </tr>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2088 <tr class="layout">
2089 <td class="left"><tt>&lt;{ i8, i32 }&gt;</tt></td>
2090 <td class="left">A packed struct known to be 5 bytes in size.</td>
2091 </tr>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2092</table>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]2093
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2094</div>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2095
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2096<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2097<h4>
Chris Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]2098 <a name="t_opaque">Opaque Structure Types</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2099</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2100
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2101<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2102
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2103<h5>Overview:</h5>
Chris Lattner628ed392011-07-23 19:59:08 +0000[diff] [blame]2104<p>Opaque structure types are used to represent named structure types that do
2105 not have a body specified. This corresponds (for example) to the C notion of
2106 a forward declared structure.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2107
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2108<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2109<pre>
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2110 %X = type opaque
2111 %52 = type opaque
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2112</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2113
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2114<h5>Examples:</h5>
2115<table class="layout">
2116 <tr class="layout">
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2117 <td class="left"><tt>opaque</tt></td>
2118 <td class="left">An opaque type.</td>
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2119 </tr>
2120</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2121
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2122</div>
2123
Chris Lattner1afcace2011-07-09 17:41:24 +0000[diff] [blame]2124
2125
Andrew Lenharth75e10682006-12-08 17:13:00 +0000[diff] [blame]2126<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2127<h4>
2128 <a name="t_pointer">Pointer Type</a>
2129</h4>
Chris Lattner0fd4a272009-02-08 19:53:29 +0000[diff] [blame]2130
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2131<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2132
2133<h5>Overview:</h5>
Dan Gohmanff3ef322010-02-25 16:50:07 +0000[diff] [blame]2134<p>The pointer type is used to specify memory locations.
2135 Pointers are commonly used to reference objects in memory.</p>
2136
2137<p>Pointer types may have an optional address space attribute defining the
2138 numbered address space where the pointed-to object resides. The default
2139 address space is number zero. The semantics of non-zero address
2140 spaces are target-specific.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2141
2142<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
2143 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner0fd4a272009-02-08 19:53:29 +0000[diff] [blame]2144
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2145<h5>Syntax:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]2146<pre>
2147 &lt;type&gt; *
2148</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2149
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]2150<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2151<table class="layout">
2152 <tr class="layout">
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]2153 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2154 <td class="left">A <a href="#t_pointer">pointer</a> to <a
2155 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
2156 </tr>
2157 <tr class="layout">
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]2158 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2159 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]2160 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2161 <tt>i32</tt>.</td>
2162 </tr>
2163 <tr class="layout">
2164 <td class="left"><tt>i32 addrspace(5)*</tt></td>
2165 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
2166 that resides in address space #5.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2167 </tr>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2168</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2169
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2170</div>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2171
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2172<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2173<h4>
2174 <a name="t_vector">Vector Type</a>
2175</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2176
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2177<div>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2178
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2179<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2180<p>A vector type is a simple derived type that represents a vector of elements.
2181 Vector types are used when multiple primitive data are operated in parallel
2182 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sandsd40d14e2009-11-27 13:38:03 +0000[diff] [blame]2183 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2184 <a href="#t_firstclass">first class</a>.</p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2185
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2186<h5>Syntax:</h5>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2187<pre>
2188 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
2189</pre>
2190
Chris Lattner7d2e7be2010-10-10 18:20:35 +0000[diff] [blame]2191<p>The number of elements is a constant integer value larger than 0; elementtype
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]2192 may be any integer or floating point type, or a pointer to these types.
2193 Vectors of size zero are not allowed. </p>
Chris Lattner69c11bb2005-04-25 17:34:15 +0000[diff] [blame]2194
Chris Lattnera58561b2004-08-12 19:12:28 +0000[diff] [blame]2195<h5>Examples:</h5>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2196<table class="layout">
2197 <tr class="layout">
Chris Lattner23ff1f92007-12-19 05:04:11 +0000[diff] [blame]2198 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
2199 <td class="left">Vector of 4 32-bit integer values.</td>
2200 </tr>
2201 <tr class="layout">
2202 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
2203 <td class="left">Vector of 8 32-bit floating-point values.</td>
2204 </tr>
2205 <tr class="layout">
2206 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
2207 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2208 </tr>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]2209 <tr class="layout">
2210 <td class="left"><tt>&lt;4 x i64*&gt;</tt></td>
2211 <td class="left">Vector of 4 pointers to 64-bit integer values.</td>
2212 </tr>
Reid Spencerd3f876c2004-11-01 08:19:36 +0000[diff] [blame]2213</table>
Dan Gohmand8791e52009-01-24 15:58:40 +0000[diff] [blame]2214
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]2215</div>
2216
Bill Wendlingaf75f0c2011-07-31 06:47:33 +0000[diff] [blame]2217</div>
2218
NAKAMURA Takumi4b2e07a2011-10-31 13:04:26 +0000[diff] [blame]2219</div>
2220
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2221<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2222<h2><a name="constants">Constants</a></h2>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2223<!-- *********************************************************************** -->
2224
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2225<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2226
2227<p>LLVM has several different basic types of constants. This section describes
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2228 them all and their syntax.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2229
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2230<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2231<h3>
2232 <a name="simpleconstants">Simple Constants</a>
2233</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2234
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2235<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2236
2237<dl>
2238 <dt><b>Boolean constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2239 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]2240 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2241
2242 <dt><b>Integer constants</b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2243 <dd>Standard integers (such as '4') are constants of
2244 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2245 with integer types.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2246
2247 <dt><b>Floating point constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2248 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2249 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2250 notation (see below). The assembler requires the exact decimal value of a
2251 floating-point constant. For example, the assembler accepts 1.25 but
2252 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2253 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2254
2255 <dt><b>Null pointer constants</b></dt>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]2256 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2257 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2258</dl>
2259
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2260<p>The one non-intuitive notation for constants is the hexadecimal form of
2261 floating point constants. For example, the form '<tt>double
2262 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2263 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2264 constants are required (and the only time that they are generated by the
2265 disassembler) is when a floating point constant must be emitted but it cannot
2266 be represented as a decimal floating point number in a reasonable number of
2267 digits. For example, NaN's, infinities, and other special values are
2268 represented in their IEEE hexadecimal format so that assembly and disassembly
2269 do not cause any bits to change in the constants.</p>
2270
Dale Johannesenbd5e5a82009-02-11 22:14:51 +0000[diff] [blame]2271<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2272 represented using the 16-digit form shown above (which matches the IEEE754
2273 representation for double); float values must, however, be exactly
2274 representable as IEE754 single precision. Hexadecimal format is always used
2275 for long double, and there are three forms of long double. The 80-bit format
2276 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2277 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2278 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2279 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2280 currently supported target uses this format. Long doubles will only work if
2281 they match the long double format on your target. All hexadecimal formats
2282 are big-endian (sign bit at the left).</p>
2283
Dale Johannesen21fe99b2010-10-01 00:48:59 +0000[diff] [blame]2284<p>There are no constants of type x86mmx.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2285</div>
2286
2287<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2288<h3>
Bill Wendlingd9fe2982009-07-20 02:32:41 +0000[diff] [blame]2289<a name="aggregateconstants"></a> <!-- old anchor -->
2290<a name="complexconstants">Complex Constants</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2291</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2292
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2293<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2294
Chris Lattner70882792009-02-28 18:32:25 +0000[diff] [blame]2295<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2296 constants and smaller complex constants.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2297
2298<dl>
2299 <dt><b>Structure constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2300 <dd>Structure constants are represented with notation similar to structure
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2301 type definitions (a comma separated list of elements, surrounded by braces
2302 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2303 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2304 Structure constants must have <a href="#t_struct">structure type</a>, and
2305 the number and types of elements must match those specified by the
2306 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2307
2308 <dt><b>Array constants</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2309 <dd>Array constants are represented with notation similar to array type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2310 definitions (a comma separated list of elements, surrounded by square
2311 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2312 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2313 the number and types of elements must match those specified by the
2314 type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2315
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2316 <dt><b>Vector constants</b></dt>
Reid Spencer485bad12007-02-15 03:07:05 +0000[diff] [blame]2317 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2318 definitions (a comma separated list of elements, surrounded by
2319 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2320 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2321 have <a href="#t_vector">vector type</a>, and the number and types of
2322 elements must match those specified by the type.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2323
2324 <dt><b>Zero initialization</b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2325 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]2326 value to zero of <em>any</em> type, including scalar and
2327 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2328 This is often used to avoid having to print large zero initializers
2329 (e.g. for large arrays) and is always exactly equivalent to using explicit
2330 zero initializers.</dd>
Nick Lewycky21cc4462009-04-04 07:22:01 +0000[diff] [blame]2331
2332 <dt><b>Metadata node</b></dt>
Nick Lewycky1e8c7a62009-05-30 16:08:30 +0000[diff] [blame]2333 <dd>A metadata node is a structure-like constant with
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2334 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2335 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2336 be interpreted as part of the instruction stream, metadata is a place to
2337 attach additional information such as debug info.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2338</dl>
2339
2340</div>
2341
2342<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2343<h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2344 <a name="globalconstants">Global Variable and Function Addresses</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2345</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2346
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2347<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2348
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2349<p>The addresses of <a href="#globalvars">global variables</a>
2350 and <a href="#functionstructure">functions</a> are always implicitly valid
2351 (link-time) constants. These constants are explicitly referenced when
2352 the <a href="#identifiers">identifier for the global</a> is used and always
2353 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2354 legal LLVM file:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2355
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2356<pre class="doc_code">
Chris Lattnera18a4242007-06-06 18:28:13 +0000[diff] [blame]2357@X = global i32 17
2358@Y = global i32 42
2359@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2360</pre>
2361
2362</div>
2363
2364<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2365<h3>
2366 <a name="undefvalues">Undefined Values</a>
2367</h3>
2368
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2369<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2370
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2371<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]2372 indicates that the user of the value may receive an unspecified bit-pattern.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2373 Undefined values may be of any type (other than '<tt>label</tt>'
2374 or '<tt>void</tt>') and be used anywhere a constant is permitted.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2375
Chris Lattnerc608cb12009-09-11 01:49:31 +0000[diff] [blame]2376<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2377 program is well defined no matter what value is used. This gives the
2378 compiler more freedom to optimize. Here are some examples of (potentially
2379 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2380
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2381
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2382<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2383 %A = add %X, undef
2384 %B = sub %X, undef
2385 %C = xor %X, undef
2386Safe:
2387 %A = undef
2388 %B = undef
2389 %C = undef
2390</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2391
2392<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]2393 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]2394
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2395<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2396 %A = or %X, undef
2397 %B = and %X, undef
2398Safe:
2399 %A = -1
2400 %B = 0
2401Unsafe:
2402 %A = undef
2403 %B = undef
2404</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2405
2406<p>These logical operations have bits that are not always affected by the input.
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2407 For example, if <tt>%X</tt> has a zero bit, then the output of the
2408 '<tt>and</tt>' operation will always be a zero for that bit, no matter what
2409 the corresponding bit from the '<tt>undef</tt>' is. As such, it is unsafe to
2410 optimize or assume that the result of the '<tt>and</tt>' is '<tt>undef</tt>'.
2411 However, it is safe to assume that all bits of the '<tt>undef</tt>' could be
2412 0, and optimize the '<tt>and</tt>' to 0. Likewise, it is safe to assume that
2413 all the bits of the '<tt>undef</tt>' operand to the '<tt>or</tt>' could be
2414 set, allowing the '<tt>or</tt>' to be folded to -1.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2415
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2416<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2417 %A = select undef, %X, %Y
2418 %B = select undef, 42, %Y
2419 %C = select %X, %Y, undef
2420Safe:
2421 %A = %X (or %Y)
2422 %B = 42 (or %Y)
2423 %C = %Y
2424Unsafe:
2425 %A = undef
2426 %B = undef
2427 %C = undef
2428</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2429
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2430<p>This set of examples shows that undefined '<tt>select</tt>' (and conditional
2431 branch) conditions can go <em>either way</em>, but they have to come from one
2432 of the two operands. In the <tt>%A</tt> example, if <tt>%X</tt> and
2433 <tt>%Y</tt> were both known to have a clear low bit, then <tt>%A</tt> would
2434 have to have a cleared low bit. However, in the <tt>%C</tt> example, the
2435 optimizer is allowed to assume that the '<tt>undef</tt>' operand could be the
2436 same as <tt>%Y</tt>, allowing the whole '<tt>select</tt>' to be
2437 eliminated.</p>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2438
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2439<pre class="doc_code">
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2440 %A = xor undef, undef
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2441
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2442 %B = undef
2443 %C = xor %B, %B
2444
2445 %D = undef
2446 %E = icmp lt %D, 4
2447 %F = icmp gte %D, 4
2448
2449Safe:
2450 %A = undef
2451 %B = undef
2452 %C = undef
2453 %D = undef
2454 %E = undef
2455 %F = undef
2456</pre>
Chris Lattner48a109c2009-09-07 22:52:39 +0000[diff] [blame]2457
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2458<p>This example points out that two '<tt>undef</tt>' operands are not
2459 necessarily the same. This can be surprising to people (and also matches C
2460 semantics) where they assume that "<tt>X^X</tt>" is always zero, even
2461 if <tt>X</tt> is undefined. This isn't true for a number of reasons, but the
2462 short answer is that an '<tt>undef</tt>' "variable" can arbitrarily change
2463 its value over its "live range". This is true because the variable doesn't
2464 actually <em>have a live range</em>. Instead, the value is logically read
2465 from arbitrary registers that happen to be around when needed, so the value
2466 is not necessarily consistent over time. In fact, <tt>%A</tt> and <tt>%C</tt>
2467 need to have the same semantics or the core LLVM "replace all uses with"
2468 concept would not hold.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2469
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2470<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2471 %A = fdiv undef, %X
2472 %B = fdiv %X, undef
2473Safe:
2474 %A = undef
2475b: unreachable
2476</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2477
2478<p>These examples show the crucial difference between an <em>undefined
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2479 value</em> and <em>undefined behavior</em>. An undefined value (like
2480 '<tt>undef</tt>') is allowed to have an arbitrary bit-pattern. This means that
2481 the <tt>%A</tt> operation can be constant folded to '<tt>undef</tt>', because
2482 the '<tt>undef</tt>' could be an SNaN, and <tt>fdiv</tt> is not (currently)
2483 defined on SNaN's. However, in the second example, we can make a more
2484 aggressive assumption: because the <tt>undef</tt> is allowed to be an
2485 arbitrary value, we are allowed to assume that it could be zero. Since a
2486 divide by zero has <em>undefined behavior</em>, we are allowed to assume that
2487 the operation does not execute at all. This allows us to delete the divide and
2488 all code after it. Because the undefined operation "can't happen", the
2489 optimizer can assume that it occurs in dead code.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2490
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2491<pre class="doc_code">
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2492a: store undef -> %X
2493b: store %X -> undef
2494Safe:
2495a: &lt;deleted&gt;
2496b: unreachable
2497</pre>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2498
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2499<p>These examples reiterate the <tt>fdiv</tt> example: a store <em>of</em> an
2500 undefined value can be assumed to not have any effect; we can assume that the
2501 value is overwritten with bits that happen to match what was already there.
2502 However, a store <em>to</em> an undefined location could clobber arbitrary
2503 memory, therefore, it has undefined behavior.</p>
Chris Lattner6e9057b2009-09-07 23:33:52 +0000[diff] [blame]2504
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2505</div>
2506
2507<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2508<h3>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2509 <a name="poisonvalues">Poison Values</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2510</h3>
2511
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2512<div>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2513
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2514<p>Poison values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2515 they also represent the fact that an instruction or constant expression which
2516 cannot evoke side effects has nevertheless detected a condition which results
2517 in undefined behavior.</p>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2518
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2519<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]2520 only exist when produced by operations such as
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2521 <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]2522
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2523<p>Poison value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2524
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2525<ul>
2526<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2527 their operands.</li>
2528
2529<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2530 to their dynamic predecessor basic block.</li>
2531
2532<li>Function arguments depend on the corresponding actual argument values in
2533 the dynamic callers of their functions.</li>
2534
2535<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2536 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2537 control back to them.</li>
2538
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2539<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2540 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2541 or exception-throwing call instructions that dynamically transfer control
2542 back to them.</li>
2543
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2544<li>Non-volatile loads and stores depend on the most recent stores to all of the
2545 referenced memory addresses, following the order in the IR
2546 (including loads and stores implied by intrinsics such as
2547 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2548
Dan Gohman7c24ff12010-05-03 14:59:34 +0000[diff] [blame]2549<!-- TODO: In the case of multiple threads, this only applies if the store
2550 "happens-before" the load or store. -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2551
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2552<!-- TODO: floating-point exception state -->
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2553
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2554<li>An instruction with externally visible side effects depends on the most
2555 recent preceding instruction with externally visible side effects, following
Dan Gohmanff70fe42010-07-06 15:26:33 +0000[diff] [blame]2556 the order in the IR. (This includes
2557 <a href="#volatile">volatile operations</a>.)</li>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2558
Dan Gohmanb5328162010-05-03 14:55:22 +0000[diff] [blame]2559<li>An instruction <i>control-depends</i> on a
2560 <a href="#terminators">terminator instruction</a>
2561 if the terminator instruction has multiple successors and the instruction
2562 is always executed when control transfers to one of the successors, and
Chris Lattner7a2bdde2011-04-15 05:18:47 +0000[diff] [blame]2563 may not be executed when control is transferred to another.</li>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2564
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2565<li>Additionally, an instruction also <i>control-depends</i> on a terminator
2566 instruction if the set of instructions it otherwise depends on would be
Chris Lattner7a2bdde2011-04-15 05:18:47 +0000[diff] [blame]2567 different if the terminator had transferred control to a different
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2568 successor.</li>
2569
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2570<li>Dependence is transitive.</li>
2571
2572</ul>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2573
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2574<p>Poison Values have the same behavior as <a href="#undefvalues">undef values</a>,
2575 with the additional affect that any instruction which has a <i>dependence</i>
2576 on a poison value has undefined behavior.</p>
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2577
2578<p>Here are some examples:</p>
Dan Gohmanc30f6e12010-04-26 20:54:53 +0000[diff] [blame]2579
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2580<pre class="doc_code">
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2581entry:
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2582 %poison = sub nuw i32 0, 1 ; Results in a poison value.
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2583 %still_poison = and i32 %poison, 0 ; 0, but also poison.
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2584 %poison_yet_again = getelementptr i32* @h, i32 %still_poison
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2585 store i32 0, i32* %poison_yet_again ; memory at @h[0] is poisoned
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2586
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2587 store i32 %poison, i32* @g ; Poison value stored to memory.
2588 %poison2 = load i32* @g ; Poison value loaded back from memory.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2589
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2590 store volatile i32 %poison, i32* @g ; External observation; undefined behavior.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2591
2592 %narrowaddr = bitcast i32* @g to i16*
2593 %wideaddr = bitcast i32* @g to i64*
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]2594 %poison3 = load i16* %narrowaddr ; Returns a poison value.
2595 %poison4 = load i64* %wideaddr ; Returns a poison value.
Dan Gohman34b3d992010-04-28 00:49:41 +0000[diff] [blame]2596
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2597 %cmp = icmp slt i32 %poison, 0 ; Returns a poison value.
2598 br i1 %cmp, label %true, label %end ; Branch to either destination.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2599
2600true:
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2601 store volatile i32 0, i32* @g ; This is control-dependent on %cmp, so
2602 ; it has undefined behavior.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2603 br label %end
2604
2605end:
2606 %p = phi i32 [ 0, %entry ], [ 1, %true ]
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2607 ; Both edges into this PHI are
2608 ; control-dependent on %cmp, so this
2609 ; always results in a poison value.
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2610
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2611 store volatile i32 0, i32* @g ; This would depend on the store in %true
2612 ; if %cmp is true, or the store in %entry
2613 ; otherwise, so this is undefined behavior.
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2614
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2615 br i1 %cmp, label %second_true, label %second_end
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2616 ; The same branch again, but this time the
2617 ; true block doesn't have side effects.
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2618
2619second_true:
2620 ; No side effects!
Nick Lewycky64f9fb12011-05-16 19:29:30 +0000[diff] [blame]2621 ret void
Dan Gohmanca4cac42011-04-12 23:05:59 +0000[diff] [blame]2622
2623second_end:
Dan Gohman5cdc51e2011-12-06 03:31:14 +0000[diff] [blame]2624 store volatile i32 0, i32* @g ; This time, the instruction always depends
2625 ; on the store in %end. Also, it is
2626 ; control-equivalent to %end, so this is
Dan Gohmane1a29842011-12-06 03:35:58 +0000[diff] [blame]2627 ; well-defined (ignoring earlier undefined
2628 ; behavior in this example).
