Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / b9d66609986c1e4aed96e7a30fb4e3ed48511a92 / . / llvm / docs / LangRef.html
blob: 694fed821e2911e84ac7c20e6639832a6785106d [file] [log] [blame]
Misha Brukmanc501f552004-03-01 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 Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
Reid Spencercb84e432004-08-26 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">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]12
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]13<body>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]14
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]15<div class="doc_title"> LLVM Language Reference Manual </div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]16<ol>
Misha Brukman76307852003-11-08 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 Lattner6af02f32004-12-09 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>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a></li>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]24 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]25 <li><a href="#globalvars">Global Variables</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +0000[diff] [blame]26 <li><a href="#functionstructure">Functions</a></li>
Anton Korobeynikov546ea7e2007-04-29 18:02:48 +0000[diff] [blame]27 <li><a href="#aliasstructure">Aliases</a>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]28 <li><a href="#paramattrs">Parameter Attributes</a></li>
Gordon Henriksen71183b62007-12-10 03:18:06 +0000[diff] [blame]29 <li><a href="#gc">Garbage Collector Names</a></li>
Chris Lattner91c15c42006-01-23 23:23:47 +0000[diff] [blame]30 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
Reid Spencer50c723a2007-02-19 23:54:10 +0000[diff] [blame]31 <li><a href="#datalayout">Data Layout</a></li>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]32 </ol>
33 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]34 <li><a href="#typesystem">Type System</a>
35 <ol>
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]36 <li><a href="#t_classifications">Type Classifications</a></li>
Robert Bocchino820bc75b2006-02-17 21:18:08 +0000[diff] [blame]37 <li><a href="#t_primitive">Primitive Types</a>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]38 <ol>
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]39 <li><a href="#t_floating">Floating Point Types</a></li>
40 <li><a href="#t_void">Void Type</a></li>
41 <li><a href="#t_label">Label Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]42 </ol>
43 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]44 <li><a href="#t_derived">Derived Types</a>
45 <ol>
Chris Lattner9a2e3cb2007-12-18 06:18:21 +0000[diff] [blame]46 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]47 <li><a href="#t_array">Array Type</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]48 <li><a href="#t_function">Function Type</a></li>
49 <li><a href="#t_pointer">Pointer Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]50 <li><a href="#t_struct">Structure Type</a></li>
Andrew Lenharth8df88e22006-12-08 17:13:00 +0000[diff] [blame]51 <li><a href="#t_pstruct">Packed Structure Type</a></li>
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]52 <li><a href="#t_vector">Vector Type</a></li>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]53 <li><a href="#t_opaque">Opaque Type</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]54 </ol>
55 </li>
56 </ol>
57 </li>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]58 <li><a href="#constants">Constants</a>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]59 <ol>
60 <li><a href="#simpleconstants">Simple Constants</a>
61 <li><a href="#aggregateconstants">Aggregate Constants</a>
62 <li><a href="#globalconstants">Global Variable and Function Addresses</a>
63 <li><a href="#undefvalues">Undefined Values</a>
64 <li><a href="#constantexprs">Constant Expressions</a>
65 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]66 </li>
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]67 <li><a href="#othervalues">Other Values</a>
68 <ol>
69 <li><a href="#inlineasm">Inline Assembler Expressions</a>
70 </ol>
71 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]72 <li><a href="#instref">Instruction Reference</a>
73 <ol>
74 <li><a href="#terminators">Terminator Instructions</a>
75 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]76 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
77 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]78 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
79 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]80 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]81 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]82 </ol>
83 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]84 <li><a href="#binaryops">Binary Operations</a>
85 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]86 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
87 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
88 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]89 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
90 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
91 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]92 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
93 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
94 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]95 </ol>
96 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]97 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
98 <ol>
Reid Spencer2ab01932007-02-02 13:57:07 +0000[diff] [blame]99 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
100 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
101 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]102 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]103 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]104 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]105 </ol>
106 </li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]107 <li><a href="#vectorops">Vector Operations</a>
108 <ol>
109 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
110 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
111 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]112 </ol>
113 </li>
Dan Gohmanb9d66602008-05-12 23:51:09 +0000[diff] [blame^]114 <li><a href="#aggregateops">Aggregate Operations</a>
115 <ol>
116 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
117 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
118 </ol>
119 </li>
Chris Lattner6ab66722006-08-15 00:45:58 +0000[diff] [blame]120 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]121 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]122 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
123 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
124 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
Robert Bocchino820bc75b2006-02-17 21:18:08 +0000[diff] [blame]125 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
126 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
127 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]128 </ol>
129 </li>
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]130 <li><a href="#convertops">Conversion Operations</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]131 <ol>
132 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
133 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
134 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
135 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
136 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]137 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
138 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
139 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
140 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]141 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
142 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]143 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]144 </ol>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]145 <li><a href="#otherops">Other Operations</a>
146 <ol>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]147 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
148 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
Nate Begemand2195702008-05-12 19:01:56 +0000[diff] [blame]149 <li><a href="#i_vicmp">'<tt>vicmp</tt>' Instruction</a></li>
150 <li><a href="#i_vfcmp">'<tt>vfcmp</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]151 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]152 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]153 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
Chris Lattner33337472006-01-13 23:26:01 +0000[diff] [blame]154 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]155 <li><a href="#i_getresult">'<tt>getresult</tt>' Instruction</a></li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]156 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]157 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]158 </ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]159 </li>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]160 <li><a href="#intrinsics">Intrinsic Functions</a>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]161 <ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]162 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
163 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]164 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
165 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
166 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]167 </ol>
168 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]169 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
170 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]171 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
172 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
173 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]174 </ol>
175 </li>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]176 <li><a href="#int_codegen">Code Generator Intrinsics</a>
177 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]178 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
179 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
180 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
181 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
182 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
183 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
184 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
John Criswellaa1c3c12004-04-09 16:43:20 +0000[diff] [blame]185 </ol>
186 </li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]187 <li><a href="#int_libc">Standard C Library Intrinsics</a>
188 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]189 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
190 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
191 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
192 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
193 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]194 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
195 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]197 </ol>
198 </li>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]199 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]200 <ol>
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]201 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
Chris Lattnerb748c672006-01-16 22:34:14 +0000[diff] [blame]202 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
203 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
204 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Reid Spencer5bf54c82007-04-11 23:23:49 +0000[diff] [blame]205 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
206 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]207 </ol>
208 </li>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]209 <li><a href="#int_debugger">Debugger intrinsics</a></li>
Jim Laskey2211f492007-03-14 19:31:19 +0000[diff] [blame]210 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]211 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]212 <ol>
213 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]214 </ol>
215 </li>
Andrew Lenharth9b254ee2008-02-16 01:24:58 +0000[diff] [blame]216 <li><a href="#int_atomics">Atomic intrinsics</a>
217 <ol>
Andrew Lenharth95528942008-02-21 06:45:13 +0000[diff] [blame]218 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
219 <li><a href="#int_atomic_lcs"><tt>llvm.atomic.lcs</tt></a></li>
220 <li><a href="#int_atomic_las"><tt>llvm.atomic.las</tt></a></li>
221 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
Andrew Lenharth9b254ee2008-02-16 01:24:58 +0000[diff] [blame]222 </ol>
223 </li>
Reid Spencer5b2cb0f2007-07-20 19:59:11 +0000[diff] [blame]224 <li><a href="#int_general">General intrinsics</a>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]225 <ol>
Reid Spencer5b2cb0f2007-07-20 19:59:11 +0000[diff] [blame]226 <li><a href="#int_var_annotation">
Tanya Lattner08abc812007-09-22 00:01:26 +0000[diff] [blame]227 <tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]228 <li><a href="#int_annotation">
Tanya Lattner08abc812007-09-22 00:01:26 +0000[diff] [blame]229 <tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +0000[diff] [blame]230 <li><a href="#int_trap">
231 <tt>llvm.trap</tt>' Intrinsic</a></li>
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]232 </ol>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]233 </li>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]234 </ol>
235 </li>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]236</ol>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]237
238<div class="doc_author">
239 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
240 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]241</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]242
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]243<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]244<div class="doc_section"> <a name="abstract">Abstract </a></div>
245<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]246
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]247<div class="doc_text">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]248<p>This document is a reference manual for the LLVM assembly language.