Dan Gohmanae11c3f2010-04-26 23:36:52 +0000[diff] [blame]2629</pre>
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2630
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]2631</div>
2632
2633<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2634<h3>
2635 <a name="blockaddress">Addresses of Basic Blocks</a>
2636</h3>
2637
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2638<div>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2639
Chris Lattnercdfc9402009-11-01 01:27:45 +0000[diff] [blame]2640<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2641
2642<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner2dfdf2a2009-10-27 21:49:40 +0000[diff] [blame]2643 basic block in the specified function, and always has an i8* type. Taking
Chris Lattnercdfc9402009-11-01 01:27:45 +0000[diff] [blame]2644 the address of the entry block is illegal.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2645
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2646<p>This value only has defined behavior when used as an operand to the
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2647 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction, or for
2648 comparisons against null. Pointer equality tests between labels addresses
2649 results in undefined behavior &mdash; though, again, comparison against null
2650 is ok, and no label is equal to the null pointer. This may be passed around
2651 as an opaque pointer sized value as long as the bits are not inspected. This
2652 allows <tt>ptrtoint</tt> and arithmetic to be performed on these values so
2653 long as the original value is reconstituted before the <tt>indirectbr</tt>
2654 instruction.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]2655
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2656<p>Finally, some targets may provide defined semantics when using the value as
2657 the operand to an inline assembly, but that is target specific.</p>
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]2658
2659</div>
2660
2661
2662<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2663<h3>
2664 <a name="constantexprs">Constant Expressions</a>
2665</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2666
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2667<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2668
2669<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2670 to be used as constants. Constant expressions may be of
2671 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2672 operation that does not have side effects (e.g. load and call are not
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]2673 supported). The following is the syntax for constant expressions:</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2674
2675<dl>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2676 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2677 <dd>Truncate a constant to another type. The bit size of CST must be larger
2678 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2679
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2680 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2681 <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]2682 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2683
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2684 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2685 <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]2686 smaller than the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2687
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2688 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2689 <dd>Truncate a floating point constant to another floating point type. The
2690 size of CST must be larger than the size of TYPE. Both types must be
2691 floating point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2692
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2693 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2694 <dd>Floating point extend a constant to another type. The size of CST must be
2695 smaller or equal to the size of TYPE. Both types must be floating
2696 point.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2697
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2698 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2699 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2700 constant. TYPE must be a scalar or vector integer type. CST must be of
2701 scalar or vector floating point type. Both CST and TYPE must be scalars,
2702 or vectors of the same number of elements. If the value won't fit in the
2703 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2704
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2705 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2706 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2707 constant. TYPE must be a scalar or vector integer type. CST must be of
2708 scalar or vector floating point type. Both CST and TYPE must be scalars,
2709 or vectors of the same number of elements. If the value won't fit in the
2710 integer type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2711
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2712 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2713 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2714 constant. TYPE must be a scalar or vector floating point type. CST must be
2715 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2716 vectors of the same number of elements. If the value won't fit in the
2717 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2718
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2719 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2720 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2721 constant. TYPE must be a scalar or vector floating point type. CST must be
2722 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2723 vectors of the same number of elements. If the value won't fit in the
2724 floating point type, the results are undefined.</dd>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]2725
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2726 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2727 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2728 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2729 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2730 make it fit in <tt>TYPE</tt>.</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2731
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2732 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2733 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2734 type. CST must be of integer type. The CST value is zero extended,
2735 truncated, or unchanged to make it fit in a pointer size. This one is
2736 <i>really</i> dangerous!</dd>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]2737
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2738 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner03bbad62009-02-28 18:27:03 +0000[diff] [blame]2739 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2740 are the same as those for the <a href="#i_bitcast">bitcast
2741 instruction</a>.</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2742
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2743 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2744 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2745 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2746 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2747 instruction, the index list may have zero or more indexes, which are
2748 required to make sense for the type of "CSTPTR".</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2749
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2750 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2751 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2752
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2753 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2754 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2755
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2756 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Reid Spencer01c42592006-12-04 19:23:19 +0000[diff] [blame]2757 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]2758
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2759 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2760 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2761 constants.</dd>
Robert Bocchino9fbe1452006-01-10 19:31:34 +0000[diff] [blame]2762
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2763 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2764 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2765 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]2766
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2767 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2768 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2769 constants.</dd>
Chris Lattnerc1989542006-04-08 00:13:41 +0000[diff] [blame]2770
Nick Lewycky9e130ce2010-05-29 06:44:15 +0000[diff] [blame]2771 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2772 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2773 constants. The index list is interpreted in a similar manner as indices in
2774 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2775 index value must be specified.</dd>
2776
2777 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2778 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2779 constants. The index list is interpreted in a similar manner as indices in
2780 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2781 index value must be specified.</dd>
2782
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]2783 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2784 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2785 be any of the <a href="#binaryops">binary</a>
2786 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2787 on operands are the same as those for the corresponding instruction
2788 (e.g. no bitwise operations on floating point values are allowed).</dd>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2789</dl>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2790
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]2791</div>
Chris Lattner9ee5d222004-03-08 16:49:10 +0000[diff] [blame]2792
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2793</div>
2794
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]2795<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2796<h2><a name="othervalues">Other Values</a></h2>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2797<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2798<div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2799<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2800<h3>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2801<a name="inlineasm">Inline Assembler Expressions</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2802</h3>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2803
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2804<div>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2805
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2806<p>LLVM supports inline assembler expressions (as opposed
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2807 to <a href="#moduleasm">Module-Level Inline Assembly</a>) through the use of
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2808 a special value. This value represents the inline assembler as a string
2809 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2810 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2811 expression has side effects, and a flag indicating whether the function
2812 containing the asm needs to align its stack conservatively. An example
2813 inline assembler expression is:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2814
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2815<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2816i32 (i32) asm "bswap $0", "=r,r"
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2817</pre>
2818
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2819<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2820 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2821 have:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2822
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2823<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2824%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]2825</pre>
2826
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2827<p>Inline asms with side effects not visible in the constraint list must be
2828 marked as having side effects. This is done through the use of the
2829 '<tt>sideeffect</tt>' keyword, like so:</p>
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2830
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2831<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]2832call void asm sideeffect "eieio", ""()
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2833</pre>
2834
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2835<p>In some cases inline asms will contain code that will not work unless the
2836 stack is aligned in some way, such as calls or SSE instructions on x86,
2837 yet will not contain code that does that alignment within the asm.
2838 The compiler should make conservative assumptions about what the asm might
2839 contain and should generate its usual stack alignment code in the prologue
2840 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2841
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2842<pre class="doc_code">
Dale Johannesen8ba2d5b2009-10-21 23:28:00 +0000[diff] [blame]2843call void asm alignstack "eieio", ""()
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2844</pre>
Dale Johannesen09fed252009-10-13 21:56:55 +0000[diff] [blame]2845
2846<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2847 first.</p>
2848
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2849<!--
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]2850<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]2851 documented here. Constraints on what can be done (e.g. duplication, moving,
2852 etc need to be documented). This is probably best done by reference to
2853 another document that covers inline asm from a holistic perspective.</p>
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2854 -->
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2855
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2856<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2857<h4>
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2858 <a name="inlineasm_md">Inline Asm Metadata</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2859</h4>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2860
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2861<div>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2862
Bill Wendlingaee0f452011-11-30 21:52:43 +0000[diff] [blame]2863<p>The call instructions that wrap inline asm nodes may have a
2864 "<tt>!srcloc</tt>" MDNode attached to it that contains a list of constant
2865 integers. If present, the code generator will use the integer as the
2866 location cookie value when report errors through the <tt>LLVMContext</tt>
2867 error reporting mechanisms. This allows a front-end to correlate backend
2868 errors that occur with inline asm back to the source code that produced it.
2869 For example:</p>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2870
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]2871<pre class="doc_code">
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2872call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2873...
2874!42 = !{ i32 1234567 }
2875</pre>
Chris Lattnercf9a4152010-04-07 05:38:05 +0000[diff] [blame]2876
2877<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]2878 IR. If the MDNode contains multiple constants, the code generator will use
Chris Lattnerce1b9ad2010-11-17 08:20:42 +0000[diff] [blame]2879 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]2880
2881</div>
2882
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2883</div>
2884
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2885<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]2886<h3>
2887 <a name="metadata">Metadata Nodes and Metadata Strings</a>
2888</h3>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2889
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]2890<div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2891
2892<p>LLVM IR allows metadata to be attached to instructions in the program that
2893 can convey extra information about the code to the optimizers and code
2894 generator. One example application of metadata is source-level debug
2895 information. There are two metadata primitives: strings and nodes. All
2896 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2897 preceding exclamation point ('<tt>!</tt>').</p>
2898
2899<p>A metadata string is a string surrounded by double quotes. It can contain
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2900 any character by escaping non-printable characters with "<tt>\xx</tt>" where
2901 "<tt>xx</tt>" is the two digit hex code. For example:
2902 "<tt>!"test\00"</tt>".</p>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2903
2904<p>Metadata nodes are represented with notation similar to structure constants
2905 (a comma separated list of elements, surrounded by braces and preceded by an
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2906 exclamation point). Metadata nodes can have any values as their operand. For
2907 example:</p>
2908
2909<div class="doc_code">
2910<pre>
2911!{ metadata !"test\00", i32 10}
2912</pre>
2913</div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2914
2915<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2916 metadata nodes, which can be looked up in the module symbol table. For
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2917 example:</p>
2918
2919<div class="doc_code">
2920<pre>
2921!foo = metadata !{!4, !3}
2922</pre>
2923</div>
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]2924
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2925<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2926 function is using two metadata arguments:</p>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2927
Bill Wendling9ff5de92011-03-02 02:17:11 +0000[diff] [blame]2928<div class="doc_code">
2929<pre>
2930call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2931</pre>
2932</div>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2933
2934<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2935 attached to the <tt>add</tt> instruction using the <tt>!dbg</tt>
2936 identifier:</p>
Devang Patele1d50cd2010-03-04 23:44:48 +0000[diff] [blame]2937
Bill Wendling9ff5de92011-03-02 02:17:11 +0000[diff] [blame]2938<div class="doc_code">
2939<pre>
2940%indvar.next = add i64 %indvar, 1, !dbg !21
2941</pre>
2942</div>
2943
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]2944<p>More information about specific metadata nodes recognized by the optimizers
2945 and code generator is found below.</p>
2946
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2947<!-- _______________________________________________________________________ -->
Peter Collingbourne249d9532011-10-27 19:19:07 +0000[diff] [blame]2948<h4>
2949 <a name="tbaa">'<tt>tbaa</tt>' Metadata</a>
2950</h4>
2951
2952<div>
2953
2954<p>In LLVM IR, memory does not have types, so LLVM's own type system is not
2955 suitable for doing TBAA. Instead, metadata is added to the IR to describe
2956 a type system of a higher level language. This can be used to implement
2957 typical C/C++ TBAA, but it can also be used to implement custom alias
2958 analysis behavior for other languages.</p>
2959
2960<p>The current metadata format is very simple. TBAA metadata nodes have up to
2961 three fields, e.g.:</p>
2962
2963<div class="doc_code">
2964<pre>
2965!0 = metadata !{ metadata !"an example type tree" }
2966!1 = metadata !{ metadata !"int", metadata !0 }
2967!2 = metadata !{ metadata !"float", metadata !0 }
2968!3 = metadata !{ metadata !"const float", metadata !2, i64 1 }
2969</pre>
2970</div>
2971
2972<p>The first field is an identity field. It can be any value, usually
2973 a metadata string, which uniquely identifies the type. The most important
2974 name in the tree is the name of the root node. Two trees with
2975 different root node names are entirely disjoint, even if they
2976 have leaves with common names.</p>
2977
2978<p>The second field identifies the type's parent node in the tree, or
2979 is null or omitted for a root node. A type is considered to alias
2980 all of its descendants and all of its ancestors in the tree. Also,
2981 a type is considered to alias all types in other trees, so that
2982 bitcode produced from multiple front-ends is handled conservatively.</p>
2983
2984<p>If the third field is present, it's an integer which if equal to 1
2985 indicates that the type is "constant" (meaning
2986 <tt>pointsToConstantMemory</tt> should return true; see
2987 <a href="AliasAnalysis.html#OtherItfs">other useful
2988 <tt>AliasAnalysis</tt> methods</a>).</p>
2989
2990</div>
2991
Bill Wendlingf6cc4c22011-11-30 21:43:43 +0000[diff] [blame]2992<!-- _______________________________________________________________________ -->
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]2993<h4>
2994 <a name="fpaccuracy">'<tt>fpaccuracy</tt>' Metadata</a>
2995</h4>
2996
2997<div>
2998
2999<p><tt>fpaccuracy</tt> metadata may be attached to any instruction of floating
3000 point type. It expresses the maximum relative error of the result of
3001 that instruction, in ULPs. ULP is defined as follows:</p>
3002
Bill Wendling0656e252011-11-09 19:33:56 +0000[diff] [blame]3003<blockquote>
3004
3005<p>If <tt>x</tt> is a real number that lies between two finite consecutive
3006 floating-point numbers <tt>a</tt> and <tt>b</tt>, without being equal to one
3007 of them, then <tt>ulp(x) = |b - a|</tt>, otherwise <tt>ulp(x)</tt> is the
3008 distance between the two non-equal finite floating-point numbers nearest
3009 <tt>x</tt>. Moreover, <tt>ulp(NaN)</tt> is <tt>NaN</tt>.</p>
3010
3011</blockquote>
Peter Collingbourne999f90b2011-10-27 19:19:14 +0000[diff] [blame]3012
3013<p>The maximum relative error may be any rational number. The metadata node
3014 shall consist of a pair of unsigned integers respectively representing
3015 the numerator and denominator. For example, 2.5 ULP:</p>
3016
3017<div class="doc_code">
3018<pre>
3019!0 = metadata !{ i32 5, i32 2 }
3020</pre>
3021</div>
3022
3023</div>
3024
Chris Lattnere6a5ddd2010-01-15 21:50:19 +0000[diff] [blame]3025</div>
3026
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3027</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3028
3029<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3030<h2>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3031 <a name="intrinsic_globals">Intrinsic Global Variables</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3032</h2>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3033<!-- *********************************************************************** -->
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3034<div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3035<p>LLVM has a number of "magic" global variables that contain data that affect
3036code generation or other IR semantics. These are documented here. All globals
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3037of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
3038section and all globals that start with "<tt>llvm.</tt>" are reserved for use
3039by LLVM.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3040
3041<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3042<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3043<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3044</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3045
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3046<div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3047
3048<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
3049href="#linkage_appending">appending linkage</a>. This array contains a list of
3050pointers to global variables and functions which may optionally have a pointer
3051cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
3052
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3053<div class="doc_code">
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3054<pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3055@X = global i8 4
3056@Y = global i32 123
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3057
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3058@llvm.used = appending global [2 x i8*] [
3059 i8* @X,
3060 i8* bitcast (i32* @Y to i8*)
3061], section "llvm.metadata"
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3062</pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3063</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3064
3065<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]3066 compiler, assembler, and linker are required to treat the symbol as if there
3067 is a reference to the global that it cannot see. For example, if a variable
3068 has internal linkage and no references other than that from
3069 the <tt>@llvm.used</tt> list, it cannot be deleted. This is commonly used to
3070 represent references from inline asms and other things the compiler cannot
3071 "see", and corresponds to "<tt>attribute((used))</tt>" in GNU C.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3072
3073<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]3074 object file to prevent the assembler and linker from molesting the
3075 symbol.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3076
3077</div>
3078
3079<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3080<h3>
3081 <a name="intg_compiler_used">
3082 The '<tt>llvm.compiler.used</tt>' Global Variable
3083 </a>
3084</h3>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3085
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3086<div>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3087
3088<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3089 <tt>@llvm.used</tt> directive, except that it only prevents the compiler from
3090 touching the symbol. On targets that support it, this allows an intelligent
3091 linker to optimize references to the symbol without being impeded as it would
3092 be by <tt>@llvm.used</tt>.</p>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3093
3094<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]3095 should not be exposed to source languages.</p>
Chris Lattner401e10c2009-07-20 06:14:25 +0000[diff] [blame]3096
3097</div>
3098
3099<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3100<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3101<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3102</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3103
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3104<div>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3105
3106<div class="doc_code">
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3107<pre>
3108%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +0000[diff] [blame]3109@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3110</pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3111</div>
3112
3113<p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor
3114 functions and associated priorities. The functions referenced by this array
3115 will be called in ascending order of priority (i.e. lowest first) when the
3116 module is loaded. The order of functions with the same priority is not
3117 defined.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3118
3119</div>
3120
3121<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3122<h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3123<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3124</h3>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3125
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3126<div>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3127
3128<div class="doc_code">
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3129<pre>
3130%0 = type { i32, void ()* }
David Chisnall27195a52010-04-30 19:27:35 +0000[diff] [blame]3131@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnalle31e9962010-04-30 19:23:49 +0000[diff] [blame]3132</pre>
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3133</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3134
Bill Wendling9ae75632011-11-08 00:32:45 +0000[diff] [blame]3135<p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions
3136 and associated priorities. The functions referenced by this array will be
3137 called in descending order of priority (i.e. highest first) when the module
3138 is loaded. The order of functions with the same priority is not defined.</p>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3139
3140</div>
3141
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3142</div>
Chris Lattner857755c2009-07-20 05:55:19 +0000[diff] [blame]3143
Chris Lattnere87d6532006-01-25 23:47:57 +0000[diff] [blame]3144<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3145<h2><a name="instref">Instruction Reference</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3146<!-- *********************************************************************** -->
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3147
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3148<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3149
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3150<p>The LLVM instruction set consists of several different classifications of
3151 instructions: <a href="#terminators">terminator
3152 instructions</a>, <a href="#binaryops">binary instructions</a>,
3153 <a href="#bitwiseops">bitwise binary instructions</a>,
3154 <a href="#memoryops">memory instructions</a>, and
3155 <a href="#otherops">other instructions</a>.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3156
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3157<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3158<h3>
3159 <a name="terminators">Terminator Instructions</a>
3160</h3>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3161
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3162<div>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3163
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3164<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
3165 in a program ends with a "Terminator" instruction, which indicates which
3166 block should be executed after the current block is finished. These
3167 terminator instructions typically yield a '<tt>void</tt>' value: they produce
3168 control flow, not values (the one exception being the
3169 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
3170
Chris Lattner6445ecb2011-08-02 20:29:13 +0000[diff] [blame]3171<p>The terminator instructions are:
3172 '<a href="#i_ret"><tt>ret</tt></a>',
3173 '<a href="#i_br"><tt>br</tt></a>',
3174 '<a href="#i_switch"><tt>switch</tt></a>',
3175 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>',
3176 '<a href="#i_invoke"><tt>invoke</tt></a>',
3177 '<a href="#i_unwind"><tt>unwind</tt></a>',
3178 '<a href="#i_resume"><tt>resume</tt></a>', and
3179 '<a href="#i_unreachable"><tt>unreachable</tt></a>'.</p>
Chris Lattnerc3f59762004-12-09 17:30:23 +0000[diff] [blame]3180
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3181<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3182<h4>
3183 <a name="i_ret">'<tt>ret</tt>' Instruction</a>
3184</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3185
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3186<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3187
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3188<h5>Syntax:</h5>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]3189<pre>
3190 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]3191 ret void <i>; Return from void function</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3192</pre>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3193
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3194<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3195<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
3196 a value) from a function back to the caller.</p>
3197
3198<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
3199 value and then causes control flow, and one that just causes control flow to
3200 occur.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3201
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3202<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3203<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
3204 return value. The type of the return value must be a
3205 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]3206
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3207<p>A function is not <a href="#wellformed">well formed</a> if it it has a
3208 non-void return type and contains a '<tt>ret</tt>' instruction with no return
3209 value or a return value with a type that does not match its type, or if it
3210 has a void return type and contains a '<tt>ret</tt>' instruction with a
3211 return value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3212
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3213<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3214<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
3215 the calling function's context. If the caller is a
3216 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
3217 instruction after the call. If the caller was an
3218 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
3219 the beginning of the "normal" destination block. If the instruction returns
3220 a value, that value shall set the call or invoke instruction's return
3221 value.</p>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3222
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3223<h5>Example:</h5>
Chris Lattnerf4cde4e2008-04-23 04:59:35 +0000[diff] [blame]3224<pre>
3225 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]3226 ret void <i>; Return from a void function</i>
Bill Wendling0a4bbbf2009-02-28 22:12:54 +0000[diff] [blame]3227 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]3228</pre>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +0000[diff] [blame]3229
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3230</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3231<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3232<h4>
3233 <a name="i_br">'<tt>br</tt>' Instruction</a>
3234</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3235
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3236<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3237
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3238<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3239<pre>
Bill Wendlingb3aa4712011-07-26 10:41:15 +0000[diff] [blame]3240 br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;
3241 br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3242</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3243
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3244<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3245<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
3246 different basic block in the current function. There are two forms of this
3247 instruction, corresponding to a conditional branch and an unconditional
3248 branch.</p>
3249
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3250<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3251<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
3252 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
3253 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
3254 target.</p>
3255
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3256<h5>Semantics:</h5>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3257<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3258 argument is evaluated. If the value is <tt>true</tt>, control flows to the
3259 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
3260 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
3261
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3262<h5>Example:</h5>
Bill Wendlingc39e3e02009-07-20 02:39:26 +0000[diff] [blame]3263<pre>
3264Test:
3265 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
3266 br i1 %cond, label %IfEqual, label %IfUnequal
3267IfEqual:
3268 <a href="#i_ret">ret</a> i32 1
3269IfUnequal:
3270 <a href="#i_ret">ret</a> i32 0
3271</pre>
3272
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3273</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3274
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3275<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3276<h4>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3277 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3278</h4>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3279
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3280<div>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3281
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3282<h5>Syntax:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3283<pre>
3284 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
3285</pre>
3286
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3287<h5>Overview:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3288<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]3289 several different places. It is a generalization of the '<tt>br</tt>'
3290 instruction, allowing a branch to occur to one of many possible
3291 destinations.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3292
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3293<h5>Arguments:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3294<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3295 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
3296 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
3297 The table is not allowed to contain duplicate constant entries.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3298
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3299<h5>Semantics:</h5>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3300<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3301 destinations. When the '<tt>switch</tt>' instruction is executed, this table
3302 is searched for the given value. If the value is found, control flow is
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]3303 transferred to the corresponding destination; otherwise, control flow is
3304 transferred to the default destination.</p>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3305
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3306<h5>Implementation:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3307<p>Depending on properties of the target machine and the particular
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3308 <tt>switch</tt> instruction, this instruction may be code generated in
3309 different ways. For example, it could be generated as a series of chained
3310 conditional branches or with a lookup table.</p>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3311
3312<h5>Example:</h5>
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3313<pre>
3314 <i>; Emulate a conditional br instruction</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]3315 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]3316 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3317
3318 <i>; Emulate an unconditional br instruction</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3319 switch i32 0, label %dest [ ]
Chris Lattnerc88c17b2004-02-24 04:54:45 +0000[diff] [blame]3320
3321 <i>; Implement a jump table:</i>
Dan Gohman2a08c532009-01-04 23:44:43 +0000[diff] [blame]3322 switch i32 %val, label %otherwise [ i32 0, label %onzero
3323 i32 1, label %onone
3324 i32 2, label %ontwo ]
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3325</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3326
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3327</div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3328
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3329
3330<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3331<h4>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3332 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3333</h4>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3334
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3335<div>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3336
3337<h5>Syntax:</h5>
3338<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3339 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]3340</pre>
3341
3342<h5>Overview:</h5>
3343
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3344<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3345 within the current function, whose address is specified by
Chris Lattnerc6f44362009-10-27 21:01:34 +0000[diff] [blame]3346 "<tt>address</tt>". Address must be derived from a <a
3347 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3348
3349<h5>Arguments:</h5>
3350
3351<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3352 rest of the arguments indicate the full set of possible destinations that the
3353 address may point to. Blocks are allowed to occur multiple times in the
3354 destination list, though this isn't particularly useful.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3355
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3356<p>This destination list is required so that dataflow analysis has an accurate
3357 understanding of the CFG.</p>
3358
3359<h5>Semantics:</h5>
3360
3361<p>Control transfers to the block specified in the address argument. All
3362 possible destination blocks must be listed in the label list, otherwise this
3363 instruction has undefined behavior. This implies that jumps to labels
3364 defined in other functions have undefined behavior as well.</p>
3365
3366<h5>Implementation:</h5>
3367
3368<p>This is typically implemented with a jump through a register.</p>
3369
3370<h5>Example:</h5>
3371<pre>
Chris Lattnerab21db72009-10-28 00:19:10 +0000[diff] [blame]3372 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnerf9be95f2009-10-27 19:13:16 +0000[diff] [blame]3373</pre>
3374
3375</div>
3376
3377
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3378<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3379<h4>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3380 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3381</h4>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3382
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3383<div>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3384
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3385<h5>Syntax:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3386<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]3387 &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]3388 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3389</pre>
3390
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]3391<h5>Overview:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3392<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3393 function, with the possibility of control flow transfer to either the
3394 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3395 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3396 control flow will return to the "normal" label. If the callee (or any
3397 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3398 instruction, control is interrupted and continued at the dynamically nearest
3399 "exception" label.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3400
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3401<p>The '<tt>exception</tt>' label is a
3402 <i><a href="ExceptionHandling.html#overview">landing pad</a></i> for the
3403 exception. As such, '<tt>exception</tt>' label is required to have the
3404 "<a href="#i_landingpad"><tt>landingpad</tt></a>" instruction, which contains
Chad Rosier85f5a1a2011-12-09 02:00:44 +0000[diff] [blame]3405 the information about the behavior of the program after unwinding
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3406 happens, as its first non-PHI instruction. The restrictions on the
3407 "<tt>landingpad</tt>" instruction's tightly couples it to the
3408 "<tt>invoke</tt>" instruction, so that the important information contained
3409 within the "<tt>landingpad</tt>" instruction can't be lost through normal
3410 code motion.</p>
3411
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3412<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3413<p>This instruction requires several arguments:</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3414
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3415<ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3416 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3417 convention</a> the call should use. If none is specified, the call
3418 defaults to using C calling conventions.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]3419
3420 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3421 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3422 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]3423
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3424 <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]3425 function value being invoked. In most cases, this is a direct function
3426 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3427 off an arbitrary pointer to function value.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3428
3429 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3430 function to be invoked. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3431
3432 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]3433 signature argument types and parameter attributes. All arguments must be
3434 of <a href="#t_firstclass">first class</a> type. If the function
3435 signature indicates the function accepts a variable number of arguments,
3436 the extra arguments can be specified.</li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3437
3438 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3439 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3440
3441 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3442 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3443
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]3444 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3445 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3446 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3447</ol>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3448
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3449<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3450<p>This instruction is designed to operate as a standard
3451 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3452 primary difference is that it establishes an association with a label, which
3453 is used by the runtime library to unwind the stack.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3454
3455<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3456 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3457 exception. Additionally, this is important for implementation of
3458 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3459
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3460<p>For the purposes of the SSA form, the definition of the value returned by the
3461 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3462 block to the "normal" label. If the callee unwinds then no return value is
3463 available.</p>
Dan Gohmanf96a4992009-05-22 21:47:08 +0000[diff] [blame]3464
Chris Lattnerdf7a6802010-01-15 18:08:37 +0000[diff] [blame]3465<p>Note that the code generator does not yet completely support unwind, and
3466that the invoke/unwind semantics are likely to change in future versions.</p>
3467
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3468<h5>Example:</h5>
Chris Lattnerbad10ee2005-05-06 22:57:40 +0000[diff] [blame]3469<pre>
Nick Lewyckyd703f652008-03-16 07:18:12 +0000[diff] [blame]3470 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3471 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckyd703f652008-03-16 07:18:12 +0000[diff] [blame]3472 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3473 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3474</pre>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3475
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3476</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3477
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3478<!-- _______________________________________________________________________ -->
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3479
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3480<h4>
3481 <a name="i_unwind">'<tt>unwind</tt>' Instruction</a>
3482</h4>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3483
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3484<div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3485
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3486<h5>Syntax:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3487<pre>
3488 unwind
3489</pre>
3490
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3491<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3492<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3493 at the first callee in the dynamic call stack which used
3494 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3495 This is primarily used to implement exception handling.</p>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3496
Chris Lattner27f71f22003-09-03 00:41:47 +0000[diff] [blame]3497<h5>Semantics:</h5>
Chris Lattner72ed2002008-04-19 21:01:16 +0000[diff] [blame]3498<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3499 immediately halt. The dynamic call stack is then searched for the