249LLVM is an SSA based representation that provides type safety,
250low-level operations, flexibility, and the capability of representing
251'all' high-level languages cleanly. It is the common code
252representation used throughout all phases of the LLVM compilation
253strategy.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]254</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]255
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]256<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]257<div class="doc_section"> <a name="introduction">Introduction</a> </div>
258<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]259
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]260<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]261
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]262<p>The LLVM code representation is designed to be used in three
Gabor Greifa54634a2007-07-06 22:07:22 +0000[diff] [blame]263different forms: as an in-memory compiler IR, as an on-disk bitcode
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]264representation (suitable for fast loading by a Just-In-Time compiler),
265and as a human readable assembly language representation. This allows
266LLVM to provide a powerful intermediate representation for efficient
267compiler transformations and analysis, while providing a natural means
268to debug and visualize the transformations. The three different forms
269of LLVM are all equivalent. This document describes the human readable
270representation and notation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]271
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]272<p>The LLVM representation aims to be light-weight and low-level
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]273while being expressive, typed, and extensible at the same time. It
274aims to be a "universal IR" of sorts, by being at a low enough level
275that high-level ideas may be cleanly mapped to it (similar to how
276microprocessors are "universal IR's", allowing many source languages to
277be mapped to them). By providing type information, LLVM can be used as
278the target of optimizations: for example, through pointer analysis, it
279can be proven that a C automatic variable is never accessed outside of
280the current function... allowing it to be promoted to a simple SSA
281value instead of a memory location.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]282
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]283</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]284
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]285<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]286<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]287
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]288<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]289
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]290<p>It is important to note that this document describes 'well formed'
291LLVM assembly language. There is a difference between what the parser
292accepts and what is considered 'well formed'. For example, the
293following instruction is syntactically okay, but not well formed:</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]294
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]295<div class="doc_code">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]296<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]297%x = <a href="#i_add">add</a> i32 1, %x
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]298</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]299</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]300
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]301<p>...because the definition of <tt>%x</tt> does not dominate all of
302its uses. The LLVM infrastructure provides a verification pass that may
303be used to verify that an LLVM module is well formed. This pass is
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]304automatically run by the parser after parsing input assembly and by
Gabor Greifa54634a2007-07-06 22:07:22 +0000[diff] [blame]305the optimizer before it outputs bitcode. The violations pointed out
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]306by the verifier pass indicate bugs in transformation passes or input to
307the parser.</p>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]308</div>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]309
Chris Lattner87a3dbe2007-10-03 17:34:29 +0000[diff] [blame]310<!-- Describe the typesetting conventions here. -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]311
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]312<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]313<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]314<!-- *********************************************************************** -->
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]315
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]316<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]317
Reid Spencerb23b65f2007-08-07 14:34:28 +0000[diff] [blame]318 <p>LLVM identifiers come in two basic types: global and local. Global
319 identifiers (functions, global variables) begin with the @ character. Local
320 identifiers (register names, types) begin with the % character. Additionally,
321 there are three different formats for identifiers, for different purposes:
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]322
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]323<ol>
Reid Spencerb23b65f2007-08-07 14:34:28 +0000[diff] [blame]324 <li>Named values are represented as a string of characters with their prefix.