3500 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3501 Once found, execution continues at the "exceptional" destination block
3502 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3503 instruction in the dynamic call chain, undefined behavior results.</p>
3504
Chris Lattnerdf7a6802010-01-15 18:08:37 +0000[diff] [blame]3505<p>Note that the code generator does not yet completely support unwind, and
3506that the invoke/unwind semantics are likely to change in future versions.</p>
3507
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3508</div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3509
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3510 <!-- _______________________________________________________________________ -->
3511
3512<h4>
3513 <a name="i_resume">'<tt>resume</tt>' Instruction</a>
3514</h4>
3515
3516<div>
3517
3518<h5>Syntax:</h5>
3519<pre>
3520 resume &lt;type&gt; &lt;value&gt;
3521</pre>
3522
3523<h5>Overview:</h5>
3524<p>The '<tt>resume</tt>' instruction is a terminator instruction that has no
3525 successors.</p>
3526
3527<h5>Arguments:</h5>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3528<p>The '<tt>resume</tt>' instruction requires one argument, which must have the
Bill Wendlinge4ad50b2011-08-03 18:37:32 +0000[diff] [blame]3529 same type as the result of any '<tt>landingpad</tt>' instruction in the same
3530 function.</p>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3531
3532<h5>Semantics:</h5>
3533<p>The '<tt>resume</tt>' instruction resumes propagation of an existing
3534 (in-flight) exception whose unwinding was interrupted with
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3535 a <a href="#i_landingpad"><tt>landingpad</tt></a> instruction.</p>
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3536
3537<h5>Example:</h5>
3538<pre>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]3539 resume { i8*, i32 } %exn
Bill Wendlingdccc03b2011-07-31 06:30:59 +0000[diff] [blame]3540</pre>
3541
3542</div>
3543
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3544<!-- _______________________________________________________________________ -->
3545
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3546<h4>
3547 <a name="i_unreachable">'<tt>unreachable</tt>' Instruction</a>
3548</h4>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3549
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3550<div>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3551
3552<h5>Syntax:</h5>
3553<pre>
3554 unreachable
3555</pre>
3556
3557<h5>Overview:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3558<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3559 instruction is used to inform the optimizer that a particular portion of the
3560 code is not reachable. This can be used to indicate that the code after a
3561 no-return function cannot be reached, and other facts.</p>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3562
3563<h5>Semantics:</h5>
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3564<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3565
Chris Lattner35eca582004-10-16 18:04:13 +0000[diff] [blame]3566</div>
3567
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3568</div>
3569
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3570<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3571<h3>
3572 <a name="binaryops">Binary Operations</a>
3573</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3574
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3575<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3576
3577<p>Binary operators are used to do most of the computation in a program. They
3578 require two operands of the same type, execute an operation on them, and
3579 produce a single value. The operands might represent multiple data, as is
3580 the case with the <a href="#t_vector">vector</a> data type. The result value
3581 has the same type as its operands.</p>
3582
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3583<p>There are several different binary operators:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3584
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3585<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3586<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3587 <a name="i_add">'<tt>add</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3588</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3589
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3590<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3591
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3592<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3593<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3594 &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]3595 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3596 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3597 &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]3598</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3599
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3600<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3601<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3602
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3603<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3604<p>The two arguments to the '<tt>add</tt>' instruction must
3605 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3606 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3607
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3608<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3609<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3610
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3611<p>If the sum has unsigned overflow, the result returned is the mathematical
3612 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]3613
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3614<p>Because LLVM integers use a two's complement representation, this instruction
3615 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3616
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3617<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3618 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3619 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3620 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3621 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3622
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3623<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3624<pre>
3625 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3626</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3627
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3628</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3629
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3630<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3631<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3632 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3633</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3634
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3635<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3636
3637<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3638<pre>
3639 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3640</pre>
3641
3642<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3643<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3644
3645<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3646<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3647 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3648 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3649
3650<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3651<p>The value produced is the floating point sum of the two operands.</p>
3652
3653<h5>Example:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3654<pre>
3655 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3656</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3657
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3658</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3659
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3660<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3661<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3662 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3663</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3664
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3665<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3666
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3667<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3668<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3669 &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]3670 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3671 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3672 &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]3673</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3674
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3675<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3676<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3677 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3678
3679<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3680 '<tt>neg</tt>' instruction present in most other intermediate
3681 representations.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3682
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3683<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3684<p>The two arguments to the '<tt>sub</tt>' instruction must
3685 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3686 integer values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3687
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3688<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3689<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3690
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3691<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3692 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3693 result.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3694
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3695<p>Because LLVM integers use a two's complement representation, this instruction
3696 is appropriate for both signed and unsigned integers.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3697
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3698<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3699 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3700 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3701 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3702 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3703
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3704<h5>Example:</h5>
Bill Wendlingaac388b2007-05-29 09:42:13 +0000[diff] [blame]3705<pre>
3706 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]3707 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3708</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3709
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3710</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3711
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3712<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3713<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3714 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3715</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3716
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3717<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3718
3719<h5>Syntax:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3720<pre>
3721 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3722</pre>
3723
3724<h5>Overview:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3725<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3726 operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3727
3728<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3729 '<tt>fneg</tt>' instruction present in most other intermediate
3730 representations.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3731
3732<h5>Arguments:</h5>
Bill Wendlingd9fe2982009-07-20 02:32:41 +0000[diff] [blame]3733<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3734 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3735 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3736
3737<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3738<p>The value produced is the floating point difference of the two operands.</p>
3739
3740<h5>Example:</h5>
3741<pre>
3742 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3743 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3744</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3745
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3746</div>
3747
3748<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3749<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3750 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3751</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3752
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3753<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3754
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3755<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3756<pre>
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3757 &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]3758 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3759 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3760 &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]3761</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3762
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3763<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3764<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3765
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3766<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3767<p>The two arguments to the '<tt>mul</tt>' instruction must
3768 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3769 integer values. Both arguments must have identical types.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3770
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3771<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3772<p>The value produced is the integer product of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3773
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3774<p>If the result of the multiplication has unsigned overflow, the result
3775 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3776 width of the result.</p>
3777
3778<p>Because LLVM integers use a two's complement representation, and the result
3779 is the same width as the operands, this instruction returns the correct
3780 result for both signed and unsigned integers. If a full product
3781 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3782 be sign-extended or zero-extended as appropriate to the width of the full
3783 product.</p>
3784
Dan Gohman08d012e2009-07-22 22:44:56 +0000[diff] [blame]3785<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3786 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3787 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3788 is a <a href="#poisonvalues">poison value</a> if unsigned and/or signed overflow,
Dan Gohmanfff6c532010-04-22 23:14:21 +0000[diff] [blame]3789 respectively, occurs.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3790
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3791<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3792<pre>
3793 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3794</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3795
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]3796</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3797
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3798<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3799<h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3800 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3801</h4>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3802
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3803<div>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3804
3805<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3806<pre>
3807 &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]3808</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3809
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3810<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3811<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3812
3813<h5>Arguments:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3814<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3815 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3816 floating point values. Both arguments must have identical types.</p>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3817
3818<h5>Semantics:</h5>
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3819<p>The value produced is the floating point product of the two operands.</p>
3820
3821<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3822<pre>
3823 &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]3824</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3825
Dan Gohmanae3a0be2009-06-04 22:49:04 +0000[diff] [blame]3826</div>
3827
3828<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3829<h4>
3830 <a name="i_udiv">'<tt>udiv</tt>' Instruction</a>
3831</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3832
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3833<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3834
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3835<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3836<pre>
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]3837 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3838 &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]3839</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3840
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3841<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3842<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3843
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3844<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3845<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3846 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3847 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3848
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3849<h5>Semantics:</h5>
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3850<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3851
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3852<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3853 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3854
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3855<p>Division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3856
Chris Lattner35bda892011-02-06 21:44:57 +0000[diff] [blame]3857<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3858 <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]3859 multiple of %op2 (as such, "((a udiv exact b) mul b) == a").</p>
3860
3861
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3862<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3863<pre>
3864 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3865</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3866
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3867</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3868
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3869<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3870<h4>
3871 <a name="i_sdiv">'<tt>sdiv</tt>' Instruction</a>
3872</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3873
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3874<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3875
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3876<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3877<pre>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3878 &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]3879 &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]3880</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3881
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3882<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3883<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3884
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3885<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3886<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3887 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3888 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3889
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3890<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3891<p>The value produced is the signed integer quotient of the two operands rounded
3892 towards zero.</p>
3893
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3894<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3895 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3896
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3897<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3898 undefined behavior; this is a rare case, but can occur, for example, by doing
3899 a 32-bit division of -2147483648 by -1.</p>
3900
Dan Gohman9c5beed2009-07-22 00:04:19 +0000[diff] [blame]3901<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]3902 <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]3903 be rounded.</p>
Dan Gohmancbb38f22009-07-20 22:41:19 +0000[diff] [blame]3904
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3905<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3906<pre>
3907 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3908</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3909
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3910</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3911
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3912<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3913<h4>
3914 <a name="i_fdiv">'<tt>fdiv</tt>' Instruction</a>
3915</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3916
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3917<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3918
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]3919<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3920<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]3921 &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]3922</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3923
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3924<h5>Overview:</h5>
3925<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3926
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3927<h5>Arguments:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]3928<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3929 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3930 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3931
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3932<h5>Semantics:</h5>
Reid Spencer1628cec2006-10-26 06:15:43 +0000[diff] [blame]3933<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3934
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3935<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3936<pre>
3937 &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]3938</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3939
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3940</div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3941
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3942<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3943<h4>
3944 <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3945</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3946
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3947<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3948
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3949<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3950<pre>
3951 &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]3952</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3953
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3954<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3955<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3956 division of its two arguments.</p>
3957
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3958<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3959<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3960 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3961 values. Both arguments must have identical types.</p>
3962
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3963<h5>Semantics:</h5>
3964<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3965 This instruction always performs an unsigned division to get the
3966 remainder.</p>
3967
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3968<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3969 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3970
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]3971<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3972
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3973<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3974<pre>
3975 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3976</pre>
3977
3978</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3979
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]3980<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3981<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3982 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]3983</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3984
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]3985<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3986
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3987<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3988<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]3989 &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]3990</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]3991
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3992<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]3993<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3994 division of its two operands. This instruction can also take
3995 <a href="#t_vector">vector</a> versions of the values in which case the
3996 elements must be integers.</p>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +0000[diff] [blame]3997
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]3998<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]3999<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4000 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4001 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4002
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4003<h5>Semantics:</h5>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4004<p>This instruction returns the <i>remainder</i> of a division (where the result
Duncan Sandsdea3a5e2011-03-07 09:12:24 +0000[diff] [blame]4005 is either zero or has the same sign as the dividend, <tt>op1</tt>), not the
4006 <i>modulo</i> operator (where the result is either zero or has the same sign
4007 as the divisor, <tt>op2</tt>) of a value.
4008 For more information about the difference,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4009 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
4010 Math Forum</a>. For a table of how this is implemented in various languages,
4011 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
4012 Wikipedia: modulo operation</a>.</p>
4013
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4014<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4015 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
4016
Chris Lattner5ec89832008-01-28 00:36:27 +0000[diff] [blame]4017<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4018 Overflow also leads to undefined behavior; this is a rare case, but can
4019 occur, for example, by taking the remainder of a 32-bit division of
4020 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
4021 lets srem be implemented using instructions that return both the result of
4022 the division and the remainder.)</p>
4023
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4024<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4025<pre>
4026 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4027</pre>
4028
4029</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4030
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4031<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4032<h4>
4033 <a name="i_frem">'<tt>frem</tt>' Instruction</a>
4034</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4035
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4036<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4037
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4038<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4039<pre>
4040 &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]4041</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4042
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4043<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4044<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
4045 its two operands.</p>
4046
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4047<h5>Arguments:</h5>
4048<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4049 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
4050 floating point values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4051
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4052<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4053<p>This instruction returns the <i>remainder</i> of a division. The remainder
4054 has the same sign as the dividend.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4055
Reid Spencer0a783f72006-11-02 01:53:59 +0000[diff] [blame]4056<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4057<pre>
4058 &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]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>
Robert Bocchino7b81c752006-02-17 21:18:08 +0000[diff] [blame]4062
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4063</div>
4064
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]4065<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4066<h3>
4067 <a name="bitwiseops">Bitwise Binary Operations</a>
4068</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4069
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4070<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4071
4072<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
4073 program. They are generally very efficient instructions and can commonly be
4074 strength reduced from other instructions. They require two operands of the
4075 same type, execute an operation on them, and produce a single value. The
4076 resulting value is the same type as its operands.</p>
4077
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4078<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4079<h4>
4080 <a name="i_shl">'<tt>shl</tt>' Instruction</a>
4081</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4082
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4083<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4084
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4085<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4086<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4087 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4088 &lt;result&gt; = shl nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4089 &lt;result&gt; = shl nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4090 &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]4091</pre>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4092
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4093<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4094<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
4095 a specified number of bits.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4096
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4097<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4098<p>Both arguments to the '<tt>shl</tt>' instruction must be the
4099 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
4100 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4101
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4102<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4103<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
4104 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
4105 is (statically or dynamically) negative or equal to or larger than the number
4106 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4107 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4108 shift amount in <tt>op2</tt>.</p>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4109
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4110<p>If the <tt>nuw</tt> keyword is present, then the shift produces a
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4111 <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]4112 the <tt>nsw</tt> keyword is present, then the shift produces a
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4113 <a href="#poisonvalues">poison value</a> if it shifts out any bits that disagree
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4114 with the resultant sign bit. As such, NUW/NSW have the same semantics as
4115 they would if the shift were expressed as a mul instruction with the same
4116 nsw/nuw bits in (mul %op1, (shl 1, %op2)).</p>
4117
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4118<h5>Example:</h5>
4119<pre>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4120 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
4121 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
4122 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4123 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4124 &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]4125</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4126
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4127</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4128
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4129<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4130<h4>
4131 <a name="i_lshr">'<tt>lshr</tt>' Instruction</a>
4132</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4133
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4134<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4135
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4136<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4137<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4138 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4139 &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]4140</pre>
4141
4142<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4143<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
4144 operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4145
4146<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4147<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4148 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4149 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4150
4151<h5>Semantics:</h5>
4152<p>This instruction always performs a logical shift right operation. The most
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4153 significant bits of the result will be filled with zero bits after the shift.
4154 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
4155 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
4156 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
4157 shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4158
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4159<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4160 <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]4161 shifted out are non-zero.</p>
4162
4163
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4164<h5>Example:</h5>
4165<pre>
4166 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
4167 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
4168 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
4169 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4170 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4171 &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]4172</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4173
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4174</div>
4175
Reid Spencer8e11bf82007-02-02 13:57:07 +0000[diff] [blame]4176<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4177<h4>
4178 <a name="i_ashr">'<tt>ashr</tt>' Instruction</a>
4179</h4>
4180
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4181<div>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4182
4183<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4184<pre>
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4185 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4186 &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]4187</pre>
4188
4189<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4190<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
4191 operand shifted to the right a specified number of bits with sign
4192 extension.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4193
4194<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4195<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4196 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4197 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4198
4199<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4200<p>This instruction always performs an arithmetic shift right operation, The
4201 most significant bits of the result will be filled with the sign bit
4202 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
4203 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
4204 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
4205 the corresponding shift amount in <tt>op2</tt>.</p>
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4206
Chris Lattnerf067d582011-02-07 16:40:21 +0000[diff] [blame]4207<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]4208 <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]4209 shifted out are non-zero.</p>
4210
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4211<h5>Example:</h5>
4212<pre>
4213 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
4214 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
4215 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
4216 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattner6ccc2d52007-10-03 21:01:14 +0000[diff] [blame]4217 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wange9f10152008-12-09 05:46:39 +0000[diff] [blame]4218 &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]4219</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4220
Reid Spencer569f2fa2007-01-31 21:39:12 +0000[diff] [blame]4221</div>
4222
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4223<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4224<h4>
4225 <a name="i_and">'<tt>and</tt>' Instruction</a>
4226</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4227
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4228<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4229
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4230<h5>Syntax:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4231<pre>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]4232 &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]4233</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4234
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4235<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4236<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
4237 operands.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4238
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4239<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4240<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4241 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4242 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4243
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4244<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4245<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4246
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4247<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4248 <tbody>
4249 <tr>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4250 <th>In0</th>
4251 <th>In1</th>
4252 <th>Out</th>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4253 </tr>
4254 <tr>
4255 <td>0</td>
4256 <td>0</td>
4257 <td>0</td>
4258 </tr>
4259 <tr>
4260 <td>0</td>
4261 <td>1</td>
4262 <td>0</td>
4263 </tr>
4264 <tr>
4265 <td>1</td>
4266 <td>0</td>
4267 <td>0</td>
4268 </tr>
4269 <tr>
4270 <td>1</td>
4271 <td>1</td>
4272 <td>1</td>
4273 </tr>
4274 </tbody>
4275</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4276
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4277<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4278<pre>
4279 &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]4280 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
4281 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4282</pre>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4283</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4284<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4285<h4>
4286 <a name="i_or">'<tt>or</tt>' Instruction</a>
4287</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4288
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4289<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4290
4291<h5>Syntax:</h5>
4292<pre>
4293 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
4294</pre>
4295
4296<h5>Overview:</h5>
4297<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
4298 two operands.</p>
4299
4300<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4301<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4302 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4303 values. Both arguments must have identical types.</p>
4304
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4305<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4306<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4307
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4308<table border="1" cellspacing="0" cellpadding="4">
4309 <tbody>
4310 <tr>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4311 <th>In0</th>
4312 <th>In1</th>
4313 <th>Out</th>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4314 </tr>
4315 <tr>
4316 <td>0</td>
4317 <td>0</td>
4318 <td>0</td>
4319 </tr>
4320 <tr>
4321 <td>0</td>
4322 <td>1</td>
4323 <td>1</td>
4324 </tr>
4325 <tr>
4326 <td>1</td>
4327 <td>0</td>
4328 <td>1</td>
4329 </tr>
4330 <tr>
4331 <td>1</td>
4332 <td>1</td>
4333 <td>1</td>
4334 </tr>
4335 </tbody>
4336</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4337
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4338<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4339<pre>
4340 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4341 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
4342 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4343</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4344
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4345</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4346
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4347<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4348<h4>
4349 <a name="i_xor">'<tt>xor</tt>' Instruction</a>
4350</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4351
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4352<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4353
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4354<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4355<pre>
4356 &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]4357</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4358
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4359<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4360<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
4361 its two operands. The <tt>xor</tt> is used to implement the "one's
4362 complement" operation, which is the "~" operator in C.</p>
4363
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4364<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4365<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4366 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
4367 values. Both arguments must have identical types.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]4368
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4369<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4370<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4371
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4372<table border="1" cellspacing="0" cellpadding="4">
4373 <tbody>
4374 <tr>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4375 <th>In0</th>
4376 <th>In1</th>
4377 <th>Out</th>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4378 </tr>
4379 <tr>
4380 <td>0</td>
4381 <td>0</td>
4382 <td>0</td>
4383 </tr>
4384 <tr>
4385 <td>0</td>
4386 <td>1</td>
4387 <td>1</td>
4388 </tr>
4389 <tr>
4390 <td>1</td>
4391 <td>0</td>
4392 <td>1</td>
4393 </tr>
4394 <tr>
4395 <td>1</td>
4396 <td>1</td>
4397 <td>0</td>
4398 </tr>
4399 </tbody>
4400</table>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4401
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4402<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4403<pre>
4404 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4405 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4406 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4407 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4408</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4409
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4410</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4411
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4412</div>
4413
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4414<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4415<h3>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4416 <a name="vectorops">Vector Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4417</h3>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4418
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4419<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4420
4421<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4422 target-independent manner. These instructions cover the element-access and
4423 vector-specific operations needed to process vectors effectively. While LLVM
4424 does directly support these vector operations, many sophisticated algorithms
4425 will want to use target-specific intrinsics to take full advantage of a
4426 specific target.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4427
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4428<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4429<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4430 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4431</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4432
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4433<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4434
4435<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4436<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4437 &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]4438</pre>
4439
4440<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4441<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4442 from a vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4443
4444
4445<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4446<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4447 of <a href="#t_vector">vector</a> type. The second operand is an index
4448 indicating the position from which to extract the element. The index may be
4449 a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4450
4451<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4452<p>The result is a scalar of the same type as the element type of
4453 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4454 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4455 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4456
4457<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4458<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4459 &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]4460</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4461
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4462</div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4463
4464<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4465<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4466 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4467</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4468
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4469<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4470
4471<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4472<pre>
Dan Gohmanf3480b92008-05-12 23:38:42 +0000[diff] [blame]4473 &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]4474</pre>
4475
4476<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4477<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4478 vector at a specified index.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4479
4480<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4481<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4482 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4483 whose type must equal the element type of the first operand. The third
4484 operand is an index indicating the position at which to insert the value.
4485 The index may be a variable.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4486
4487<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4488<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4489 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4490 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4491 results are undefined.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4492
4493<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4494<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4495 &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]4496</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4497
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4498</div>
4499
4500<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4501<h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4502 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4503</h4>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4504
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4505<div>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4506
4507<h5>Syntax:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4508<pre>
Mon P Wangaeb06d22008-11-10 04:46:22 +0000[diff] [blame]4509 &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]4510</pre>
4511
4512<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4513<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4514 from two input vectors, returning a vector with the same element type as the
4515 input and length that is the same as the shuffle mask.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4516
4517<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4518<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4519 with types that match each other. The third argument is a shuffle mask whose
4520 element type is always 'i32'. The result of the instruction is a vector
4521 whose length is the same as the shuffle mask and whose element type is the
4522 same as the element type of the first two operands.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4523
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4524<p>The shuffle mask operand is required to be a constant vector with either
4525 constant integer or undef values.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4526
4527<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4528<p>The elements of the two input vectors are numbered from left to right across
4529 both of the vectors. The shuffle mask operand specifies, for each element of
4530 the result vector, which element of the two input vectors the result element
4531 gets. The element selector may be undef (meaning "don't care") and the
4532 second operand may be undef if performing a shuffle from only one vector.</p>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4533
4534<h5>Example:</h5>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4535<pre>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]4536 &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]4537 &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]4538 &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]4539 &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]4540 &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]4541 &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]4542 &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]4543 &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]4544</pre>
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4545
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4546</div>
Tanya Lattner09474292006-04-14 19:24:33 +0000[diff] [blame]4547
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4548</div>
4549
Chris Lattner3df241e2006-04-08 23:07:04 +0000[diff] [blame]4550<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4551<h3>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4552 <a name="aggregateops">Aggregate Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4553</h3>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4554
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4555<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4556
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4557<p>LLVM supports several instructions for working with
4558 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4559
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4560<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4561<h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4562 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4563</h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4564
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4565<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4566
4567<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4568<pre>
4569 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4570</pre>
4571
4572<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4573<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4574 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4575
4576<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4577<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4578 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4579 <a href="#t_array">array</a> type. The operands are constant indices to
4580 specify which value to extract in a similar manner as indices in a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4581 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Frits van Bommel13242892010-12-05 20:54:38 +0000[diff] [blame]4582 <p>The major differences to <tt>getelementptr</tt> indexing are:</p>
4583 <ul>
4584 <li>Since the value being indexed is not a pointer, the first index is
4585 omitted and assumed to be zero.</li>
4586 <li>At least one index must be specified.</li>
4587 <li>Not only struct indices but also array indices must be in
4588 bounds.</li>
4589 </ul>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4590
4591<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4592<p>The result is the value at the position in the aggregate specified by the
4593 index operands.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4594
4595<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4596<pre>
Gabor Greifa5b6f452009-10-28 13:14:50 +0000[diff] [blame]4597 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4598</pre>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4599
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4600</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4601
4602<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4603<h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4604 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4605</h4>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4606
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4607<div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4608
4609<h5>Syntax:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4610<pre>
Bill Wendling194229e2011-07-26 20:42:28 +0000[diff] [blame]4611 &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]4612</pre>
4613
4614<h5>Overview:</h5>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4615<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4616 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4617
4618<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4619<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]4620 of <a href="#t_struct">struct</a> or
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]4621 <a href="#t_array">array</a> type. The second operand is a first-class
4622 value to insert. The following operands are constant indices indicating
4623 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]4624 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' instruction. The
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4625 value to insert must have the same type as the value identified by the
4626 indices.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4627
4628<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4629<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4630 that of <tt>val</tt> except that the value at the position specified by the
4631 indices is that of <tt>elt</tt>.</p>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4632
4633<h5>Example:</h5>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4634<pre>
Chris Lattner8645d1a2011-05-22 07:18:08 +0000[diff] [blame]4635 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4636 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
4637 %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]4638</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4639
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4640</div>
4641
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4642</div>
Dan Gohmana334d5f2008-05-12 23:51:09 +0000[diff] [blame]4643
4644<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4645<h3>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]4646 <a name="memoryops">Memory Access and Addressing Operations</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4647</h3>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4648
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4649<div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4650
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4651<p>A key design point of an SSA-based representation is how it represents
4652 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez2fee2942009-10-26 23:44:29 +0000[diff] [blame]4653 very simple. This section describes how to read, write, and allocate
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4654 memory in LLVM.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4655
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4656<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4657<h4>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4658 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4659</h4>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4660
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4661<div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4662
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4663<h5>Syntax:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4664<pre>
Dan Gohmanf75a7d32010-05-28 01:14:11 +0000[diff] [blame]4665 &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]4666</pre>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4667
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4668<h5>Overview:</h5>
Jeff Cohen6f1cc772007-04-22 01:17:39 +0000[diff] [blame]4669<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4670 currently executing function, to be automatically released when this function
4671 returns to its caller. The object is always allocated in the generic address
4672 space (address space zero).</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4673
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4674<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4675<p>The '<tt>alloca</tt>' instruction
4676 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4677 runtime stack, returning a pointer of the appropriate type to the program.
4678 If "NumElements" is specified, it is the number of elements allocated,
4679 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4680 specified, the value result of the allocation is guaranteed to be aligned to
4681 at least that boundary. If not specified, or if zero, the target can choose
4682 to align the allocation on any convenient boundary compatible with the
4683 type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4684
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4685<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4686
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4687<h5>Semantics:</h5>
Bill Wendling871eb0a2009-05-08 20:49:29 +0000[diff] [blame]4688<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4689 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4690 memory is automatically released when the function returns. The
4691 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4692 variables that must have an address available. When the function returns
4693 (either with the <tt><a href="#i_ret">ret</a></tt>
4694 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4695 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4696
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4697<h5>Example:</h5>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4698<pre>
Dan Gohman81e21672009-01-04 23:49:44 +0000[diff] [blame]4699 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4700 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4701 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4702 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4703</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4704
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4705</div>
Chris Lattner2cbdc452005-11-06 08:02:57 +0000[diff] [blame]4706
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]4707<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4708<h4>
4709 <a name="i_load">'<tt>load</tt>' Instruction</a>
4710</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4711
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4712<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4713
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4714<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4715<pre>
Eli Friedmanf03bb262011-08-12 22:50:01 +0000[diff] [blame]4716 &lt;result&gt; = load [volatile] &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4717 &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]4718 !&lt;index&gt; = !{ i32 1 }
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4719</pre>
4720
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4721<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4722<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4723
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4724<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4725<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4726 from which to load. The pointer must point to
4727 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4728 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]4729 number or order of execution of this <tt>load</tt> with other <a
4730 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4731
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]4732<p>If the <code>load</code> is marked as <code>atomic</code>, it takes an extra
4733 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4734 argument. The <code>release</code> and <code>acq_rel</code> orderings are
4735 not valid on <code>load</code> instructions. Atomic loads produce <a
4736 href="#memorymodel">defined</a> results when they may see multiple atomic
4737 stores. The type of the pointee must be an integer type whose bit width
4738 is a power of two greater than or equal to eight and less than or equal
4739 to a target-specific size limit. <code>align</code> must be explicitly
4740 specified on atomic loads, and the load has undefined behavior if the
4741 alignment is not set to a value which is at least the size in bytes of
4742 the pointee. <code>!nontemporal</code> does not have any defined semantics
4743 for atomic loads.</p>
4744
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4745<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4746 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]4747 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4748 alignment for the target. It is the responsibility of the code emitter to
4749 ensure that the alignment information is correct. Overestimating the
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4750 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4751 produce less efficient code. An alignment of 1 is always safe.</p>
4752
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4753<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4754 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4755 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendling7c78dbb2010-02-25 21:23:24 +0000[diff] [blame]4756 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4757 and code generator that this load is not expected to be reused in the cache.
4758 The code generator may select special instructions to save cache bandwidth,
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4759 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4760
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4761<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4762<p>The location of memory pointed to is loaded. If the value being loaded is of
4763 scalar type then the number of bytes read does not exceed the minimum number
4764 of bytes needed to hold all bits of the type. For example, loading an
4765 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4766 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4767 is undefined if the value was not originally written using a store of the
4768 same type.</p>
4769
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4770<h5>Examples:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4771<pre>
4772 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4773 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]4774 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4775</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4776
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4777</div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4778
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4779<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]4780<h4>
4781 <a name="i_store">'<tt>store</tt>' Instruction</a>
4782</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4783
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]4784<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4785
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4786<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4787<pre>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4788 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>
4789 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]4790</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4791
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4792<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]4793<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4794
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4795<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4796<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4797 and an address at which to store it. The type of the
4798 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4799 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]4800 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
4801 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4802 order of execution of this <tt>store</tt> with other <a
4803 href="#volatile">volatile operations</a>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4804
Eli Friedman21006d42011-08-09 23:02:53 +0000[diff] [blame]4805<p>If the <code>store</code> is marked as <code>atomic</code>, it takes an extra
4806 <a href="#ordering">ordering</a> and optional <code>singlethread</code>
4807 argument. The <code>acquire</code> and <code>acq_rel</code> orderings aren't
4808 valid on <code>store</code> instructions. Atomic loads produce <a
4809 href="#memorymodel">defined</a> results when they may see multiple atomic
4810 stores. The type of the pointee must be an integer type whose bit width
4811 is a power of two greater than or equal to eight and less than or equal
4812 to a target-specific size limit. <code>align</code> must be explicitly
4813 specified on atomic stores, and the store has undefined behavior if the
4814 alignment is not set to a value which is at least the size in bytes of
4815 the pointee. <code>!nontemporal</code> does not have any defined semantics
4816 for atomic stores.</p>
4817
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4818<p>The optional constant "align" argument specifies the alignment of the
4819 operation (that is, the alignment of the memory address). A value of 0 or an
4820 omitted "align" argument means that the operation has the preferential
4821 alignment for the target. It is the responsibility of the code emitter to
4822 ensure that the alignment information is correct. Overestimating the
4823 alignment results in an undefined behavior. Underestimating the alignment may
4824 produce less efficient code. An alignment of 1 is always safe.</p>
4825
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4826<p>The optional !nontemporal metadata must reference a single metatadata
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]4827 name &lt;index&gt; corresponding to a metadata node with one i32 entry of
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4828 value 1. The existence of the !nontemporal metatadata on the
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4829 instruction tells the optimizer and code generator that this load is
4830 not expected to be reused in the cache. The code generator may
4831 select special instructions to save cache bandwidth, such as the
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]4832 MOVNT instruction on x86.</p>
David Greene8939b0d2010-02-16 20:50:18 +0000[diff] [blame]4833
4834
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]4835<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4836<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4837 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4838 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4839 does not exceed the minimum number of bytes needed to hold all bits of the
4840 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4841 writing a value of a type like <tt>i20</tt> with a size that is not an
4842 integral number of bytes, it is unspecified what happens to the extra bits
4843 that do not belong to the type, but they will typically be overwritten.</p>
4844
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4845<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4846<pre>
4847 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8c6c72d2007-10-22 05:10:05 +0000[diff] [blame]4848 store i32 3, i32* %ptr <i>; yields {void}</i>
4849 %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]4850</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]4851
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]4852</div>
4853
Chris Lattner2b7d3202002-05-06 03:03:22 +0000[diff] [blame]4854<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]4855<h4>
4856<a name="i_fence">'<tt>fence</tt>' Instruction</a>
4857</h4>
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]4858
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]4859<div>
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]4860
4861<h5>Syntax:</h5>
4862<pre>
4863 fence [singlethread] &lt;ordering&gt; <i>; yields {void}</i>
4864</pre>
4865
4866<h5>Overview:</h5>
4867<p>The '<tt>fence</tt>' instruction is used to introduce happens-before edges
4868between operations.</p>
4869
4870<h5>Arguments:</h5> <p>'<code>fence</code>' instructions take an <a
4871href="#ordering">ordering</a> argument which defines what
4872<i>synchronizes-with</i> edges they add. They can only be given
4873<code>acquire</code>, <code>release</code>, <code>acq_rel</code>, and
4874<code>seq_cst</code> orderings.</p>
4875
4876<h5>Semantics:</h5>
4877<p>A fence <var>A</var> which has (at least) <code>release</code> ordering
4878semantics <i>synchronizes with</i> a fence <var>B</var> with (at least)
4879<code>acquire</code> ordering semantics if and only if there exist atomic
4880operations <var>X</var> and <var>Y</var>, both operating on some atomic object
4881<var>M</var>, such that <var>A</var> is sequenced before <var>X</var>,
4882<var>X</var> modifies <var>M</var> (either directly or through some side effect
4883of a sequence headed by <var>X</var>), <var>Y</var> is sequenced before
4884<var>B</var>, and <var>Y</var> observes <var>M</var>. This provides a
4885<i>happens-before</i> dependency between <var>A</var> and <var>B</var>. Rather
4886than an explicit <code>fence</code>, one (but not both) of the atomic operations
4887<var>X</var> or <var>Y</var> might provide a <code>release</code> or
4888<code>acquire</code> (resp.) ordering constraint and still
4889<i>synchronize-with</i> the explicit <code>fence</code> and establish the
4890<i>happens-before</i> edge.</p>
4891
4892<p>A <code>fence</code> which has <code>seq_cst</code> ordering, in addition to
4893having both <code>acquire</code> and <code>release</code> semantics specified
4894above, participates in the global program order of other <code>seq_cst</code>
4895operations and/or fences.</p>
4896
4897<p>The optional "<a href="#singlethread"><code>singlethread</code></a>" argument
4898specifies that the fence only synchronizes with other fences in the same
4899thread. (This is useful for interacting with signal handlers.)</p>
4900
Eli Friedman47f35132011-07-25 23:16:38 +0000[diff] [blame]4901<h5>Example:</h5>
4902<pre>
4903 fence acquire <i>; yields {void}</i>
4904 fence singlethread seq_cst <i>; yields {void}</i>
4905</pre>
4906
4907</div>
4908
4909<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]4910<h4>
4911<a name="i_cmpxchg">'<tt>cmpxchg</tt>' Instruction</a>
4912</h4>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4913
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]4914<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4915
4916<h5>Syntax:</h5>
4917<pre>
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4918 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]4919</pre>
4920
4921<h5>Overview:</h5>
4922<p>The '<tt>cmpxchg</tt>' instruction is used to atomically modify memory.
4923It loads a value in memory and compares it to a given value. If they are
4924equal, it stores a new value into the memory.</p>
4925
4926<h5>Arguments:</h5>
4927<p>There are three arguments to the '<code>cmpxchg</code>' instruction: an
4928address to operate on, a value to compare to the value currently be at that
4929address, and a new value to place at that address if the compared values are
4930equal. The type of '<var>&lt;cmp&gt;</var>' must be an integer type whose
4931bit width is a power of two greater than or equal to eight and less than
4932or equal to a target-specific size limit. '<var>&lt;cmp&gt;</var>' and
4933'<var>&lt;new&gt;</var>' must have the same type, and the type of
4934'<var>&lt;pointer&gt;</var>' must be a pointer to that type. If the
4935<code>cmpxchg</code> is marked as <code>volatile</code>, then the
4936optimizer is not allowed to modify the number or order of execution
4937of this <code>cmpxchg</code> with other <a href="#volatile">volatile
4938operations</a>.</p>
4939
4940<!-- FIXME: Extend allowed types. -->
4941
4942<p>The <a href="#ordering"><var>ordering</var></a> argument specifies how this
4943<code>cmpxchg</code> synchronizes with other atomic operations.</p>
4944
4945<p>The optional "<code>singlethread</code>" argument declares that the
4946<code>cmpxchg</code> is only atomic with respect to code (usually signal
4947handlers) running in the same thread as the <code>cmpxchg</code>. Otherwise the
4948cmpxchg is atomic with respect to all other code in the system.</p>
4949
4950<p>The pointer passed into cmpxchg must have alignment greater than or equal to
4951the size in memory of the operand.
4952
4953<h5>Semantics:</h5>
4954<p>The contents of memory at the location specified by the
4955'<tt>&lt;pointer&gt;</tt>' operand is read and compared to
4956'<tt>&lt;cmp&gt;</tt>'; if the read value is the equal,
4957'<tt>&lt;new&gt;</tt>' is written. The original value at the location
4958is returned.
4959
4960<p>A successful <code>cmpxchg</code> is a read-modify-write instruction for the
4961purpose of identifying <a href="#release_sequence">release sequences</a>. A
4962failed <code>cmpxchg</code> is equivalent to an atomic load with an ordering
4963parameter determined by dropping any <code>release</code> part of the
4964<code>cmpxchg</code>'s ordering.</p>
4965
4966<!--
4967FIXME: Is compare_exchange_weak() necessary? (Consider after we've done
4968optimization work on ARM.)
4969
4970FIXME: Is a weaker ordering constraint on failure helpful in practice?
4971-->
4972
4973<h5>Example:</h5>
4974<pre>
4975entry:
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4976 %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]4977 <a href="#i_br">br</a> label %loop
4978
4979loop:
4980 %cmp = <a href="#i_phi">phi</a> i32 [ %orig, %entry ], [%old, %loop]
4981 %squared = <a href="#i_mul">mul</a> i32 %cmp, %cmp
Bill Wendling262396b2011-12-09 22:41:40 +0000[diff] [blame]4982 %old = cmpxchg i32* %ptr, i32 %cmp, i32 %squared <i>; yields {i32}</i>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4983 %success = <a href="#i_icmp">icmp</a> eq i32 %cmp, %old
4984 <a href="#i_br">br</a> i1 %success, label %done, label %loop
4985
4986done:
4987 ...
4988</pre>
4989
4990</div>
4991
4992<!-- _______________________________________________________________________ -->
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]4993<h4>
4994<a name="i_atomicrmw">'<tt>atomicrmw</tt>' Instruction</a>
4995</h4>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4996
NAKAMURA Takumib5bb29c2011-08-12 06:17:17 +0000[diff] [blame]4997<div>
Eli Friedmanff030482011-07-28 21:48:00 +0000[diff] [blame]4998
4999<h5>Syntax:</h5>
5000<pre>
Eli Friedmanf03bb262011-08-12 22:50:01 +0000[diff] [blame]5001 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]5002</pre>
5003
5004<h5>Overview:</h5>
5005<p>The '<tt>atomicrmw</tt>' instruction is used to atomically modify memory.</p>
5006
5007<h5>Arguments:</h5>
5008<p>There are three arguments to the '<code>atomicrmw</code>' instruction: an
5009operation to apply, an address whose value to modify, an argument to the
5010operation. The operation must be one of the following keywords:</p>
5011<ul>
5012 <li>xchg</li>
5013 <li>add</li>
5014 <li>sub</li>
5015 <li>and</li>
5016 <li>nand</li>
5017 <li>or</li>
5018 <li>xor</li>
5019 <li>max</li>
5020 <li>min</li>
5021 <li>umax</li>
5022 <li>umin</li>
5023</ul>
5024
5025<p>The type of '<var>&lt;value&gt;</var>' must be an integer type whose
5026bit width is a power of two greater than or equal to eight and less than
5027or equal to a target-specific size limit. The type of the
5028'<code>&lt;pointer&gt;</code>' operand must be a pointer to that type.
5029If the <code>atomicrmw</code> is marked as <code>volatile</code>, then the
5030optimizer is not allowed to modify the number or order of execution of this
5031<code>atomicrmw</code> with other <a href="#volatile">volatile
5032 operations</a>.</p>
5033
5034<!-- FIXME: Extend allowed types. -->
5035
5036<h5>Semantics:</h5>
5037<p>The contents of memory at the location specified by the
5038'<tt>&lt;pointer&gt;</tt>' operand are atomically read, modified, and written
5039back. The original value at the location is returned. The modification is
5040specified by the <var>operation</var> argument:</p>
5041
5042<ul>
5043 <li>xchg: <code>*ptr = val</code></li>
5044 <li>add: <code>*ptr = *ptr + val</code></li>
5045 <li>sub: <code>*ptr = *ptr - val</code></li>
5046 <li>and: <code>*ptr = *ptr &amp; val</code></li>
5047 <li>nand: <code>*ptr = ~(*ptr &amp; val)</code></li>
5048 <li>or: <code>*ptr = *ptr | val</code></li>
5049 <li>xor: <code>*ptr = *ptr ^ val</code></li>
5050 <li>max: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using a signed comparison)</li>
5051 <li>min: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using a signed comparison)</li>
5052 <li>umax: <code>*ptr = *ptr &gt; val ? *ptr : val</code> (using an unsigned comparison)</li>
5053 <li>umin: <code>*ptr = *ptr &lt; val ? *ptr : val</code> (using an unsigned comparison)</li>
5054</ul>
5055
5056<h5>Example:</h5>
5057<pre>
5058 %old = atomicrmw add i32* %ptr, i32 1 acquire <i>; yields {i32}</i>
5059</pre>
5060
5061</div>
5062
5063<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5064<h4>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5065 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5066</h4>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5067
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5068<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5069
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5070<h5>Syntax:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5071<pre>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5072 &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]5073 &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]5074 &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]5075</pre>
5076
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5077<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5078<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5079 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
5080 It performs address calculation only and does not access memory.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5081
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5082<h5>Arguments:</h5>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5083<p>The first argument is always a pointer or a vector of pointers,
5084 and forms the basis of the
Chris Lattnerc8eef442009-07-29 06:44:13 +0000[diff] [blame]5085 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5086 elements of the aggregate object are indexed. The interpretation of each
5087 index is dependent on the type being indexed into. The first index always
5088 indexes the pointer value given as the first argument, the second index
5089 indexes a value of the type pointed to (not necessarily the value directly
5090 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5091 indexed into must be a pointer value, subsequent types can be arrays,
Chris Lattner61c70e92010-08-28 04:09:24 +0000[diff] [blame]5092 vectors, and structs. Note that subsequent types being indexed into
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5093 can never be pointers, since that would require loading the pointer before
5094 continuing calculation.</p>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5095
5096<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]5097 When indexing into a (optionally packed) structure, only <tt>i32</tt>
Chris Lattnerfdfeb692010-02-12 20:49:41 +0000[diff] [blame]5098 integer <b>constants</b> are allowed. When indexing into an array, pointer
5099 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]5100 constant. These integers are treated as signed values where relevant.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5101
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5102<p>For example, let's consider a C code fragment and how it gets compiled to
5103 LLVM:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5104
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]5105<pre class="doc_code">
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5106struct RT {
5107 char A;
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]5108 int B[10][20];
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5109 char C;
5110};
5111struct ST {
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]5112 int X;
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5113 double Y;
5114 struct RT Z;
5115};
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5116
Chris Lattnercabc8462007-05-29 15:43:56 +0000[diff] [blame]5117int *foo(struct ST *s) {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5118 return &amp;s[1].Z.B[5][13];
5119}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5120</pre>
5121
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5122<p>The LLVM code generated by Clang is:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5123
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]5124<pre class="doc_code">
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5125%struct.RT = <a href="#namedtypes">type</a> { i8, [10 x [20 x i32]], i8 }
5126%struct.ST = <a href="#namedtypes">type</a> { i32, double, %struct.RT }
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5127
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5128define i32* @foo(%struct.ST* %s) nounwind uwtable readnone optsize ssp {
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5129entry:
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5130 %arrayidx = getelementptr inbounds %struct.ST* %s, i64 1, i32 2, i32 1, i64 5, i64 13
5131 ret i32* %arrayidx
Bill Wendling2f7a8b02007-05-29 09:04:49 +0000[diff] [blame]5132}
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5133</pre>
5134
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5135<h5>Semantics:</h5>
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5136<p>In the example above, the first index is indexing into the
5137 '<tt>%struct.ST*</tt>' type, which is a pointer, yielding a
5138 '<tt>%struct.ST</tt>' = '<tt>{ i32, double, %struct.RT }</tt>' type, a
5139 structure. The second index indexes into the third element of the structure,
5140 yielding a '<tt>%struct.RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]], i8 }</tt>'
5141 type, another structure. The third index indexes into the second element of
5142 the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an array. The
5143 two dimensions of the array are subscripted into, yielding an '<tt>i32</tt>'
5144 type. The '<tt>getelementptr</tt>' instruction returns a pointer to this
5145 element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5146
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5147<p>Note that it is perfectly legal to index partially through a structure,
5148 returning a pointer to an inner element. Because of this, the LLVM code for
5149 the given testcase is equivalent to:</p>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5150
Bill Wendlinga3495392011-12-13 01:07:07 +0000[diff] [blame]5151<pre class="doc_code">
5152define i32* @foo(%struct.ST* %s) {
5153 %t1 = getelementptr %struct.ST* %s, i32 1 <i>; yields %struct.ST*:%t1</i>
5154 %t2 = getelementptr %struct.ST* %t1, i32 0, i32 2 <i>; yields %struct.RT*:%t2</i>
5155 %t3 = getelementptr %struct.RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
5156 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
5157 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
5158 ret i32* %t5
5159}
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5160</pre>
Chris Lattnere67a9512005-06-24 17:22:57 +0000[diff] [blame]5161
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5162<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohmanbfb056d2011-12-06 03:18:47 +0000[diff] [blame]5163 <tt>getelementptr</tt> is a <a href="#poisonvalues">poison value</a> if the
Dan Gohman27ef9972010-04-23 15:23:32 +0000[diff] [blame]5164 base pointer is not an <i>in bounds</i> address of an allocated object,
5165 or if any of the addresses that would be formed by successive addition of
5166 the offsets implied by the indices to the base address with infinitely
Eli Friedman266246c2011-08-12 23:37:55 +0000[diff] [blame]5167 precise signed arithmetic are not an <i>in bounds</i> address of that
5168 allocated object. The <i>in bounds</i> addresses for an allocated object
5169 are all the addresses that point into the object, plus the address one
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5170 byte past the end.