325 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
326 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]327 Identifiers which require other characters in their names can be surrounded
Reid Spencerb23b65f2007-08-07 14:34:28 +0000[diff] [blame]328 with quotes. In this way, anything except a <tt>&quot;</tt> character can
329 be used in a named value.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]330
Reid Spencerb23b65f2007-08-07 14:34:28 +0000[diff] [blame]331 <li>Unnamed values are represented as an unsigned numeric value with their
332 prefix. For example, %12, @2, %44.</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]333
Reid Spencer8f08d802004-12-09 18:02:53 +0000[diff] [blame]334 <li>Constants, which are described in a <a href="#constants">section about
335 constants</a>, below.</li>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]336</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]337
Reid Spencerb23b65f2007-08-07 14:34:28 +0000[diff] [blame]338<p>LLVM requires that values start with a prefix for two reasons: Compilers
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]339don't need to worry about name clashes with reserved words, and the set of
340reserved words may be expanded in the future without penalty. Additionally,
341unnamed identifiers allow a compiler to quickly come up with a temporary
342variable without having to avoid symbol table conflicts.</p>
343
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]344<p>Reserved words in LLVM are very similar to reserved words in other
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]345languages. There are keywords for different opcodes
346('<tt><a href="#i_add">add</a></tt>',
347 '<tt><a href="#i_bitcast">bitcast</a></tt>',
348 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]349href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]350and others. These reserved words cannot conflict with variable names, because
Reid Spencerb23b65f2007-08-07 14:34:28 +0000[diff] [blame]351none of them start with a prefix character ('%' or '@').</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]352
353<p>Here is an example of LLVM code to multiply the integer variable
354'<tt>%X</tt>' by 8:</p>
355
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]356<p>The easy way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]357
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]358<div class="doc_code">
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]359<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]360%result = <a href="#i_mul">mul</a> i32 %X, 8
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]361</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]362</div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]363
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]364<p>After strength reduction:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]365
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]366<div class="doc_code">
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]367<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]368%result = <a href="#i_shl">shl</a> i32 %X, i8 3
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]369</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]370</div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]371
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]372<p>And the hard way:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]373
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]374<div class="doc_code">
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]375<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]376<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
377<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
378%result = <a href="#i_add">add</a> i32 %1, %1
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]379</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]380</div>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]381
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]382<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
383important lexical features of LLVM:</p>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]384
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]385<ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]386
387 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
388 line.</li>
389
390 <li>Unnamed temporaries are created when the result of a computation is not
391 assigned to a named value.</li>
392
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]393 <li>Unnamed temporaries are numbered sequentially</li>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]394
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]395</ol>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]396
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]397<p>...and it also shows a convention that we follow in this document. When
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]398demonstrating instructions, we will follow an instruction with a comment that
399defines the type and name of value produced. Comments are shown in italic
400text.</p>
401
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]402</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]403
404<!-- *********************************************************************** -->
405<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
406<!-- *********************************************************************** -->
407
408<!-- ======================================================================= -->
409<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
410</div>
411
412<div class="doc_text">
413
414<p>LLVM programs are composed of "Module"s, each of which is a
415translation unit of the input programs. Each module consists of
416functions, global variables, and symbol table entries. Modules may be
417combined together with the LLVM linker, which merges function (and
418global variable) definitions, resolves forward declarations, and merges
419symbol table entries. Here is an example of the "hello world" module:</p>
420
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]421<div class="doc_code">
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]422<pre><i>; Declare the string constant as a global constant...</i>
Chris Lattner2b0bf4f2007-06-12 17:00:26 +0000[diff] [blame]423<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
424 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]425
426<i>; External declaration of the puts function</i>
Chris Lattner2b0bf4f2007-06-12 17:00:26 +0000[diff] [blame]427<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]428
429<i>; Definition of main function</i>
Chris Lattner2b0bf4f2007-06-12 17:00:26 +0000[diff] [blame]430define i32 @main() { <i>; i32()* </i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]431 <i>; Convert [13x i8 ]* to i8 *...</i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]432 %cast210 = <a
Chris Lattner2150cde2007-06-12 17:01:15 +0000[diff] [blame]433 href="#i_getelementptr">getelementptr</a> [13 x i8 ]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]434
435 <i>; Call puts function to write out the string to stdout...</i>
436 <a
Chris Lattner2b0bf4f2007-06-12 17:00:26 +0000[diff] [blame]437 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]438 <a
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]439 href="#i_ret">ret</a> i32 0<br>}<br>
440</pre>
441</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]442
443<p>This example is made up of a <a href="#globalvars">global variable</a>
444named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
445function, and a <a href="#functionstructure">function definition</a>
446for "<tt>main</tt>".</p>
447
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]448<p>In general, a module is made up of a list of global values,
449where both functions and global variables are global values. Global values are
450represented by a pointer to a memory location (in this case, a pointer to an
451array of char, and a pointer to a function), and have one of the following <a
452href="#linkage">linkage types</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]453
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]454</div>
455
456<!-- ======================================================================= -->
457<div class="doc_subsection">
458 <a name="linkage">Linkage Types</a>
459</div>
460
461<div class="doc_text">
462
463<p>
464All Global Variables and Functions have one of the following types of linkage:
465</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]466
467<dl>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]468
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]469 <dt><tt><b><a name="linkage_internal">internal</a></b></tt> </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]470
471 <dd>Global values with internal linkage are only directly accessible by
472 objects in the current module. In particular, linking code into a module with
473 an internal global value may cause the internal to be renamed as necessary to
474 avoid collisions. Because the symbol is internal to the module, all
475 references can be updated. This corresponds to the notion of the
Chris Lattnere20b4702007-01-14 06:51:48 +0000[diff] [blame]476 '<tt>static</tt>' keyword in C.