5171 In cases where the base is a vector of pointers the <tt>inbounds</tt> keyword
5172 applies to each of the computations element-wise. </p>
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5173
5174<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]5175 the base address with silently-wrapping two's complement arithmetic. If the
5176 offsets have a different width from the pointer, they are sign-extended or
5177 truncated to the width of the pointer. The result value of the
5178 <tt>getelementptr</tt> may be outside the object pointed to by the base
5179 pointer. The result value may not necessarily be used to access memory
5180 though, even if it happens to point into allocated storage. See the
5181 <a href="#pointeraliasing">Pointer Aliasing Rules</a> section for more
5182 information.</p>
Dan Gohmandd8004d2009-07-27 21:53:46 +0000[diff] [blame]5183
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5184<p>The getelementptr instruction is often confusing. For some more insight into
5185 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Chris Lattner884a9702006-08-15 00:45:58 +0000[diff] [blame]5186
Chris Lattner7faa8832002-04-14 06:13:44 +0000[diff] [blame]5187<h5>Example:</h5>
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5188<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5189 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijmane49d0bc2008-10-13 13:44:15 +0000[diff] [blame]5190 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
5191 <i>; yields i8*:vptr</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5192 %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]5193 <i>; yields i8*:eptr</i>
5194 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta9f805c22009-04-25 07:27:44 +0000[diff] [blame]5195 <i>; yields i32*:iptr</i>
Sanjiv Gupta16ffa802009-04-24 16:38:13 +0000[diff] [blame]5196 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5197</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5198
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5199<p>In cases where the pointer argument is a vector of pointers, only a
5200 single index may be used, and the number of vector elements has to be
5201 the same. For example: </p>
5202<pre class="doc_code">
5203 %A = getelementptr <4 x i8*> %ptrs, <4 x i64> %offsets,
5204</pre>
5205
Chris Lattnerf74d5c72004-04-05 01:30:49 +0000[diff] [blame]5206</div>
Reid Spencer47ce1792006-11-09 21:15:49 +0000[diff] [blame]5207
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5208</div>
5209
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]5210<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5211<h3>
5212 <a name="convertops">Conversion Operations</a>
5213</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5214
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5215<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5216
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5217<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5218 which all take a single operand and a type. They perform various bit
5219 conversions on the operand.</p>
5220
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5221<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5222<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5223 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5224</h4>
5225
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5226<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5227
5228<h5>Syntax:</h5>
5229<pre>
5230 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5231</pre>
5232
5233<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5234<p>The '<tt>trunc</tt>' instruction truncates its operand to the
5235 type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5236
5237<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5238<p>The '<tt>trunc</tt>' instruction takes a value to trunc, and a type to trunc it to.
5239 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5240 of the same number of integers.
5241 The bit size of the <tt>value</tt> must be larger than
5242 the bit size of the destination type, <tt>ty2</tt>.
5243 Equal sized types are not allowed.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5244
5245<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5246<p>The '<tt>trunc</tt>' instruction truncates the high order bits
5247 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
5248 source size must be larger than the destination size, <tt>trunc</tt> cannot
5249 be a <i>no-op cast</i>. It will always truncate bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5250
5251<h5>Example:</h5>
5252<pre>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5253 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
5254 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
5255 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
5256 %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]5257</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5258
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5259</div>
5260
5261<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5262<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5263 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5264</h4>
5265
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5266<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5267
5268<h5>Syntax:</h5>
5269<pre>
5270 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5271</pre>
5272
5273<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5274<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5275 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5276
5277
5278<h5>Arguments:</h5>
Nadav Rotemed9b9342011-02-20 12:37:50 +0000[diff] [blame]5279<p>The '<tt>zext</tt>' instruction takes a value to cast, and a type to cast it to.
5280 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5281 of the same number of integers.
5282 The bit size of the <tt>value</tt> must be smaller than
5283 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5284 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5285
5286<h5>Semantics:</h5>
5287<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]5288 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5289
Reid Spencerb5929522007-01-12 15:46:11 +0000[diff] [blame]5290<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]5291
5292<h5>Example:</h5>
5293<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5294 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5295 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Nadav Rotemed9b9342011-02-20 12:37:50 +0000[diff] [blame]5296 %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]5297</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5298
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5299</div>
5300
5301<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5302<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5303 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5304</h4>
5305
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5306<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5307
5308<h5>Syntax:</h5>
5309<pre>
5310 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5311</pre>
5312
5313<h5>Overview:</h5>
5314<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
5315
5316<h5>Arguments:</h5>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5317<p>The '<tt>sext</tt>' instruction takes a value to cast, and a type to cast it to.
5318 Both types must be of <a href="#t_integer">integer</a> types, or vectors
5319 of the same number of integers.
5320 The bit size of the <tt>value</tt> must be smaller than
5321 the bit size of the destination type,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5322 <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5323
5324<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5325<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
5326 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
5327 of the type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5328
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5329<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]5330
5331<h5>Example:</h5>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5332<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5333 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5334 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Nadav Rotem8c20ec52011-02-24 21:01:34 +0000[diff] [blame]5335 %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]5336</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5337
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5338</div>
5339
5340<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5341<h4>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5342 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5343</h4>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5344
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5345<div>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5346
5347<h5>Syntax:</h5>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5348<pre>
5349 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5350</pre>
5351
5352<h5>Overview:</h5>
5353<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5354 <tt>ty2</tt>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5355
5356<h5>Arguments:</h5>
5357<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5358 point</a> value to cast and a <a href="#t_floating">floating point</a> type
5359 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]5360 <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]5361 <i>no-op cast</i>.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5362
5363<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5364<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5365 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5366 <a href="#t_floating">floating point</a> type. If the value cannot fit
5367 within the destination type, <tt>ty2</tt>, then the results are
5368 undefined.</p>
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5369
5370<h5>Example:</h5>
5371<pre>
5372 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
5373 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
5374</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5375
Reid Spencer3fa91b02006-11-09 21:48:10 +0000[diff] [blame]5376</div>
5377
5378<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5379<h4>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5380 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5381</h4>
5382
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5383<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5384
5385<h5>Syntax:</h5>
5386<pre>
5387 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5388</pre>
5389
5390<h5>Overview:</h5>
5391<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]5392 floating point value.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5393
5394<h5>Arguments:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5395<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5396 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
5397 a <a href="#t_floating">floating point</a> type to cast it to. The source
5398 type must be smaller than the destination type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5399
5400<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5401<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5402 <a href="#t_floating">floating point</a> type to a larger
5403 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
5404 used to make a <i>no-op cast</i> because it always changes bits. Use
5405 <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]5406
5407<h5>Example:</h5>
5408<pre>
Nick Lewycky5bb3ece2011-03-31 18:20:19 +0000[diff] [blame]5409 %X = fpext float 3.125 to double <i>; yields double:3.125000e+00</i>
5410 %Y = fpext double %X to fp128 <i>; yields fp128:0xL00000000000000004000900000000000</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5411</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5412
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5413</div>
5414
5415<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5416<h4>
Reid Spencer24d6da52007-01-21 00:29:26 +0000[diff] [blame]5417 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5418</h4>
5419
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5420<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5421
5422<h5>Syntax:</h5>
5423<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5424 &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]5425</pre>
5426
5427<h5>Overview:</h5>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5428<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5429 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5430
5431<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5432<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
5433 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5434 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5435 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5436 vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5437
5438<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5439<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5440 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5441 towards zero) unsigned integer value. If the value cannot fit
5442 in <tt>ty2</tt>, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5443
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5444<h5>Example:</h5>
5445<pre>
Reid Spencer1539a1c2007-07-31 14:40:14 +0000[diff] [blame]5446 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner88519042007-09-22 03:17:52 +0000[diff] [blame]5447 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5448 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5449</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5450
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5451</div>
5452
5453<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5454<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5455 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5456</h4>
5457
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5458<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5459
5460<h5>Syntax:</h5>
5461<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5462 &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]5463</pre>
5464
5465<h5>Overview:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5466<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5467 <a href="#t_floating">floating point</a> <tt>value</tt> to
5468 type <tt>ty2</tt>.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5469
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5470<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5471<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
5472 scalar or vector <a href="#t_floating">floating point</a> value, and a type
5473 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
5474 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
5475 vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5476
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]5477<h5>Semantics:</h5>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5478<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5479 <a href="#t_floating">floating point</a> operand into the nearest (rounding
5480 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
5481 the results are undefined.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5482
Chris Lattner33ba0d92001-07-09 00:26:23 +0000[diff] [blame]5483<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5484<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]5485 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner88519042007-09-22 03:17:52 +0000[diff] [blame]5486 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif6a292012009-10-28 09:21:30 +0000[diff] [blame]5487 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5488</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5489
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5490</div>
5491
5492<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5493<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5494 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5495</h4>
5496
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5497<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5498
5499<h5>Syntax:</h5>
5500<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5501 &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]5502</pre>
5503
5504<h5>Overview:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5505<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5506 integer and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5507
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5508<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +0000[diff] [blame]5509<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]5510 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5511 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5512 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5513 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5514
5515<h5>Semantics:</h5>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5516<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5517 integer quantity and converts it to the corresponding floating point
5518 value. If the value cannot fit in the floating point value, the results are
5519 undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5520
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5521<h5>Example:</h5>
5522<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5523 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5524 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5525</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5526
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5527</div>
5528
5529<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5530<h4>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5531 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5532</h4>
5533
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5534<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5535
5536<h5>Syntax:</h5>
5537<pre>
Reid Spencerd4448792006-11-09 23:03:26 +0000[diff] [blame]5538 &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]5539</pre>
5540
5541<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5542<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
5543 and converts that value to the <tt>ty2</tt> type.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5544
5545<h5>Arguments:</h5>
Nate Begemanb348d182007-11-17 03:58:34 +0000[diff] [blame]5546<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]5547 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
5548 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
5549 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
5550 floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5551
5552<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5553<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
5554 quantity and converts it to the corresponding floating point value. If the
5555 value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5556
5557<h5>Example:</h5>
5558<pre>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5559 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5560 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5561</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5562
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5563</div>
5564
5565<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5566<h4>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5567 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5568</h4>
5569
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5570<div>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5571
5572<h5>Syntax:</h5>
5573<pre>
5574 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5575</pre>
5576
5577<h5>Overview:</h5>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5578<p>The '<tt>ptrtoint</tt>' instruction converts the pointer or a vector of
5579 pointers <tt>value</tt> to
5580 the integer (or vector of integers) type <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5581
5582<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5583<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5584 must be a a value of type <a href="#t_pointer">pointer</a> or a vector of
5585 pointers, and a type to cast it to
5586 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> or a vector
5587 of integers type.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5588
5589<h5>Semantics:</h5>
5590<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5591 <tt>ty2</tt> by interpreting the pointer value as an integer and either
5592 truncating or zero extending that value to the size of the integer type. If
5593 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
5594 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
5595 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
5596 change.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5597
5598<h5>Example:</h5>
5599<pre>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5600 %X = ptrtoint i32* %P to i8 <i>; yields truncation on 32-bit architecture</i>
5601 %Y = ptrtoint i32* %P to i64 <i>; yields zero extension on 32-bit architecture</i>
5602 %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]5603</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5604
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5605</div>
5606
5607<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5608<h4>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5609 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5610</h4>
5611
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5612<div>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5613
5614<h5>Syntax:</h5>
5615<pre>
5616 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
5617</pre>
5618
5619<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5620<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
5621 pointer type, <tt>ty2</tt>.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5622
5623<h5>Arguments:</h5>
Duncan Sands8036ca42007-03-30 12:22:09 +0000[diff] [blame]5624<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5625 value to cast, and a type to cast it to, which must be a
5626 <a href="#t_pointer">pointer</a> type.</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5627
5628<h5>Semantics:</h5>
5629<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5630 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
5631 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
5632 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
5633 than the size of a pointer then a zero extension is done. If they are the
5634 same size, nothing is done (<i>no-op cast</i>).</p>
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5635
5636<h5>Example:</h5>
5637<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5638 %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]5639 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
5640 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5641 %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]5642</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5643
Reid Spencer72679252006-11-11 21:00:47 +0000[diff] [blame]5644</div>
5645
5646<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5647<h4>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5648 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5649</h4>
5650
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5651<div>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5652
5653<h5>Syntax:</h5>
5654<pre>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5655 &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]5656</pre>
5657
5658<h5>Overview:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5659<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5660 <tt>ty2</tt> without changing any bits.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5661
5662<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5663<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
5664 non-aggregate first class value, and a type to cast it to, which must also be
5665 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
5666 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
5667 identical. If the source type is a pointer, the destination type must also be
5668 a pointer. This instruction supports bitwise conversion of vectors to
5669 integers and to vectors of other types (as long as they have the same
5670 size).</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5671
5672<h5>Semantics:</h5>
Reid Spencer5c0ef472006-11-11 23:08:07 +0000[diff] [blame]5673<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5674 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
5675 this conversion. The conversion is done as if the <tt>value</tt> had been
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5676 stored to memory and read back as type <tt>ty2</tt>.
5677 Pointer (or vector of pointers) types may only be converted to other pointer
5678 (or vector of pointers) types with this instruction. To convert
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5679 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5680 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer9dee3ac2006-11-08 01:11:31 +0000[diff] [blame]5681
5682<h5>Example:</h5>
5683<pre>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5684 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5685 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5686 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
5687 %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]5688</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5689
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]5690</div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5691
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5692</div>
5693
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5694<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5695<h3>
5696 <a name="otherops">Other Operations</a>
5697</h3>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5698
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5699<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5700
5701<p>The instructions in this category are the "miscellaneous" instructions, which
5702 defy better classification.</p>
5703
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5704<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5705<h4>
5706 <a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5707</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5708
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5709<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5710
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5711<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5712<pre>
5713 &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]5714</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5715
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5716<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5717<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
Nadav Rotem16087692011-12-05 06:29:09 +0000[diff] [blame]5718 boolean values based on comparison of its two integer, integer vector,
5719 pointer, or pointer vector operands.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5720
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5721<h5>Arguments:</h5>
5722<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5723 the condition code indicating the kind of comparison to perform. It is not a
5724 value, just a keyword. The possible condition code are:</p>
5725
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5726<ol>
5727 <li><tt>eq</tt>: equal</li>
5728 <li><tt>ne</tt>: not equal </li>
5729 <li><tt>ugt</tt>: unsigned greater than</li>
5730 <li><tt>uge</tt>: unsigned greater or equal</li>
5731 <li><tt>ult</tt>: unsigned less than</li>
5732 <li><tt>ule</tt>: unsigned less or equal</li>
5733 <li><tt>sgt</tt>: signed greater than</li>
5734 <li><tt>sge</tt>: signed greater or equal</li>
5735 <li><tt>slt</tt>: signed less than</li>
5736 <li><tt>sle</tt>: signed less or equal</li>
5737</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5738
Chris Lattner3b19d652007-01-15 01:54:13 +0000[diff] [blame]5739<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5740 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5741 typed. They must also be identical types.</p>
5742
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5743<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5744<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5745 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]5746 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]5747 result, as follows:</p>
5748
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5749<ol>
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5750 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5751 <tt>false</tt> otherwise. No sign interpretation is necessary or
5752 performed.</li>
5753
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5754 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5755 <tt>false</tt> otherwise. No sign interpretation is necessary or
5756 performed.</li>
5757
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5758 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5759 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5760
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5761 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5762 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5763 to <tt>op2</tt>.</li>
5764
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5765 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5766 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5767
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5768 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5769 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5770
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5771 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5772 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5773
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5774 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5775 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5776 to <tt>op2</tt>.</li>
5777
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5778 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5779 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5780
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5781 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5782 <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]5783</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5784
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5785<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5786 values are compared as if they were integers.</p>
5787
5788<p>If the operands are integer vectors, then they are compared element by
5789 element. The result is an <tt>i1</tt> vector with the same number of elements
5790 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5791
5792<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5793<pre>
5794 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]5795 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
5796 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
5797 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5798 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5799 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5800</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]5801
5802<p>Note that the code generator does not yet support vector types with
5803 the <tt>icmp</tt> instruction.</p>
5804
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5805</div>
5806
5807<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5808<h4>
5809 <a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5810</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5811
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5812<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5813
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5814<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5815<pre>
5816 &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]5817</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5818
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5819<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5820<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5821 values based on comparison of its operands.</p>
5822
5823<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]5824(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5825
5826<p>If the operands are floating point vectors, then the result type is a vector
5827 of boolean with the same number of elements as the operands being
5828 compared.</p>
5829
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5830<h5>Arguments:</h5>
5831<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5832 the condition code indicating the kind of comparison to perform. It is not a
5833 value, just a keyword. The possible condition code are:</p>
5834
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5835<ol>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5836 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5837 <li><tt>oeq</tt>: ordered and equal</li>
5838 <li><tt>ogt</tt>: ordered and greater than </li>
5839 <li><tt>oge</tt>: ordered and greater than or equal</li>
5840 <li><tt>olt</tt>: ordered and less than </li>
5841 <li><tt>ole</tt>: ordered and less than or equal</li>
5842 <li><tt>one</tt>: ordered and not equal</li>
5843 <li><tt>ord</tt>: ordered (no nans)</li>
5844 <li><tt>ueq</tt>: unordered or equal</li>
5845 <li><tt>ugt</tt>: unordered or greater than </li>
5846 <li><tt>uge</tt>: unordered or greater than or equal</li>
5847 <li><tt>ult</tt>: unordered or less than </li>
5848 <li><tt>ule</tt>: unordered or less than or equal</li>
5849 <li><tt>une</tt>: unordered or not equal</li>
5850 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5851 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5852</ol>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5853
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5854<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5855 <i>unordered</i> means that either operand may be a QNAN.</p>
5856
5857<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5858 a <a href="#t_floating">floating point</a> type or
5859 a <a href="#t_vector">vector</a> of floating point type. They must have
5860 identical types.</p>
5861
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5862<h5>Semantics:</h5>
Gabor Greiffb224a22008-08-07 21:46:00 +0000[diff] [blame]5863<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5864 according to the condition code given as <tt>cond</tt>. If the operands are
5865 vectors, then the vectors are compared element by element. Each comparison
Nick Lewyckyec38da42009-09-27 00:45:11 +0000[diff] [blame]5866 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5867 follows:</p>
5868
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5869<ol>
5870 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5871
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5872 <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]5873 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5874
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5875 <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]5876 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5877
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5878 <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]5879 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5880
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5881 <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]5882 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5883
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5884 <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]5885 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5886
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5887 <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]5888 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5889
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5890 <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]5891
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5892 <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]5893 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5894
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5895 <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]5896 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5897
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5898 <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]5899 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5900
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5901 <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]5902 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5903
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5904 <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]5905 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5906
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]5907 <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]5908 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5909
Reid Spencerb7f26282006-11-19 03:00:14 +0000[diff] [blame]5910 <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]5911
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5912 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5913</ol>
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5914
5915<h5>Example:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5916<pre>
5917 &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]5918 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5919 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5920 &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]5921</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]5922
5923<p>Note that the code generator does not yet support vector types with
5924 the <tt>fcmp</tt> instruction.</p>
5925
Reid Spencerf3a70a62006-11-18 21:50:54 +0000[diff] [blame]5926</div>
5927
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5928<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5929<h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5930 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5931</h4>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5932
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5933<div>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5934
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5935<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5936<pre>
5937 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
5938</pre>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5939
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5940<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5941<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
5942 SSA graph representing the function.</p>
5943
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5944<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5945<p>The type of the incoming values is specified with the first type field. After
5946 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5947 one pair for each predecessor basic block of the current block. Only values
5948 of <a href="#t_firstclass">first class</a> type may be used as the value
5949 arguments to the PHI node. Only labels may be used as the label
5950 arguments.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5951
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5952<p>There must be no non-phi instructions between the start of a basic block and
5953 the PHI instructions: i.e. PHI instructions must be first in a basic
5954 block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5955
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5956<p>For the purposes of the SSA form, the use of each incoming value is deemed to
5957 occur on the edge from the corresponding predecessor block to the current
5958 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5959 value on the same edge).</p>
Jay Foadd2449092009-06-03 10:20:10 +0000[diff] [blame]5960
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5961<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]5962<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5963 specified by the pair corresponding to the predecessor basic block that
5964 executed just prior to the current block.</p>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5965
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5966<h5>Example:</h5>
Chris Lattner5568e942008-05-20 20:48:21 +0000[diff] [blame]5967<pre>
5968Loop: ; Infinite loop that counts from 0 on up...