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]477 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]478
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]479 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]480
Chris Lattnere20b4702007-01-14 06:51:48 +0000[diff] [blame]481 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
482 the same name when linkage occurs. This is typically used to implement
483 inline functions, templates, or other code which must be generated in each
484 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
485 allowed to be discarded.
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]486 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]487
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]488 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]489
490 <dd>"<tt>weak</tt>" linkage is exactly the same as <tt>linkonce</tt> linkage,
491 except that unreferenced <tt>weak</tt> globals may not be discarded. This is
Chris Lattnere20b4702007-01-14 06:51:48 +0000[diff] [blame]492 used for globals that may be emitted in multiple translation units, but that
493 are not guaranteed to be emitted into every translation unit that uses them.
494 One example of this are common globals in C, such as "<tt>int X;</tt>" at
495 global scope.
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]496 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]497
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]498 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]499
500 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
501 pointer to array type. When two global variables with appending linkage are
502 linked together, the two global arrays are appended together. This is the
503 LLVM, typesafe, equivalent of having the system linker append together
504 "sections" with identical names when .o files are linked.
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]505 </dd>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]506
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]507 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
508 <dd>The semantics of this linkage follow the ELF model: the symbol is weak
509 until linked, if not linked, the symbol becomes null instead of being an
510 undefined reference.
511 </dd>
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]512
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]513 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Chris Lattnerd79749a2004-12-09 16:36:40 +0000[diff] [blame]514
515 <dd>If none of the above identifiers are used, the global is externally
516 visible, meaning that it participates in linkage and can be used to resolve
517 external symbol references.
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]518 </dd>
Reid Spencer7972c472007-04-11 23:49:50 +0000[diff] [blame]519</dl>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000[diff] [blame]520
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000[diff] [blame]521 <p>
522 The next two types of linkage are targeted for Microsoft Windows platform
523 only. They are designed to support importing (exporting) symbols from (to)
524 DLLs.
525 </p>
526
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]527 <dl>
Anton Korobeynikovd61d39e2006-09-14 18:23:27 +0000[diff] [blame]528 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
529
530 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
531 or variable via a global pointer to a pointer that is set up by the DLL
532 exporting the symbol. On Microsoft Windows targets, the pointer name is
533 formed by combining <code>_imp__</code> and the function or variable name.
534 </dd>
535
536 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
537
538 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
539 pointer to a pointer in a DLL, so that it can be referenced with the
540 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
541 name is formed by combining <code>_imp__</code> and the function or variable
542 name.
543 </dd>
544
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]545</dl>
546
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]547<p><a name="linkage_external"></a>For example, since the "<tt>.LC0</tt>"
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]548variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
549variable and was linked with this one, one of the two would be renamed,
550preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
551external (i.e., lacking any linkage declarations), they are accessible
Reid Spencer92c671e2007-01-05 00:59:10 +0000[diff] [blame]552outside of the current module.</p>
553<p>It is illegal for a function <i>declaration</i>
554to have any linkage type other than "externally visible", <tt>dllimport</tt>,
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]555or <tt>extern_weak</tt>.</p>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000[diff] [blame]556<p>Aliases can have only <tt>external</tt>, <tt>internal</tt> and <tt>weak</tt>
557linkages.
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]558</div>
559
560<!-- ======================================================================= -->
561<div class="doc_subsection">
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]562 <a name="callingconv">Calling Conventions</a>
563</div>
564
565<div class="doc_text">
566
567<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
568and <a href="#i_invoke">invokes</a> can all have an optional calling convention
569specified for the call. The calling convention of any pair of dynamic
570caller/callee must match, or the behavior of the program is undefined. The
571following calling conventions are supported by LLVM, and more may be added in
572the future:</p>
573
574<dl>
575 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
576
577 <dd>This calling convention (the default if no other calling convention is
578 specified) matches the target C calling conventions. This calling convention
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]579 supports varargs function calls and tolerates some mismatch in the declared
Reid Spencer72ba4992006-12-31 21:30:18 +0000[diff] [blame]580 prototype and implemented declaration of the function (as does normal C).
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]581 </dd>
582
583 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
584
585 <dd>This calling convention attempts to make calls as fast as possible
586 (e.g. by passing things in registers). This calling convention allows the
587 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattnerc792eb32005-05-06 23:08:23 +0000[diff] [blame]588 without having to conform to an externally specified ABI. Implementations of
589 this convention should allow arbitrary tail call optimization to be supported.
590 This calling convention does not support varargs and requires the prototype of
591 all callees to exactly match the prototype of the function definition.
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]592 </dd>
593
594 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
595
596 <dd>This calling convention attempts to make code in the caller as efficient
597 as possible under the assumption that the call is not commonly executed. As
598 such, these calls often preserve all registers so that the call does not break
599 any live ranges in the caller side. This calling convention does not support
600 varargs and requires the prototype of all callees to exactly match the
601 prototype of the function definition.
602 </dd>
603
Chris Lattner573f64e2005-05-07 01:46:40 +0000[diff] [blame]604 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]605
606 <dd>Any calling convention may be specified by number, allowing
607 target-specific calling conventions to be used. Target specific calling
608 conventions start at 64.
609 </dd>
Chris Lattner573f64e2005-05-07 01:46:40 +0000[diff] [blame]610</dl>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]611
612<p>More calling conventions can be added/defined on an as-needed basis, to
613support pascal conventions or any other well-known target-independent
614convention.</p>
615
616</div>
617
618<!-- ======================================================================= -->
619<div class="doc_subsection">
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000[diff] [blame]620 <a name="visibility">Visibility Styles</a>
621</div>
622
623<div class="doc_text">
624
625<p>
626All Global Variables and Functions have one of the following visibility styles:
627</p>
628
629<dl>
630 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
631
632 <dd>On ELF, default visibility means that the declaration is visible to other
633 modules and, in shared libraries, means that the declared entity may be
634 overridden. On Darwin, default visibility means that the declaration is
635 visible to other modules. Default visibility corresponds to "external
636 linkage" in the language.
637 </dd>
638
639 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
640
641 <dd>Two declarations of an object with hidden visibility refer to the same
642 object if they are in the same shared object. Usually, hidden visibility
643 indicates that the symbol will not be placed into the dynamic symbol table,
644 so no other module (executable or shared library) can reference it
645 directly.
646 </dd>
647
Anton Korobeynikov39f3cff2007-04-29 18:35:00 +0000[diff] [blame]648 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
649
650 <dd>On ELF, protected visibility indicates that the symbol will be placed in
651 the dynamic symbol table, but that references within the defining module will
652 bind to the local symbol. That is, the symbol cannot be overridden by another
653 module.