5969 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5970 %nextindvar = add i32 %indvar, 1
5971 br label %Loop
5972</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5973
Reid Spencer2fd21e62006-11-08 01:18:52 +0000[diff] [blame]5974</div>
5975
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5976<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5977<h4>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5978 <a name="i_select">'<tt>select</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]5979</h4>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5980
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]5981<div>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5982
5983<h5>Syntax:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5984<pre>
Dan Gohmanf72fb672008-09-09 01:02:47 +0000[diff] [blame]5985 &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>
5986
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]5987 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5988</pre>
5989
5990<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5991<p>The '<tt>select</tt>' instruction is used to choose one value based on a
5992 condition, without branching.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]5993
5994
5995<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]5996<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5997 values indicating the condition, and two values of the
5998 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5999 vectors and the condition is a scalar, then entire vectors are selected, not
6000 individual elements.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6001
6002<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6003<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
6004 first value argument; otherwise, it returns the second value argument.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6005
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6006<p>If the condition is a vector of i1, then the value arguments must be vectors
6007 of the same size, and the selection is done element by element.</p>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6008
6009<h5>Example:</h5>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6010<pre>
Reid Spencerc78f3372007-01-12 03:35:51 +0000[diff] [blame]6011 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6012</pre>
Dan Gohmanc766f722009-01-22 01:39:38 +0000[diff] [blame]6013
6014<p>Note that the code generator does not yet support conditions
6015 with vector type.</p>
6016
Chris Lattnercc37aae2004-03-12 05:50:16 +0000[diff] [blame]6017</div>
6018
Robert Bocchino05ccd702006-01-15 20:48:27 +0000[diff] [blame]6019<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6020<h4>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6021 <a name="i_call">'<tt>call</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6022</h4>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6023
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6024<div>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6025
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6026<h5>Syntax:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6027<pre>
Devang Patel307e8ab2008-10-07 17:48:33 +0000[diff] [blame]6028 &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]6029</pre>
6030
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6031<h5>Overview:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6032<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6033
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6034<h5>Arguments:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6035<p>This instruction requires several arguments:</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6036
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]6037<ol>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6038 <li>The optional "tail" marker indicates that the callee function does not
6039 access any allocas or varargs in the caller. Note that calls may be
6040 marked "tail" even if they do not occur before
6041 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
6042 present, the function call is eligible for tail call optimization,
6043 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Chengdc444e92010-03-08 21:05:02 +0000[diff] [blame]6044 optimized into a jump</a>. The code generator may optimize calls marked
6045 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
6046 sibling call optimization</a> when the caller and callee have
6047 matching signatures, or 2) forced tail call optimization when the
6048 following extra requirements are met:
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6049 <ul>
6050 <li>Caller and callee both have the calling
6051 convention <tt>fastcc</tt>.</li>
6052 <li>The call is in tail position (ret immediately follows call and ret
6053 uses value of call or is void).</li>
6054 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohmanfbbee8d2010-03-02 01:08:11 +0000[diff] [blame]6055 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6056 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
6057 constraints are met.</a></li>
6058 </ul>
6059 </li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]6060
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6061 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
6062 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin95fa80a2010-01-09 19:44:16 +0000[diff] [blame]6063 defaults to using C calling conventions. The calling convention of the
6064 call must match the calling convention of the target function, or else the
6065 behavior is undefined.</li>
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]6066
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6067 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
6068 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
6069 '<tt>inreg</tt>' attributes are valid here.</li>
6070
6071 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
6072 type of the return value. Functions that return no value are marked
6073 <tt><a href="#t_void">void</a></tt>.</li>
6074
6075 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
6076 being invoked. The argument types must match the types implied by this
6077 signature. This type can be omitted if the function is not varargs and if
6078 the function type does not return a pointer to a function.</li>
6079
6080 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
6081 be invoked. In most cases, this is a direct function invocation, but
6082 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
6083 to function value.</li>
6084
6085 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner0724fbd2010-03-02 06:36:51 +0000[diff] [blame]6086 signature argument types and parameter attributes. All arguments must be
6087 of <a href="#t_firstclass">first class</a> type. If the function
6088 signature indicates the function accepts a variable number of arguments,
6089 the extra arguments can be specified.</li>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6090
6091 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
6092 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
6093 '<tt>readnone</tt>' attributes are valid here.</li>
Chris Lattner6536cfe2002-05-06 22:08:29 +0000[diff] [blame]6094</ol>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6095
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6096<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6097<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
6098 a specified function, with its incoming arguments bound to the specified
6099 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
6100 function, control flow continues with the instruction after the function
6101 call, and the return value of the function is bound to the result
6102 argument.</p>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6103
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]6104<h5>Example:</h5>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6105<pre>
Nick Lewyckydb7e3c92007-09-08 13:57:50 +0000[diff] [blame]6106 %retval = call i32 @test(i32 %argc)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6107 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattner772fccf2008-03-21 17:24:17 +0000[diff] [blame]6108 %X = tail call i32 @foo() <i>; yields i32</i>
6109 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
6110 call void %foo(i8 97 signext)
Devang Patelc3fc6df2008-03-10 20:49:15 +0000[diff] [blame]6111
6112 %struct.A = type { i32, i8 }
Devang Patelf642f472008-10-06 18:50:38 +0000[diff] [blame]6113 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohmanb1e6b962008-10-04 19:00:07 +0000[diff] [blame]6114 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
6115 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattner85a350f2008-10-08 06:26:11 +0000[diff] [blame]6116 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijmancb73d192008-10-07 10:03:45 +0000[diff] [blame]6117 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Chris Lattner2bff5242005-05-06 05:47:36 +0000[diff] [blame]6118</pre>
6119
Dale Johannesen07de8d12009-09-24 18:38:21 +0000[diff] [blame]6120<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen9f8380b2009-09-25 17:04:42 +0000[diff] [blame]6121standard C99 library as being the C99 library functions, and may perform
6122optimizations or generate code for them under that assumption. This is
6123something we'd like to change in the future to provide better support for
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6124freestanding environments and non-C-based languages.</p>
Dale Johannesen07de8d12009-09-24 18:38:21 +0000[diff] [blame]6125
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6126</div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6127
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6128<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6129<h4>
Chris Lattnerfb6977d2006-01-13 23:26:01 +0000[diff] [blame]6130 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6131</h4>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6132
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6133<div>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6134
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6135<h5>Syntax:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6136<pre>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6137 &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]6138</pre>
6139
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6140<h5>Overview:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6141<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6142 the "variable argument" area of a function call. It is used to implement the
6143 <tt>va_arg</tt> macro in C.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6144
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6145<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6146<p>This instruction takes a <tt>va_list*</tt> value and the type of the
6147 argument. It returns a value of the specified argument type and increments
6148 the <tt>va_list</tt> to point to the next argument. The actual type
6149 of <tt>va_list</tt> is target specific.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6150
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6151<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6152<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
6153 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
6154 to the next argument. For more information, see the variable argument
6155 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6156
6157<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]6158 take a variable number of arguments, for example, the <tt>vfprintf</tt>
6159 function.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6160
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6161<p><tt>va_arg</tt> is an LLVM instruction instead of
6162 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
6163 argument.</p>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6164
Chris Lattner8d1a81d2003-10-18 05:51:36 +0000[diff] [blame]6165<h5>Example:</h5>
Chris Lattnere19d7a72004-09-27 21:51:25 +0000[diff] [blame]6166<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
6167
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6168<p>Note that the code generator does not yet fully support va_arg on many
6169 targets. Also, it does not currently support va_arg with aggregate types on
6170 any target.</p>
Dan Gohmanf3e60bd2009-01-12 23:12:39 +0000[diff] [blame]6171
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6172</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6173
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6174<!-- _______________________________________________________________________ -->
6175<h4>
6176 <a name="i_landingpad">'<tt>landingpad</tt>' Instruction</a>
6177</h4>
6178
6179<div>
6180
6181<h5>Syntax:</h5>
6182<pre>
Bill Wendlingbf13ee12011-08-08 08:06:05 +0000[diff] [blame]6183 &lt;resultval&gt; = landingpad &lt;somety&gt; personality &lt;type&gt; &lt;pers_fn&gt; &lt;clause&gt;+
6184 &lt;resultval&gt; = landingpad &lt;somety&gt; personality &lt;type&gt; &lt;pers_fn&gt; cleanup &lt;clause&gt;*
6185
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6186 &lt;clause&gt; := catch &lt;type&gt; &lt;value&gt;
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6187 &lt;clause&gt; := filter &lt;array constant type&gt; &lt;array constant&gt;
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6188</pre>
6189
6190<h5>Overview:</h5>
6191<p>The '<tt>landingpad</tt>' instruction is used by
6192 <a href="ExceptionHandling.html#overview">LLVM's exception handling
6193 system</a> to specify that a basic block is a landing pad &mdash; one where
6194 the exception lands, and corresponds to the code found in the
6195 <i><tt>catch</tt></i> portion of a <i><tt>try/catch</tt></i> sequence. It
6196 defines values supplied by the personality function (<tt>pers_fn</tt>) upon
6197 re-entry to the function. The <tt>resultval</tt> has the
6198 type <tt>somety</tt>.</p>
6199
6200<h5>Arguments:</h5>
6201<p>This instruction takes a <tt>pers_fn</tt> value. This is the personality
6202 function associated with the unwinding mechanism. The optional
6203 <tt>cleanup</tt> flag indicates that the landing pad block is a cleanup.</p>
6204
6205<p>A <tt>clause</tt> begins with the clause type &mdash; <tt>catch</tt>
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6206 or <tt>filter</tt> &mdash; and contains the global variable representing the
6207 "type" that may be caught or filtered respectively. Unlike the
6208 <tt>catch</tt> clause, the <tt>filter</tt> clause takes an array constant as
6209 its argument. Use "<tt>[0 x i8**] undef</tt>" for a filter which cannot
6210 throw. The '<tt>landingpad</tt>' instruction must contain <em>at least</em>
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6211 one <tt>clause</tt> or the <tt>cleanup</tt> flag.</p>
6212
6213<h5>Semantics:</h5>
6214<p>The '<tt>landingpad</tt>' instruction defines the values which are set by the
6215 personality function (<tt>pers_fn</tt>) upon re-entry to the function, and
6216 therefore the "result type" of the <tt>landingpad</tt> instruction. As with
6217 calling conventions, how the personality function results are represented in
6218 LLVM IR is target specific.</p>
6219
Bill Wendlingb7a01352011-08-03 17:17:06 +0000[diff] [blame]6220<p>The clauses are applied in order from top to bottom. If two
6221 <tt>landingpad</tt> instructions are merged together through inlining, the
Bill Wendling2905c322011-08-08 07:58:58 +0000[diff] [blame]6222 clauses from the calling function are appended to the list of clauses.</p>
Bill Wendlingb7a01352011-08-03 17:17:06 +0000[diff] [blame]6223
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6224<p>The <tt>landingpad</tt> instruction has several restrictions:</p>
6225
6226<ul>
6227 <li>A landing pad block is a basic block which is the unwind destination of an
6228 '<tt>invoke</tt>' instruction.</li>
6229 <li>A landing pad block must have a '<tt>landingpad</tt>' instruction as its
6230 first non-PHI instruction.</li>
6231 <li>There can be only one '<tt>landingpad</tt>' instruction within the landing
6232 pad block.</li>
6233 <li>A basic block that is not a landing pad block may not include a
6234 '<tt>landingpad</tt>' instruction.</li>
6235 <li>All '<tt>landingpad</tt>' instructions in a function must have the same
6236 personality function.</li>
6237</ul>
6238
6239<h5>Example:</h5>
6240<pre>
6241 ;; A landing pad which can catch an integer.
6242 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6243 catch i8** @_ZTIi
6244 ;; A landing pad that is a cleanup.
6245 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6246 cleanup
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6247 ;; A landing pad which can catch an integer and can only throw a double.
6248 %res = landingpad { i8*, i32 } personality i32 (...)* @__gxx_personality_v0
6249 catch i8** @_ZTIi
Bill Wendlinge6e88262011-08-12 20:24:12 +0000[diff] [blame]6250 filter [1 x i8**] [@_ZTId]
Bill Wendlingf78faf82011-08-02 21:52:38 +0000[diff] [blame]6251</pre>
6252
6253</div>
6254
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6255</div>
6256
6257</div>
6258
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6259<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6260<h2><a name="intrinsics">Intrinsic Functions</a></h2>
Chris Lattner261efe92003-11-25 01:02:51 +0000[diff] [blame]6261<!-- *********************************************************************** -->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6262
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6263<div>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6264
6265<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6266 well known names and semantics and are required to follow certain
6267 restrictions. Overall, these intrinsics represent an extension mechanism for
6268 the LLVM language that does not require changing all of the transformations
6269 in LLVM when adding to the language (or the bitcode reader/writer, the
6270 parser, etc...).</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6271
John Criswellfc6b8952005-05-16 16:17:45 +0000[diff] [blame]6272<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6273 prefix is reserved in LLVM for intrinsic names; thus, function names may not
6274 begin with this prefix. Intrinsic functions must always be external
6275 functions: you cannot define the body of intrinsic functions. Intrinsic
6276 functions may only be used in call or invoke instructions: it is illegal to
6277 take the address of an intrinsic function. Additionally, because intrinsic
6278 functions are part of the LLVM language, it is required if any are added that
6279 they be documented here.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6280
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6281<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
6282 family of functions that perform the same operation but on different data
6283 types. Because LLVM can represent over 8 million different integer types,
6284 overloading is used commonly to allow an intrinsic function to operate on any
6285 integer type. One or more of the argument types or the result type can be
6286 overloaded to accept any integer type. Argument types may also be defined as
6287 exactly matching a previous argument's type or the result type. This allows
6288 an intrinsic function which accepts multiple arguments, but needs all of them
6289 to be of the same type, to only be overloaded with respect to a single
6290 argument or the result.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6291
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6292<p>Overloaded intrinsics will have the names of its overloaded argument types
6293 encoded into its function name, each preceded by a period. Only those types
6294 which are overloaded result in a name suffix. Arguments whose type is matched
6295 against another type do not. For example, the <tt>llvm.ctpop</tt> function
6296 can take an integer of any width and returns an integer of exactly the same
6297 integer width. This leads to a family of functions such as
6298 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
6299 %val)</tt>. Only one type, the return type, is overloaded, and only one type
6300 suffix is required. Because the argument's type is matched against the return
6301 type, it does not require its own name suffix.</p>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]6302
Eric Christopher6c7e8a02009-12-05 02:46:03 +0000[diff] [blame]6303<p>To learn how to add an intrinsic function, please see the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6304 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6305
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6306<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6307<h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6308 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6309</h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6310
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6311<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6312
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6313<p>Variable argument support is defined in LLVM with
6314 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
6315 intrinsic functions. These functions are related to the similarly named
6316 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6317
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6318<p>All of these functions operate on arguments that use a target-specific value
6319 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
6320 not define what this type is, so all transformations should be prepared to
6321 handle these functions regardless of the type used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6322
Chris Lattner374ab302006-05-15 17:26:46 +0000[diff] [blame]6323<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]6324 instruction and the variable argument handling intrinsic functions are
6325 used.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6326
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]6327<pre class="doc_code">
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6328define i32 @test(i32 %X, ...) {
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6329 ; Initialize variable argument processing
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6330 %ap = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6331 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6332 call void @llvm.va_start(i8* %ap2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6333
6334 ; Read a single integer argument
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6335 %tmp = va_arg i8** %ap, i32
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6336
6337 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6338 %aq = alloca i8*
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6339 %aq2 = bitcast i8** %aq to i8*
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6340 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6341 call void @llvm.va_end(i8* %aq2)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6342
6343 ; Stop processing of arguments.
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6344 call void @llvm.va_end(i8* %ap2)
Reid Spencerca86e162006-12-31 07:07:53 +0000[diff] [blame]6345 ret i32 %tmp
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6346}
Anton Korobeynikov5d522f32007-03-21 23:58:04 +0000[diff] [blame]6347
6348declare void @llvm.va_start(i8*)
6349declare void @llvm.va_copy(i8*, i8*)
6350declare void @llvm.va_end(i8*)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6351</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6352
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6353<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6354<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6355 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6356</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6357
6358
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6359<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6360
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6361<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6362<pre>
6363 declare void %llvm.va_start(i8* &lt;arglist&gt;)
6364</pre>
6365
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6366<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6367<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
6368 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]6369
6370<h5>Arguments:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6371<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]6372
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6373<h5>Semantics:</h5>
Dan Gohman0e451ce2008-10-14 16:51:45 +0000[diff] [blame]6374<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]6375 macro available in C. In a target-dependent way, it initializes
6376 the <tt>va_list</tt> element to which the argument points, so that the next
6377 call to <tt>va_arg</tt> will produce the first variable argument passed to
6378 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
6379 need to know the last argument of the function as the compiler can figure
6380 that out.</p>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6381
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6382</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6383
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6384<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6385<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6386 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6387</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6388
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6389<div>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6390
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6391<h5>Syntax:</h5>
6392<pre>
6393 declare void @llvm.va_end(i8* &lt;arglist&gt;)
6394</pre>
6395
6396<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6397<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6398 which has been initialized previously
6399 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
6400 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6401
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6402<h5>Arguments:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6403<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]6404
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6405<h5>Semantics:</h5>
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6406<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]6407 macro available in C. In a target-dependent way, it destroys
6408 the <tt>va_list</tt> element to which the argument points. Calls
6409 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
6410 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
6411 with calls to <tt>llvm.va_end</tt>.</p>
Chris Lattnerb75137d2007-01-08 07:55:15 +0000[diff] [blame]6412
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6413</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6414
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6415<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6416<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6417 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6418</h4>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6419
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6420<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6421
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6422<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6423<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]6424 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6425</pre>
6426
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6427<h5>Overview:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6428<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6429 from the source argument list to the destination argument list.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6430
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6431<h5>Arguments:</h5>
Andrew Lenharth8bf607a2005-06-18 18:28:17 +0000[diff] [blame]6432<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]6433 The second argument is a pointer to a <tt>va_list</tt> element to copy
6434 from.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6435
Chris Lattnerd9ad5b32003-05-08 04:57:36 +0000[diff] [blame]6436<h5>Semantics:</h5>
Jeff Cohenb627eab2007-04-29 01:07:00 +0000[diff] [blame]6437<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]6438 macro available in C. In a target-dependent way, it copies the
6439 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
6440 element. This intrinsic is necessary because
6441 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
6442 arbitrarily complex and require, for example, memory allocation.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6443
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]6444</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]6445
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6446</div>
6447
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6448<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6449<h3>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6450 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6451</h3>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6452
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6453<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6454
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6455<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattnerd3eda892008-08-05 18:29:16 +0000[diff] [blame]6456Collection</a> (GC) requires the implementation and generation of these
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6457intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
6458roots on the stack</a>, as well as garbage collector implementations that
6459require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
6460barriers. Front-ends for type-safe garbage collected languages should generate
6461these intrinsics to make use of the LLVM garbage collectors. For more details,
6462see <a href="GarbageCollection.html">Accurate Garbage Collection with
6463LLVM</a>.</p>
Christopher Lamb303dae92007-12-17 01:00:21 +0000[diff] [blame]6464
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6465<p>The garbage collection intrinsics only operate on objects in the generic
6466 address space (address space zero).</p>
Christopher Lamb303dae92007-12-17 01:00:21 +0000[diff] [blame]6467
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6468<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6469<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6470 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6471</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6472
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6473<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6474
6475<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6476<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6477 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6478</pre>
6479
6480<h5>Overview:</h5>
John Criswell9e2485c2004-12-10 15:51:16 +0000[diff] [blame]6481<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]6482 the code generator, and allows some metadata to be associated with it.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6483
6484<h5>Arguments:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6485<p>The first argument specifies the address of a stack object that contains the
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6486 root pointer. The second pointer (which must be either a constant or a
6487 global value address) contains the meta-data to be associated with the
6488 root.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6489
6490<h5>Semantics:</h5>
Chris Lattner05d67092008-04-24 05:59:56 +0000[diff] [blame]6491<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]6492 location. At compile-time, the code generator generates information to allow
6493 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
6494 intrinsic may only be used in a function which <a href="#gc">specifies a GC
6495 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6496
6497</div>
6498
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6499<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6500<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6501 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6502</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6503
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6504<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6505
6506<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6507<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6508 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6509</pre>
6510
6511<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6512<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6513 locations, allowing garbage collector implementations that require read
6514 barriers.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6515
6516<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]6517<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]6518 allocated from the garbage collector. The first object is a pointer to the
6519 start of the referenced object, if needed by the language runtime (otherwise
6520 null).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6521
6522<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6523<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6524 instruction, but may be replaced with substantially more complex code by the
6525 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
6526 may only be used in a function which <a href="#gc">specifies a GC
6527 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6528
6529</div>
6530
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6531<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6532<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6533 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6534</h4>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6535
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6536<div>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6537
6538<h5>Syntax:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6539<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6540 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6541</pre>
6542
6543<h5>Overview:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6544<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6545 locations, allowing garbage collector implementations that require write
6546 barriers (such as generational or reference counting collectors).</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6547
6548<h5>Arguments:</h5>
Chris Lattner80626e92006-03-14 20:02:51 +0000[diff] [blame]6549<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]6550 object to store it to, and the third is the address of the field of Obj to
6551 store to. If the runtime does not require a pointer to the object, Obj may
6552 be null.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6553
6554<h5>Semantics:</h5>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6555<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6556 instruction, but may be replaced with substantially more complex code by the
6557 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
6558 may only be used in a function which <a href="#gc">specifies a GC
6559 algorithm</a>.</p>
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6560
6561</div>
6562
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6563</div>
6564
Chris Lattnerd7923912004-05-23 21:06:01 +0000[diff] [blame]6565<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6566<h3>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6567 <a name="int_codegen">Code Generator Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6568</h3>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6569
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6570<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6571
6572<p>These intrinsics are provided by LLVM to expose special features that may
6573 only be implemented with code generator support.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6574
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6575<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6576<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6577 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6578</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6579
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6580<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6581
6582<h5>Syntax:</h5>
6583<pre>
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]6584 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6585</pre>
6586
6587<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6588<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
6589 target-specific value indicating the return address of the current function
6590 or one of its callers.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6591
6592<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6593<p>The argument to this intrinsic indicates which function to return the address
6594 for. Zero indicates the calling function, one indicates its caller, etc.
6595 The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6596
6597<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6598<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
6599 indicating the return address of the specified call frame, or zero if it
6600 cannot be identified. The value returned by this intrinsic is likely to be
6601 incorrect or 0 for arguments other than zero, so it should only be used for
6602 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6603
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6604<p>Note that calling this intrinsic does not prevent function inlining or other
6605 aggressive transformations, so the value returned may not be that of the
6606 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6607
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6608</div>
6609
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6610<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6611<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6612 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6613</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6614
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6615<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6616
6617<h5>Syntax:</h5>
6618<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6619 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6620</pre>
6621
6622<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6623<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
6624 target-specific frame pointer value for the specified stack frame.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6625
6626<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6627<p>The argument to this intrinsic indicates which function to return the frame
6628 pointer for. Zero indicates the calling function, one indicates its caller,
6629 etc. The argument is <b>required</b> to be a constant integer value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6630
6631<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6632<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
6633 indicating the frame address of the specified call frame, or zero if it
6634 cannot be identified. The value returned by this intrinsic is likely to be
6635 incorrect or 0 for arguments other than zero, so it should only be used for
6636 debugging purposes.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6637
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6638<p>Note that calling this intrinsic does not prevent function inlining or other
6639 aggressive transformations, so the value returned may not be that of the
6640 obvious source-language caller.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6641
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6642</div>
6643
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6644<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6645<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6646 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6647</h4>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6648
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6649<div>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6650
6651<h5>Syntax:</h5>
6652<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6653 declare i8* @llvm.stacksave()
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6654</pre>
6655
6656<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6657<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
6658 of the function stack, for use
6659 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
6660 useful for implementing language features like scoped automatic variable
6661 sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6662
6663<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6664<p>This intrinsic returns a opaque pointer value that can be passed
6665 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
6666 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
6667 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
6668 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
6669 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
6670 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6671
6672</div>
6673
6674<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6675<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6676 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6677</h4>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6678
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6679<div>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6680
6681<h5>Syntax:</h5>
6682<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6683 declare void @llvm.stackrestore(i8* %ptr)
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6684</pre>
6685
6686<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6687<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
6688 the function stack to the state it was in when the
6689 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
6690 executed. This is useful for implementing language features like scoped
6691 automatic variable sized arrays in C99.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6692
6693<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6694<p>See the description
6695 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6696
6697</div>
6698
Chris Lattner57e1f392006-01-13 02:03:13 +0000[diff] [blame]6699<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6700<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6701 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6702</h4>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6703
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6704<div>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6705
6706<h5>Syntax:</h5>
6707<pre>
Bruno Cardoso Lopes9a767332011-06-14 04:58:37 +0000[diff] [blame]6708 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]6709</pre>
6710
6711<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6712<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
6713 insert a prefetch instruction if supported; otherwise, it is a noop.