654 </dd>
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000[diff] [blame]655</dl>
656
657</div>
658
659<!-- ======================================================================= -->
660<div class="doc_subsection">
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]661 <a name="globalvars">Global Variables</a>
662</div>
663
664<div class="doc_text">
665
Chris Lattner5d5aede2005-02-12 19:30:21 +0000[diff] [blame]666<p>Global variables define regions of memory allocated at compilation time
Chris Lattner662c8722005-11-12 00:45:07 +0000[diff] [blame]667instead of run-time. Global variables may optionally be initialized, may have
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000[diff] [blame]668an explicit section to be placed in, and may have an optional explicit alignment
669specified. A variable may be defined as "thread_local", which means that it
670will not be shared by threads (each thread will have a separated copy of the
671variable). A variable may be defined as a global "constant," which indicates
672that the contents of the variable will <b>never</b> be modified (enabling better
Chris Lattner5d5aede2005-02-12 19:30:21 +0000[diff] [blame]673optimization, allowing the global data to be placed in the read-only section of
674an executable, etc). Note that variables that need runtime initialization
John Criswell4c0cf7f2005-10-24 16:17:18 +0000[diff] [blame]675cannot be marked "constant" as there is a store to the variable.</p>
Chris Lattner5d5aede2005-02-12 19:30:21 +0000[diff] [blame]676
677<p>
678LLVM explicitly allows <em>declarations</em> of global variables to be marked
679constant, even if the final definition of the global is not. This capability
680can be used to enable slightly better optimization of the program, but requires
681the language definition to guarantee that optimizations based on the
682'constantness' are valid for the translation units that do not include the
683definition.
684</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]685
686<p>As SSA values, global variables define pointer values that are in
687scope (i.e. they dominate) all basic blocks in the program. Global
688variables always define a pointer to their "content" type because they
689describe a region of memory, and all memory objects in LLVM are
690accessed through pointers.</p>
691
Christopher Lamb308121c2007-12-11 09:31:00 +0000[diff] [blame]692<p>A global variable may be declared to reside in a target-specifc numbered
693address space. For targets that support them, address spaces may affect how
694optimizations are performed and/or what target instructions are used to access
Christopher Lamb25f50762007-12-12 08:44:39 +0000[diff] [blame]695the variable. The default address space is zero. The address space qualifier
696must precede any other attributes.</p>
Christopher Lamb308121c2007-12-11 09:31:00 +0000[diff] [blame]697
Chris Lattner662c8722005-11-12 00:45:07 +0000[diff] [blame]698<p>LLVM allows an explicit section to be specified for globals. If the target
699supports it, it will emit globals to the section specified.</p>
700
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]701<p>An explicit alignment may be specified for a global. If not present, or if
702the alignment is set to zero, the alignment of the global is set by the target
703to whatever it feels convenient. If an explicit alignment is specified, the
704global is forced to have at least that much alignment. All alignments must be
705a power of 2.</p>
706
Christopher Lamb308121c2007-12-11 09:31:00 +0000[diff] [blame]707<p>For example, the following defines a global in a numbered address space with
708an initializer, section, and alignment:</p>
Chris Lattner5760c502007-01-14 00:27:09 +0000[diff] [blame]709
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]710<div class="doc_code">
Chris Lattner5760c502007-01-14 00:27:09 +0000[diff] [blame]711<pre>
Christopher Lamb308121c2007-12-11 09:31:00 +0000[diff] [blame]712@G = constant float 1.0 addrspace(5), section "foo", align 4
Chris Lattner5760c502007-01-14 00:27:09 +0000[diff] [blame]713</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]714</div>
Chris Lattner5760c502007-01-14 00:27:09 +0000[diff] [blame]715
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]716</div>
717
718
719<!-- ======================================================================= -->
720<div class="doc_subsection">
721 <a name="functionstructure">Functions</a>
722</div>
723
724<div class="doc_text">
725
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]726<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
727an optional <a href="#linkage">linkage type</a>, an optional
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000[diff] [blame]728<a href="#visibility">visibility style</a>, an optional
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]729<a href="#callingconv">calling convention</a>, a return type, an optional
730<a href="#paramattrs">parameter attribute</a> for the return type, a function
731name, a (possibly empty) argument list (each with optional
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000[diff] [blame]732<a href="#paramattrs">parameter attributes</a>), an optional section, an
Gordon Henriksendc5cafb2007-12-10 03:30:21 +0000[diff] [blame]733optional alignment, an optional <a href="#gc">garbage collector name</a>, an
Gordon Henriksen71183b62007-12-10 03:18:06 +0000[diff] [blame]734opening curly brace, a list of basic blocks, and a closing curly brace.