6714 Prefetches have no effect on the behavior of the program but can change its
6715 performance characteristics.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6716
6717<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6718<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
6719 specifier determining if the fetch should be for a read (0) or write (1),
6720 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Bruno Cardoso Lopes9a767332011-06-14 04:58:37 +0000[diff] [blame]6721 locality, to (3) - extremely local keep in cache. The <tt>cache type</tt>
6722 specifies whether the prefetch is performed on the data (1) or instruction (0)
6723 cache. The <tt>rw</tt>, <tt>locality</tt> and <tt>cache type</tt> arguments
6724 must be constant integers.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6725
6726<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6727<p>This intrinsic does not modify the behavior of the program. In particular,
6728 prefetches cannot trap and do not produce a value. On targets that support
6729 this intrinsic, the prefetch can provide hints to the processor cache for
6730 better performance.</p>
Chris Lattner9a9d7ac2005-02-28 19:24:19 +0000[diff] [blame]6731
6732</div>
6733
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6734<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6735<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6736 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6737</h4>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6738
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6739<div>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6740
6741<h5>Syntax:</h5>
6742<pre>
Chris Lattner1df4f752007-09-21 17:30:40 +0000[diff] [blame]6743 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6744</pre>
6745
6746<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6747<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
6748 Counter (PC) in a region of code to simulators and other tools. The method
6749 is target specific, but it is expected that the marker will use exported
6750 symbols to transmit the PC of the marker. The marker makes no guarantees
6751 that it will remain with any specific instruction after optimizations. It is
6752 possible that the presence of a marker will inhibit optimizations. The
6753 intended use is to be inserted after optimizations to allow correlations of
6754 simulation runs.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6755
6756<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6757<p><tt>id</tt> is a numerical id identifying the marker.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6758
6759<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6760<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6761 not support this intrinsic may ignore it.</p>
Andrew Lenharth7f4ec3b2005-03-28 20:05:49 +0000[diff] [blame]6762
6763</div>
6764
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6765<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6766<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6767 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6768</h4>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6769
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6770<div>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6771
6772<h5>Syntax:</h5>
6773<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]6774 declare i64 @llvm.readcyclecounter()
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6775</pre>
6776
6777<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6778<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6779 counter register (or similar low latency, high accuracy clocks) on those
6780 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6781 should map to RPCC. As the backing counters overflow quickly (on the order
6782 of 9 seconds on alpha), this should only be used for small timings.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6783
6784<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6785<p>When directly supported, reading the cycle counter should not modify any
6786 memory. Implementations are allowed to either return a application specific
6787 value or a system wide value. On backends without support, this is lowered
6788 to a constant 0.</p>
Andrew Lenharth51b8d542005-11-11 16:47:30 +0000[diff] [blame]6789
6790</div>
6791
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6792</div>
6793
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6794<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6795<h3>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6796 <a name="int_libc">Standard C Library Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6797</h3>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6798
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6799<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6800
6801<p>LLVM provides intrinsics for a few important standard C library functions.
6802 These intrinsics allow source-language front-ends to pass information about
6803 the alignment of the pointer arguments to the code generator, providing
6804 opportunity for more efficient code generation.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6805
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6806<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6807<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6808 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6809</h4>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6810
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6811<div>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6812
6813<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6814<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]6815 integer bit width and for different address spaces. Not all targets support
6816 all bit widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6817
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6818<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6819 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]6820 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6821 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]6822 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6823</pre>
6824
6825<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6826<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6827 source location to the destination location.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6828
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6829<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6830 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6831 and the pointers can be in specified address spaces.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6832
6833<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6834
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6835<p>The first argument is a pointer to the destination, the second is a pointer
6836 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6837 number of bytes to copy, the fourth argument is the alignment of the
6838 source and destination locations, and the fifth is a boolean indicating a
6839 volatile access.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6840
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6841<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]6842 then the caller guarantees that both the source and destination pointers are
6843 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]6844
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]6845<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6846 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6847 The detailed access behavior is not very cleanly specified and it is unwise
6848 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6849
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6850<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6851
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6852<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6853 source location to the destination location, which are not allowed to
6854 overlap. It copies "len" bytes of memory over. If the argument is known to
6855 be aligned to some boundary, this can be specified as the fourth argument,
6856 otherwise it should be set to 0 or 1.</p>
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6857
Chris Lattner33aec9e2004-02-12 17:01:32 +0000[diff] [blame]6858</div>
6859
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6860<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6861<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6862 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6863</h4>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6864
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6865<div>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6866
6867<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +0000[diff] [blame]6868<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]6869 width and for different address space. Not all targets support all bit
6870 widths however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6871
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6872<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6873 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]6874 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6875 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]6876 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6877</pre>
6878
6879<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6880<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6881 source location to the destination location. It is similar to the
6882 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6883 overlap.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6884
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6885<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6886 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6887 and the pointers can be in specified address spaces.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6888
6889<h5>Arguments:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6890
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6891<p>The first argument is a pointer to the destination, the second is a pointer
6892 to the source. The third argument is an integer argument specifying the
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6893 number of bytes to copy, the fourth argument is the alignment of the
6894 source and destination locations, and the fifth is a boolean indicating a
6895 volatile access.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6896
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6897<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]6898 then the caller guarantees that the source and destination pointers are
6899 aligned to that boundary.</p>
Chris Lattner3301ced2004-02-12 21:18:15 +0000[diff] [blame]6900
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]6901<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6902 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6903 The detailed access behavior is not very cleanly specified and it is unwise
6904 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6905
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6906<h5>Semantics:</h5>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6907
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6908<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6909 source location to the destination location, which may overlap. It copies
6910 "len" bytes of memory over. If the argument is known to be aligned to some
6911 boundary, this can be specified as the fourth argument, otherwise it should
6912 be set to 0 or 1.</p>
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6913
Chris Lattner0eb51b42004-02-12 18:10:10 +0000[diff] [blame]6914</div>
6915
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6916<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6917<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6918 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6919</h4>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6920
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6921<div>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6922
6923<h5>Syntax:</h5>
Chris Lattner824b9582008-11-21 16:42:48 +0000[diff] [blame]6924<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]6925 width and for different address spaces. However, not all targets support all
6926 bit widths.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6927
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6928<pre>
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6929 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +0000[diff] [blame]6930 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanfe47aae2010-05-28 17:13:49 +0000[diff] [blame]6931 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerff35c3f2010-04-08 00:54:34 +0000[diff] [blame]6932 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6933</pre>
6934
6935<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6936<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6937 particular byte value.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6938
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6939<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]6940 intrinsic does not return a value and takes extra alignment/volatile
6941 arguments. Also, the destination can be in an arbitrary address space.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6942
6943<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6944<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]6945 byte value with which to fill it, the third argument is an integer argument
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6946 specifying the number of bytes to fill, and the fourth argument is the known
John Criswellcdcbbfc2010-07-30 16:30:28 +0000[diff] [blame]6947 alignment of the destination location.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6948
Dan Gohmanb55a1ee2010-03-01 17:41:39 +0000[diff] [blame]6949<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]6950 then the caller guarantees that the destination pointer is aligned to that
6951 boundary.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6952
Jeffrey Yasskin93e066d2010-04-26 21:21:24 +0000[diff] [blame]6953<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6954 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6955 The detailed access behavior is not very cleanly specified and it is unwise
6956 to depend on it.</p>
Chris Lattner9f636de2010-04-08 00:53:57 +0000[diff] [blame]6957
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6958<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6959<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6960 at the destination location. If the argument is known to be aligned to some
6961 boundary, this can be specified as the fourth argument, otherwise it should
6962 be set to 0 or 1.</p>
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6963
Chris Lattner10610642004-02-14 04:08:35 +0000[diff] [blame]6964</div>
6965
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]6966<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6967<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]6968 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]6969</h4>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6970
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]6971<div>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6972
6973<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6974<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6975 floating point or vector of floating point type. Not all targets support all
6976 types however.</p>
6977
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6978<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +0000[diff] [blame]6979 declare float @llvm.sqrt.f32(float %Val)
6980 declare double @llvm.sqrt.f64(double %Val)
6981 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6982 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6983 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6984</pre>
6985
6986<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6987<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6988 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6989 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6990 behavior for negative numbers other than -0.0 (which allows for better
6991 optimization, because there is no need to worry about errno being
6992 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]6993
6994<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6995<p>The argument and return value are floating point numbers of the same
6996 type.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]6997
6998<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]6999<p>This function returns the sqrt of the specified operand if it is a
7000 nonnegative floating point number.</p>
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7001
Chris Lattnera4d74142005-07-21 01:29:16 +0000[diff] [blame]7002</div>
7003
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7004<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7005<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7006 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7007</h4>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7008
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7009<div>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7010
7011<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7012<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
7013 floating point or vector of floating point type. Not all targets support all
7014 types however.</p>
7015
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7016<pre>
Dale Johannesen408f9c12007-10-02 17:47:38 +0000[diff] [blame]7017 declare float @llvm.powi.f32(float %Val, i32 %power)
7018 declare double @llvm.powi.f64(double %Val, i32 %power)
7019 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
7020 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
7021 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattnerf4d252d2006-09-08 06:34:02 +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.powi.*</tt>' intrinsics return the first operand raised to the
7026 specified (positive or negative) power. The order of evaluation of
7027 multiplications is not defined. When a vector of floating point type is
7028 used, the second argument remains a scalar integer value.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7029
7030<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7031<p>The second argument is an integer power, and the first is a value to raise to
7032 that power.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7033
7034<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7035<p>This function returns the first value raised to the second power with an
7036 unspecified sequence of rounding operations.</p>
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7037
Chris Lattnerf4d252d2006-09-08 06:34:02 +0000[diff] [blame]7038</div>
7039
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7040<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7041<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7042 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7043</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7044
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7045<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7046
7047<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7048<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
7049 floating point or vector of floating point type. Not all targets support all
7050 types however.</p>
7051
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7052<pre>
7053 declare float @llvm.sin.f32(float %Val)
7054 declare double @llvm.sin.f64(double %Val)
7055 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
7056 declare fp128 @llvm.sin.f128(fp128 %Val)
7057 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
7058</pre>
7059
7060<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7061<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7062
7063<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7064<p>The argument and return value are floating point numbers of the same
7065 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7066
7067<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7068<p>This function returns the sine of the specified operand, returning the same
7069 values as the libm <tt>sin</tt> functions would, and handles error conditions
7070 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7071
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7072</div>
7073
7074<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7075<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7076 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7077</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7078
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7079<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7080
7081<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7082<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
7083 floating point or vector of floating point type. Not all targets support all
7084 types however.</p>
7085
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7086<pre>
7087 declare float @llvm.cos.f32(float %Val)
7088 declare double @llvm.cos.f64(double %Val)
7089 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
7090 declare fp128 @llvm.cos.f128(fp128 %Val)
7091 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
7092</pre>
7093
7094<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7095<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7096
7097<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7098<p>The argument and return value are floating point numbers of the same
7099 type.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7100
7101<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7102<p>This function returns the cosine of the specified operand, returning the same
7103 values as the libm <tt>cos</tt> functions would, and handles error conditions
7104 in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7105
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7106</div>
7107
7108<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7109<h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7110 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7111</h4>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7112
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7113<div>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7114
7115<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7116<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
7117 floating point or vector of floating point type. Not all targets support all
7118 types however.</p>
7119
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7120<pre>
7121 declare float @llvm.pow.f32(float %Val, float %Power)
7122 declare double @llvm.pow.f64(double %Val, double %Power)
7123 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
7124 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
7125 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
7126</pre>
7127
7128<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7129<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
7130 specified (positive or negative) power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7131
7132<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7133<p>The second argument is a floating point power, and the first is a value to
7134 raise to that power.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7135
7136<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7137<p>This function returns the first value raised to the second power, returning
7138 the same values as the libm <tt>pow</tt> functions would, and handles error
7139 conditions in the same way.</p>
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7140
Dan Gohman91c284c2007-10-15 20:30:11 +0000[diff] [blame]7141</div>
7142
Dan Gohman4e9011c2011-05-23 21:13:03 +0000[diff] [blame]7143<!-- _______________________________________________________________________ -->
7144<h4>
7145 <a name="int_exp">'<tt>llvm.exp.*</tt>' Intrinsic</a>
7146</h4>
7147
7148<div>
7149
7150<h5>Syntax:</h5>
7151<p>This is an overloaded intrinsic. You can use <tt>llvm.exp</tt> on any
7152 floating point or vector of floating point type. Not all targets support all
7153 types however.</p>
7154
7155<pre>
7156 declare float @llvm.exp.f32(float %Val)
7157 declare double @llvm.exp.f64(double %Val)
7158 declare x86_fp80 @llvm.exp.f80(x86_fp80 %Val)
7159 declare fp128 @llvm.exp.f128(fp128 %Val)
7160 declare ppc_fp128 @llvm.exp.ppcf128(ppc_fp128 %Val)
7161</pre>
7162
7163<h5>Overview:</h5>
7164<p>The '<tt>llvm.exp.*</tt>' intrinsics perform the exp function.</p>
7165
7166<h5>Arguments:</h5>
7167<p>The argument and return value are floating point numbers of the same
7168 type.</p>
7169
7170<h5>Semantics:</h5>
7171<p>This function returns the same values as the libm <tt>exp</tt> functions
7172 would, and handles error conditions in the same way.</p>
7173
7174</div>
7175
7176<!-- _______________________________________________________________________ -->
7177<h4>
7178 <a name="int_log">'<tt>llvm.log.*</tt>' Intrinsic</a>
7179</h4>
7180
7181<div>
7182
7183<h5>Syntax:</h5>
7184<p>This is an overloaded intrinsic. You can use <tt>llvm.log</tt> on any
7185 floating point or vector of floating point type. Not all targets support all
7186 types however.</p>
7187
7188<pre>
7189 declare float @llvm.log.f32(float %Val)
7190 declare double @llvm.log.f64(double %Val)
7191 declare x86_fp80 @llvm.log.f80(x86_fp80 %Val)
7192 declare fp128 @llvm.log.f128(fp128 %Val)
7193 declare ppc_fp128 @llvm.log.ppcf128(ppc_fp128 %Val)
7194</pre>
7195
7196<h5>Overview:</h5>
7197<p>The '<tt>llvm.log.*</tt>' intrinsics perform the log function.</p>
7198
7199<h5>Arguments:</h5>
7200<p>The argument and return value are floating point numbers of the same
7201 type.</p>
7202
7203<h5>Semantics:</h5>
7204<p>This function returns the same values as the libm <tt>log</tt> functions
7205 would, and handles error conditions in the same way.</p>
7206
Nick Lewycky1c929be2011-10-31 01:32:21 +0000[diff] [blame]7207</div>
7208
7209<!-- _______________________________________________________________________ -->
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]7210<h4>
7211 <a name="int_fma">'<tt>llvm.fma.*</tt>' Intrinsic</a>
7212</h4>
7213
7214<div>
7215
7216<h5>Syntax:</h5>
7217<p>This is an overloaded intrinsic. You can use <tt>llvm.fma</tt> on any
7218 floating point or vector of floating point type. Not all targets support all
7219 types however.</p>
7220
7221<pre>
7222 declare float @llvm.fma.f32(float %a, float %b, float %c)
7223 declare double @llvm.fma.f64(double %a, double %b, double %c)
7224 declare x86_fp80 @llvm.fma.f80(x86_fp80 %a, x86_fp80 %b, x86_fp80 %c)
7225 declare fp128 @llvm.fma.f128(fp128 %a, fp128 %b, fp128 %c)
7226 declare ppc_fp128 @llvm.fma.ppcf128(ppc_fp128 %a, ppc_fp128 %b, ppc_fp128 %c)
7227</pre>
7228
7229<h5>Overview:</h5>
Cameron Zwarichabc43e62011-07-08 22:13:55 +0000[diff] [blame]7230<p>The '<tt>llvm.fma.*</tt>' intrinsics perform the fused multiply-add
Cameron Zwarich33390842011-07-08 21:39:21 +0000[diff] [blame]7231 operation.</p>
7232
7233<h5>Arguments:</h5>
7234<p>The argument and return value are floating point numbers of the same
7235 type.</p>
7236
7237<h5>Semantics:</h5>
7238<p>This function returns the same values as the libm <tt>fma</tt> functions
7239 would.</p>
7240
Dan Gohman4e9011c2011-05-23 21:13:03 +0000[diff] [blame]7241</div>
7242
NAKAMURA Takumi4b2e07a2011-10-31 13:04:26 +0000[diff] [blame]7243</div>
7244
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7245<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7246<h3>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7247 <a name="int_manip">Bit Manipulation Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7248</h3>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7249
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7250<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7251
7252<p>LLVM provides intrinsics for a few important bit manipulation operations.
7253 These allow efficient code generation for some algorithms.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7254
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7255<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7256<h4>
Reid Spencera3e435f2007-04-04 02:42:35 +0000[diff] [blame]7257 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7258</h4>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7259
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7260<div>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7261
7262<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]7263<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7264 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
7265
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7266<pre>
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7267 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
7268 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
7269 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7270</pre>
7271
7272<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7273<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
7274 values with an even number of bytes (positive multiple of 16 bits). These
7275 are useful for performing operations on data that is not in the target's
7276 native byte order.</p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7277
7278<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7279<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
7280 and low byte of the input i16 swapped. Similarly,
7281 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
7282 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
7283 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
7284 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
7285 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
7286 more, respectively).</p>
Nate Begeman7e36c472006-01-13 23:26:38 +0000[diff] [blame]7287
7288</div>
7289
7290<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7291<h4>
Reid Spencer0b118202006-01-16 21:12:35 +0000[diff] [blame]7292 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7293</h4>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7294
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7295<div>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7296
7297<h5>Syntax:</h5>
Reid Spencer409e28f2007-04-01 08:04:23 +0000[diff] [blame]7298<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]7299 width, or on any vector with integer elements. Not all targets support all
7300 bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7301
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7302<pre>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7303 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7304 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovec43a062007-03-22 00:02:17 +0000[diff] [blame]7305 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth69940402007-08-04 01:51:18 +0000[diff] [blame]7306 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
7307 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7308 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]7309</pre>
7310
7311<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7312<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
7313 in a value.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7314
7315<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7316<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]7317 integer type, or a vector with integer elements.
7318 The return type must match the argument type.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7319
7320<h5>Semantics:</h5>
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7321<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable, or within each
7322 element of a vector.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7323
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7324</div>
7325
7326<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7327<h4>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]7328 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7329</h4>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7330
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7331<div>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7332
7333<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7334<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]7335 integer bit width, or any vector whose elements are integers. Not all
7336 targets support all bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7337
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7338<pre>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7339 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7340 declare i16 @llvm.ctlz.i16 (i16 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7341 declare i32 @llvm.ctlz.i32 (i32 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7342 declare i64 @llvm.ctlz.i64 (i64 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7343 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7344 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]7345</pre>
7346
7347<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7348<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
7349 leading zeros in a variable.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7350
7351<h5>Arguments:</h5>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7352<p>The first argument is the value to be counted. This argument may be of any
7353 integer type, or a vectory with integer element type. The return type
7354 must match the first argument type.</p>
7355
7356<p>The second argument must be a constant and is a flag to indicate whether the
7357 intrinsic should ensure that a zero as the first argument produces a defined
7358 result. Historically some architectures did not provide a defined result for
7359 zero values as efficiently, and many algorithms are now predicated on
7360 avoiding zero-value inputs.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7361
7362<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7363<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7364 zeros in a variable, or within each element of the vector.
7365 If <tt>src == 0</tt> then the result is the size in bits of the type of
7366 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7367 For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7368
Andrew Lenharthec370fd2005-05-03 18:01:48 +0000[diff] [blame]7369</div>
Chris Lattner32006282004-06-11 02:28:03 +0000[diff] [blame]7370
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7371<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7372<h4>
Chris Lattner8a886be2006-01-16 22:34:14 +0000[diff] [blame]7373 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7374</h4>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7375
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7376<div>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7377
7378<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7379<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]7380 integer bit width, or any vector of integer elements. Not all targets
7381 support all bit widths or vector types, however.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7382
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7383<pre>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7384 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7385 declare i16 @llvm.cttz.i16 (i16 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7386 declare i32 @llvm.cttz.i32 (i32 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7387 declare i64 @llvm.cttz.i64 (i64 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7388 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;, i1 &lt;is_zero_undef&gt;)
7389 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]7390</pre>
7391
7392<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7393<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
7394 trailing zeros.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7395
7396<h5>Arguments:</h5>
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7397<p>The first argument is the value to be counted. This argument may be of any
7398 integer type, or a vectory with integer element type. The return type
7399 must match the first argument type.</p>
7400
7401<p>The second argument must be a constant and is a flag to indicate whether the
7402 intrinsic should ensure that a zero as the first argument produces a defined
7403 result. Historically some architectures did not provide a defined result for
7404 zero values as efficiently, and many algorithms are now predicated on
7405 avoiding zero-value inputs.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7406
7407<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7408<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
Owen Andersonf1ac4652011-07-01 21:52:38 +0000[diff] [blame]7409 zeros in a variable, or within each element of a vector.
Chandler Carruth48b0bbf2011-12-12 04:36:04 +0000[diff] [blame]7410 If <tt>src == 0</tt> then the result is the size in bits of the type of
7411 <tt>src</tt> if <tt>is_zero_undef == 0</tt> and <tt>undef</tt> otherwise.
7412 For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7413
Chris Lattnereff29ab2005-05-15 19:39:26 +0000[diff] [blame]7414</div>
7415
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7416</div>
7417
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7418<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7419<h3>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7420 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7421</h3>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7422
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7423<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7424
7425<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendlingda01af72009-02-08 04:04:40 +0000[diff] [blame]7426
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7427<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7428<h4>
7429 <a name="int_sadd_overflow">
7430 '<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics
7431 </a>
7432</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7433
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7434<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7435
7436<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7437<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]7438 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7439
7440<pre>
7441 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
7442 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7443 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
7444</pre>
7445
7446<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7447<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7448 a signed addition of the two arguments, and indicate whether an overflow
7449 occurred during the signed summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7450
7451<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7452<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]7453 be of integer types of any bit width, but they must have the same bit
7454 width. The second element of the result structure must be of
7455 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7456 undergo signed addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7457
7458<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7459<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7460 a signed addition of the two variables. They return a structure &mdash; the
7461 first element of which is the signed summation, and the second element of
7462 which is a bit specifying if the signed summation resulted in an
7463 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7464
7465<h5>Examples:</h5>
7466<pre>
7467 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
7468 %sum = extractvalue {i32, i1} %res, 0
7469 %obit = extractvalue {i32, i1} %res, 1
7470 br i1 %obit, label %overflow, label %normal
7471</pre>
7472
7473</div>
7474
7475<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7476<h4>
7477 <a name="int_uadd_overflow">
7478 '<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics
7479 </a>
7480</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7481
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7482<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7483
7484<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7485<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]7486 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7487
7488<pre>
7489 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
7490 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7491 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
7492</pre>
7493
7494<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7495<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7496 an unsigned addition of the two arguments, and indicate whether a carry
7497 occurred during the unsigned summation.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7498
7499<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7500<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]7501 be of integer types of any bit width, but they must have the same bit
7502 width. The second element of the result structure must be of
7503 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7504 undergo unsigned addition.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7505
7506<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7507<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7508 an unsigned addition of the two arguments. They return a structure &mdash;
7509 the first element of which is the sum, and the second element of which is a
7510 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7511
7512<h5>Examples:</h5>
7513<pre>
7514 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
7515 %sum = extractvalue {i32, i1} %res, 0
7516 %obit = extractvalue {i32, i1} %res, 1
7517 br i1 %obit, label %carry, label %normal
7518</pre>
7519
7520</div>
7521
7522<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7523<h4>
7524 <a name="int_ssub_overflow">
7525 '<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics
7526 </a>
7527</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7528
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7529<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7530
7531<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7532<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]7533 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7534
7535<pre>
7536 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
7537 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7538 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
7539</pre>
7540
7541<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7542<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7543 a signed subtraction of the two arguments, and indicate whether an overflow
7544 occurred during the signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7545
7546<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7547<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]7548 be of integer types of any bit width, but they must have the same bit
7549 width. The second element of the result structure must be of
7550 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7551 undergo signed subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7552
7553<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7554<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7555 a signed subtraction of the two arguments. They return a structure &mdash;
7556 the first element of which is the subtraction, and the second element of
7557 which is a bit specifying if the signed subtraction resulted in an
7558 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7559
7560<h5>Examples:</h5>
7561<pre>
7562 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
7563 %sum = extractvalue {i32, i1} %res, 0
7564 %obit = extractvalue {i32, i1} %res, 1
7565 br i1 %obit, label %overflow, label %normal
7566</pre>
7567
7568</div>
7569
7570<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7571<h4>
7572 <a name="int_usub_overflow">
7573 '<tt>llvm.usub.with.overflow.*</tt>' Intrinsics
7574 </a>
7575</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7576
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7577<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7578
7579<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7580<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]7581 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7582
7583<pre>
7584 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
7585 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7586 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
7587</pre>
7588
7589<h5>Overview:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7590<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7591 an unsigned subtraction of the two arguments, and indicate whether an
7592 overflow occurred during the unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7593
7594<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7595<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]7596 be of integer types of any bit width, but they must have the same bit
7597 width. The second element of the result structure must be of
7598 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7599 undergo unsigned subtraction.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7600
7601<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7602<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7603 an unsigned subtraction of the two arguments. They return a structure &mdash;
7604 the first element of which is the subtraction, and the second element of
7605 which is a bit specifying if the unsigned subtraction resulted in an
7606 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7607
7608<h5>Examples:</h5>
7609<pre>
7610 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
7611 %sum = extractvalue {i32, i1} %res, 0
7612 %obit = extractvalue {i32, i1} %res, 1
7613 br i1 %obit, label %overflow, label %normal
7614</pre>
7615
7616</div>
7617
7618<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7619<h4>
7620 <a name="int_smul_overflow">
7621 '<tt>llvm.smul.with.overflow.*</tt>' Intrinsics
7622 </a>
7623</h4>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7624
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7625<div>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7626
7627<h5>Syntax:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7628<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]7629 on any integer bit width.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7630
7631<pre>
7632 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
7633 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7634 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
7635</pre>
7636
7637<h5>Overview:</h5>
7638
7639<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7640 a signed multiplication of the two arguments, and indicate whether an
7641 overflow occurred during the signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7642
7643<h5>Arguments:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7644<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]7645 be of integer types of any bit width, but they must have the same bit
7646 width. The second element of the result structure must be of
7647 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7648 undergo signed multiplication.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7649
7650<h5>Semantics:</h5>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7651<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7652 a signed multiplication of the two arguments. They return a structure &mdash;
7653 the first element of which is the multiplication, and the second element of
7654 which is a bit specifying if the signed multiplication resulted in an
7655 overflow.</p>
Bill Wendlingac1df8e2009-02-08 01:40:31 +0000[diff] [blame]7656
7657<h5>Examples:</h5>
7658<pre>
7659 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
7660 %sum = extractvalue {i32, i1} %res, 0
7661 %obit = extractvalue {i32, i1} %res, 1
7662 br i1 %obit, label %overflow, label %normal
7663</pre>
7664
Reid Spencerf86037f2007-04-11 23:23:49 +0000[diff] [blame]7665</div>
7666
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7667<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7668<h4>
7669 <a name="int_umul_overflow">
7670 '<tt>llvm.umul.with.overflow.*</tt>' Intrinsics
7671 </a>
7672</h4>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7673
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7674<div>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7675
7676<h5>Syntax:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7677<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]7678 on any integer bit width.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7679
7680<pre>
7681 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
7682 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7683 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
7684</pre>
7685
7686<h5>Overview:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7687<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7688 a unsigned multiplication of the two arguments, and indicate whether an
7689 overflow occurred during the unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7690
7691<h5>Arguments:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7692<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]7693 be of integer types of any bit width, but they must have the same bit
7694 width. The second element of the result structure must be of
7695 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
7696 undergo unsigned multiplication.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7697
7698<h5>Semantics:</h5>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7699<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7700 an unsigned multiplication of the two arguments. They return a structure
7701 &mdash; the first element of which is the multiplication, and the second
7702 element of which is a bit specifying if the unsigned multiplication resulted
7703 in an overflow.</p>
Bill Wendling41b485c2009-02-08 23:00:09 +0000[diff] [blame]7704
7705<h5>Examples:</h5>
7706<pre>
7707 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
7708 %sum = extractvalue {i32, i1} %res, 0
7709 %obit = extractvalue {i32, i1} %res, 1
7710 br i1 %obit, label %overflow, label %normal
7711</pre>
7712
7713</div>
7714
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7715</div>
7716
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7717<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7718<h3>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7719 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7720</h3>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7721
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7722<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7723
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7724<p>Half precision floating point is a storage-only format. This means that it is
7725 a dense encoding (in memory) but does not support computation in the
7726 format.</p>
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7727
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7728<p>This means that code must first load the half-precision floating point
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7729 value as an i16, then convert it to float with <a
7730 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
7731 Computation can then be performed on the float value (including extending to
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7732 double etc). To store the value back to memory, it is first converted to
7733 float if needed, then converted to i16 with
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7734 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
7735 storing as an i16 value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7736
7737<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7738<h4>
7739 <a name="int_convert_to_fp16">
7740 '<tt>llvm.convert.to.fp16</tt>' Intrinsic
7741 </a>
7742</h4>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7743
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7744<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7745
7746<h5>Syntax:</h5>
7747<pre>
7748 declare i16 @llvm.convert.to.fp16(f32 %a)
7749</pre>
7750
7751<h5>Overview:</h5>
7752<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7753 a conversion from single precision floating point format to half precision
7754 floating point format.</p>
7755
7756<h5>Arguments:</h5>
7757<p>The intrinsic function contains single argument - the value to be
7758 converted.</p>
7759
7760<h5>Semantics:</h5>
7761<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
7762 a conversion from single precision floating point format to half precision
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7763 floating point format. The return value is an <tt>i16</tt> which
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7764 contains the converted number.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7765
7766<h5>Examples:</h5>
7767<pre>
7768 %res = call i16 @llvm.convert.to.fp16(f32 %a)
7769 store i16 %res, i16* @x, align 2
7770</pre>
7771
7772</div>
7773
7774<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7775<h4>
7776 <a name="int_convert_from_fp16">
7777 '<tt>llvm.convert.from.fp16</tt>' Intrinsic
7778 </a>
7779</h4>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7780
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7781<div>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7782
7783<h5>Syntax:</h5>
7784<pre>
7785 declare f32 @llvm.convert.from.fp16(i16 %a)
7786</pre>
7787
7788<h5>Overview:</h5>
7789<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
7790 a conversion from half precision floating point format to single precision
7791 floating point format.</p>
7792
7793<h5>Arguments:</h5>
7794<p>The intrinsic function contains single argument - the value to be
7795 converted.</p>
7796
7797<h5>Semantics:</h5>
7798<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattner0cec9c82010-03-15 04:12:21 +0000[diff] [blame]7799 conversion from half single precision floating point format to single
Chris Lattner82c3dc62010-03-14 23:03:31 +0000[diff] [blame]7800 precision floating point format. The input half-float value is represented by
7801 an <tt>i16</tt> value.</p>
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7802
7803<h5>Examples:</h5>
7804<pre>
7805 %a = load i16* @x, align 2
7806 %res = call f32 @llvm.convert.from.fp16(i16 %a)
7807</pre>
7808
7809</div>
7810
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7811</div>
7812
Anton Korobeynikovf02e7302010-03-14 18:42:47 +0000[diff] [blame]7813<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7814<h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7815 <a name="int_debugger">Debugger Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7816</h3>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7817
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7818<div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7819
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7820<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
7821 prefix), are described in
7822 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
7823 Level Debugging</a> document.</p>
7824
7825</div>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]7826
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7827<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7828<h3>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7829 <a name="int_eh">Exception Handling Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7830</h3>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7831
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7832<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7833
7834<p>The LLVM exception handling intrinsics (which all start with
7835 <tt>llvm.eh.</tt> prefix), are described in
7836 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
7837 Handling</a> document.</p>
7838
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]7839</div>
7840
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]7841<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7842<h3>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]7843 <a name="int_trampoline">Trampoline Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7844</h3>
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7845
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7846<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7847
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]7848<p>These intrinsics make it possible to excise one parameter, marked with
Dan Gohmanff235352010-07-02 23:18:08 +0000[diff] [blame]7849 the <a href="#nest"><tt>nest</tt></a> attribute, from a function.
7850 The result is a callable
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7851 function pointer lacking the nest parameter - the caller does not need to
7852 provide a value for it. Instead, the value to use is stored in advance in a
7853 "trampoline", a block of memory usually allocated on the stack, which also
7854 contains code to splice the nest value into the argument list. This is used
7855 to implement the GCC nested function address extension.</p>
7856
7857<p>For example, if the function is
7858 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
7859 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
7860 follows:</p>
7861
Benjamin Kramer26fe25f2010-07-13 12:26:09 +0000[diff] [blame]7862<pre class="doc_code">
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]7863 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
7864 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]7865 call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8*, i32, i32)* @f to i8*), i8* %nval)
7866 %p = call i8* @llvm.adjust.trampoline(i8* %tramp1)
Duncan Sandsf7331b32007-09-11 14:10:23 +0000[diff] [blame]7867 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7868</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7869
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]7870<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
7871 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7872
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7873<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7874<h4>
7875 <a name="int_it">
7876 '<tt>llvm.init.trampoline</tt>' Intrinsic
7877 </a>
7878</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7879
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7880<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7881
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7882<h5>Syntax:</h5>
7883<pre>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]7884 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]7885</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7886
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7887<h5>Overview:</h5>
Duncan Sands4a544a72011-09-06 13:37:06 +0000[diff] [blame]7888<p>This fills the memory pointed to by <tt>tramp</tt> with executable code,
7889 turning it into a trampoline.</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7890
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7891<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7892<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
7893 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
7894 sufficiently aligned block of memory; this memory is written to by the
7895 intrinsic. Note that the size and the alignment are target-specific - LLVM
7896 currently provides no portable way of determining them, so a front-end that
7897 generates this intrinsic needs to have some target-specific knowledge.
7898 The <tt>func</tt> argument must hold a function bitcast to
7899 an <tt>i8*</tt>.</p>
7900
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7901<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7902<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]7903 dependent code, turning it into a function. Then <tt>tramp</tt> needs to be
7904 passed to <a href="#int_at">llvm.adjust.trampoline</a> to get a pointer
7905 which can be <a href="#int_trampoline">bitcast (to a new function) and
7906 called</a>. The new function's signature is the same as that of
7907 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
7908 removed. At most one such <tt>nest</tt> argument is allowed, and it must be of
7909 pointer type. Calling the new function is equivalent to calling <tt>func</tt>
7910 with the same argument list, but with <tt>nval</tt> used for the missing
7911 <tt>nest</tt> argument. If, after calling <tt>llvm.init.trampoline</tt>, the
7912 memory pointed to by <tt>tramp</tt> is modified, then the effect of any later call
7913 to the returned function pointer is undefined.</p>
7914</div>
7915
7916<!-- _______________________________________________________________________ -->
7917<h4>
7918 <a name="int_at">
7919 '<tt>llvm.adjust.trampoline</tt>' Intrinsic
7920 </a>
7921</h4>
7922
7923<div>
7924
7925<h5>Syntax:</h5>
7926<pre>
7927 declare i8* @llvm.adjust.trampoline(i8* &lt;tramp&gt;)
7928</pre>
7929
7930<h5>Overview:</h5>
7931<p>This performs any required machine-specific adjustment to the address of a
7932 trampoline (passed as <tt>tramp</tt>).</p>
7933
7934<h5>Arguments:</h5>
7935<p><tt>tramp</tt> must point to a block of memory which already has trampoline code
7936 filled in by a previous call to <a href="#int_it"><tt>llvm.init.trampoline</tt>
7937 </a>.</p>
7938
7939<h5>Semantics:</h5>
7940<p>On some architectures the address of the code to be executed needs to be
7941 different to the address where the trampoline is actually stored. This
7942 intrinsic returns the executable address corresponding to <tt>tramp</tt>
7943 after performing the required machine specific adjustments.
7944 The pointer returned can then be <a href="#int_trampoline"> bitcast and
7945 executed</a>.
7946</p>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]7947
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7948</div>
7949
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7950</div>
7951
Duncan Sands36397f52007-07-27 12:58:54 +0000[diff] [blame]7952<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7953<h3>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7954 <a name="int_memorymarkers">Memory Use Markers</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7955</h3>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7956
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7957<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7958
7959<p>This class of intrinsics exists to information about the lifetime of memory
7960 objects and ranges where variables are immutable.</p>
7961
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7962<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7963<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7964 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7965</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7966
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7967<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7968
7969<h5>Syntax:</h5>
7970<pre>
7971 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7972</pre>
7973
7974<h5>Overview:</h5>
7975<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7976 object's lifetime.</p>
7977
7978<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]7979<p>The first argument is a constant integer representing the size of the
7980 object, or -1 if it is variable sized. The second argument is a pointer to
7981 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7982
7983<h5>Semantics:</h5>
7984<p>This intrinsic indicates that before this point in the code, the value of the
7985 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]7986 never be used and has an undefined value. A load from the pointer that
7987 precedes this intrinsic can be replaced with
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7988 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7989
7990</div>
7991
7992<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7993<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7994 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]7995</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7996
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]7997<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]7998
7999<h5>Syntax:</h5>
8000<pre>
8001 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8002</pre>
8003
8004<h5>Overview:</h5>
8005<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
8006 object's lifetime.</p>
8007
8008<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8009<p>The first argument is a constant integer representing the size of the
8010 object, or -1 if it is variable sized. The second argument is a pointer to
8011 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8012
8013<h5>Semantics:</h5>
8014<p>This intrinsic indicates that after this point in the code, the value of the
8015 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
8016 never be used and has an undefined value. Any stores into the memory object
8017 following this intrinsic may be removed as dead.
8018
8019</div>
8020
8021<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8022<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8023 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8024</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8025
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8026<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8027
8028<h5>Syntax:</h5>
8029<pre>
Nick Lewycky29b6cb42010-11-30 04:13:41 +0000[diff] [blame]8030 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8031</pre>
8032
8033<h5>Overview:</h5>
8034<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
8035 a memory object will not change.</p>
8036
8037<h5>Arguments:</h5>
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8038<p>The first argument is a constant integer representing the size of the
8039 object, or -1 if it is variable sized. The second argument is a pointer to
8040 the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8041
8042<h5>Semantics:</h5>
8043<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
8044 the return value, the referenced memory location is constant and
8045 unchanging.</p>
8046
8047</div>
8048
8049<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8050<h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8051 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8052</h4>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8053
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8054<div>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8055
8056<h5>Syntax:</h5>
8057<pre>
8058 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
8059</pre>
8060
8061<h5>Overview:</h5>
8062<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
8063 a memory object are mutable.</p>
8064
8065<h5>Arguments:</h5>
8066<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky321333e2009-10-13 07:57:33 +0000[diff] [blame]8067 The second argument is a constant integer representing the size of the
8068 object, or -1 if it is variable sized and the third argument is a pointer
8069 to the object.</p>
Nick Lewyckycc271862009-10-13 07:03:23 +0000[diff] [blame]8070
8071<h5>Semantics:</h5>
8072<p>This intrinsic indicates that the memory is mutable again.</p>
8073
8074</div>
8075
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8076</div>
8077
Andrew Lenharth22c5c1b2008-02-16 01:24:58 +0000[diff] [blame]8078<!-- ======================================================================= -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8079<h3>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8080 <a name="int_general">General Intrinsics</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8081</h3>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8082
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8083<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8084
8085<p>This class of intrinsics is designed to be generic and has no specific
8086 purpose.</p>
8087
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8088<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8089<h4>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8090 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8091</h4>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8092
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8093<div>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8094
8095<h5>Syntax:</h5>
8096<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8097 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]8098</pre>
8099
8100<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8101<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8102
8103<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8104<p>The first argument is a pointer to a value, the second is a pointer to a
8105 global string, the third is a pointer to a global string which is the source
8106 file name, and the last argument is the line number.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8107
8108<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8109<p>This intrinsic allows annotation of local variables with arbitrary strings.
8110 This can be useful for special purpose optimizations that want to look for
John Criswelle865c032011-08-19 16:57:55 +0000[diff] [blame]8111 these annotations. These have no other defined use; they are ignored by code
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8112 generation and optimization.</p>
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8113
Tanya Lattner6d806e92007-06-15 20:50:54 +0000[diff] [blame]8114</div>
8115
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8116<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8117<h4>
Tanya Lattnere1a8da02007-09-21 23:57:59 +0000[diff] [blame]8118 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8119</h4>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8120
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8121<div>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8122
8123<h5>Syntax:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8124<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
8125 any integer bit width.</p>
8126
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8127<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8128 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8129 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8130 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8131 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
8132 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]8133</pre>
8134
8135<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8136<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8137
8138<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8139<p>The first argument is an integer value (result of some expression), the
8140 second is a pointer to a global string, the third is a pointer to a global
8141 string which is the source file name, and the last argument is the line
8142 number. It returns the value of the first argument.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8143
8144<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8145<p>This intrinsic allows annotations to be put on arbitrary expressions with
8146 arbitrary strings. This can be useful for special purpose optimizations that
John Criswelle865c032011-08-19 16:57:55 +0000[diff] [blame]8147 want to look for these annotations. These have no other defined use; they
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8148 are ignored by code generation and optimization.</p>
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8149
Tanya Lattnerb6367882007-09-21 22:59:12 +0000[diff] [blame]8150</div>
Jim Laskeydd4ef1b2007-03-14 19:31:19 +0000[diff] [blame]8151
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8152<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8153<h4>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8154 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8155</h4>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8156
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8157<div>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8158
8159<h5>Syntax:</h5>
8160<pre>
8161 declare void @llvm.trap()
8162</pre>
8163
8164<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8165<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8166
8167<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8168<p>None.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8169
8170<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8171<p>This intrinsics is lowered to the target dependent trap instruction. If the
8172 target does not have a trap instruction, this intrinsic will be lowered to
8173 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8174
Anton Korobeynikov4cb86182008-01-15 22:31:34 +0000[diff] [blame]8175</div>
8176
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8177<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8178<h4>
Misha Brukmandccb0252008-11-22 23:55:29 +0000[diff] [blame]8179 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8180</h4>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8181
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8182<div>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8183
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8184<h5>Syntax:</h5>
8185<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8186 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8187</pre>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8188
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8189<h5>Overview:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8190<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
8191 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
8192 ensure that it is placed on the stack before local variables.</p>
8193
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8194<h5>Arguments:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8195<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
8196 arguments. The first argument is the value loaded from the stack
8197 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
8198 that has enough space to hold the value of the guard.</p>
8199
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8200<h5>Semantics:</h5>
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8201<p>This intrinsic causes the prologue/epilogue inserter to force the position of
8202 the <tt>AllocaInst</tt> stack slot to be before local variables on the
8203 stack. This is to ensure that if a local variable on the stack is
8204 overwritten, it will destroy the value of the guard. When the function exits,
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8205 the guard on the stack is checked against the original guard. If they are
Bill Wendlinge910b4c2009-07-20 02:29:24 +0000[diff] [blame]8206 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
8207 function.</p>
8208
Bill Wendling69e4adb2008-11-19 05:56:17 +0000[diff] [blame]8209</div>
8210
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8211<!-- _______________________________________________________________________ -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8212<h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8213 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]8214</h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8215
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8216<div>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8217
8218<h5>Syntax:</h5>
8219<pre>
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]8220 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;type&gt;)
8221 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;type&gt;)
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8222</pre>
8223
8224<h5>Overview:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8225<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information to
8226 the optimizers to determine at compile time whether a) an operation (like
8227 memcpy) will overflow a buffer that corresponds to an object, or b) that a
8228 runtime check for overflow isn't necessary. An object in this context means
8229 an allocation of a specific class, structure, array, or other object.</p>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8230
8231<h5>Arguments:</h5>
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8232<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher8295a0a2009-12-23 00:29:49 +0000[diff] [blame]8233 argument is a pointer to or into the <tt>object</tt>. The second argument
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8234 is a boolean 0 or 1. This argument determines whether you want the
8235 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
Eric Christopher8295a0a2009-12-23 00:29:49 +0000[diff] [blame]8236 1, variables are not allowed.</p>
8237
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8238<h5>Semantics:</h5>
8239<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Bill Wendling1b383ba2010-10-27 01:07:41 +0000[diff] [blame]8240 representing the size of the object concerned, or <tt>i32/i64 -1 or 0</tt>,
8241 depending on the <tt>type</tt> argument, if the size cannot be determined at
8242 compile time.</p>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8243
8244</div>
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]8245<!-- _______________________________________________________________________ -->
8246<h4>
8247 <a name="int_expect">'<tt>llvm.expect</tt>' Intrinsic</a>
8248</h4>
Eric Christopher0e671492009-11-30 08:03:53 +0000[diff] [blame]8249
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]8250<div>
8251
8252<h5>Syntax:</h5>
8253<pre>
8254 declare i32 @llvm.expect.i32(i32 &lt;val&gt;, i32 &lt;expected_val&gt;)
8255 declare i64 @llvm.expect.i64(i64 &lt;val&gt;, i64 &lt;expected_val&gt;)
8256</pre>
8257
8258<h5>Overview:</h5>
8259<p>The <tt>llvm.expect</tt> intrinsic provides information about expected (the
8260 most probable) value of <tt>val</tt>, which can be used by optimizers.</p>
8261
8262<h5>Arguments:</h5>
8263<p>The <tt>llvm.expect</tt> intrinsic takes two arguments. The first
8264 argument is a value. The second argument is an expected value, this needs to
8265 be a constant value, variables are not allowed.</p>
8266
8267<h5>Semantics:</h5>
8268<p>This intrinsic is lowered to the <tt>val</tt>.</p>
NAKAMURA Takumif5af6ad2011-04-23 00:30:22 +0000[diff] [blame]8269</div>
8270
8271</div>
8272
Jakub Staszakb170e2d2011-12-04 18:29:26 +0000[diff] [blame]8273</div>
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]8274<!-- *********************************************************************** -->
Chris Lattner00950542001-06-06 20:29:01 +0000[diff] [blame]8275<hr>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8276<address>
8277 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]8278 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8279 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]8280 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]8281
8282 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
NAKAMURA Takumib9a33632011-04-09 02:13:37 +0000[diff] [blame]8283 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
Misha Brukmandaa4cb02004-03-01 17:47:27 +0000[diff] [blame]8284 Last modified: $Date$
8285</address>
Chris Lattnerc7d3ab32008-01-04 04:33:49 +0000[diff] [blame]8286
Misha Brukman9d0919f2003-11-08 01:05:38 +0000[diff] [blame]8287</body>
8288</html>