Anton Korobeynikovc7f9f3d2007-01-23 12:35:46 +0000[diff] [blame]735
736LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
737optional <a href="#linkage">linkage type</a>, an optional
738<a href="#visibility">visibility style</a>, an optional
739<a href="#callingconv">calling convention</a>, a return type, an optional
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]740<a href="#paramattrs">parameter attribute</a> for the return type, a function
Gordon Henriksen71183b62007-12-10 03:18:06 +0000[diff] [blame]741name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksendc5cafb2007-12-10 03:30:21 +0000[diff] [blame]742<a href="#gc">garbage collector name</a>.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]743
744<p>A function definition contains a list of basic blocks, forming the CFG for
745the function. Each basic block may optionally start with a label (giving the
746basic block a symbol table entry), contains a list of instructions, and ends
747with a <a href="#terminators">terminator</a> instruction (such as a branch or
748function return).</p>
749
Chris Lattnera59fb102007-06-08 16:52:14 +0000[diff] [blame]750<p>The first basic block in a function is special in two ways: it is immediately
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]751executed on entrance to the function, and it is not allowed to have predecessor
752basic blocks (i.e. there can not be any branches to the entry block of a
753function). Because the block can have no predecessors, it also cannot have any
754<a href="#i_phi">PHI nodes</a>.</p>
755
Chris Lattner662c8722005-11-12 00:45:07 +0000[diff] [blame]756<p>LLVM allows an explicit section to be specified for functions. If the target
757supports it, it will emit functions to the section specified.</p>
758
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]759<p>An explicit alignment may be specified for a function. If not present, or if
760the alignment is set to zero, the alignment of the function is set by the target
761to whatever it feels convenient. If an explicit alignment is specified, the
762function is forced to have at least that much alignment. All alignments must be
763a power of 2.</p>
764
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]765</div>
766
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000[diff] [blame]767
768<!-- ======================================================================= -->
769<div class="doc_subsection">
770 <a name="aliasstructure">Aliases</a>
771</div>
772<div class="doc_text">
773 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov25b2e822008-03-22 08:36:14 +0000[diff] [blame]774 function, global variable, another alias or bitcast of global value). Aliases
775 may have an optional <a href="#linkage">linkage type</a>, and an
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000[diff] [blame]776 optional <a href="#visibility">visibility style</a>.</p>
777
778 <h5>Syntax:</h5>
779
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]780<div class="doc_code">
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000[diff] [blame]781<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]782@&lt;Name&gt; = [Linkage] [Visibility] alias &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000[diff] [blame]783</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]784</div>
Anton Korobeynikova97b6942007-04-25 14:27:10 +0000[diff] [blame]785
786</div>
787
788
789
Chris Lattner91c15c42006-01-23 23:23:47 +0000[diff] [blame]790<!-- ======================================================================= -->
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]791<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
792<div class="doc_text">
793 <p>The return type and each parameter of a function type may have a set of
794 <i>parameter attributes</i> associated with them. Parameter attributes are
795 used to communicate additional information about the result or parameters of
Duncan Sandsad0ea2d2007-11-27 13:23:08 +0000[diff] [blame]796 a function. Parameter attributes are considered to be part of the function,
797 not of the function type, so functions with different parameter attributes
798 can have the same function type.</p>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]799
Reid Spencercf7ebf52007-01-15 18:27:39 +0000[diff] [blame]800 <p>Parameter attributes are simple keywords that follow the type specified. If
801 multiple parameter attributes are needed, they are space separated. For
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]802 example:</p>
803
804<div class="doc_code">
805<pre>
Duncan Sandsad0ea2d2007-11-27 13:23:08 +0000[diff] [blame]806declare i32 @printf(i8* noalias , ...) nounwind
807declare i32 @atoi(i8*) nounwind readonly
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]808</pre>
809</div>
810
Duncan Sandsad0ea2d2007-11-27 13:23:08 +0000[diff] [blame]811 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
812 <tt>readonly</tt>) come immediately after the argument list.</p>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]813
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]814 <p>Currently, only the following parameter attributes are defined:</p>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]815 <dl>
Reid Spencer314e1cb2007-07-19 23:13:04 +0000[diff] [blame]816 <dt><tt>zeroext</tt></dt>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]817 <dd>This indicates that the parameter should be zero extended just before
818 a call to this function.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]819
Reid Spencer314e1cb2007-07-19 23:13:04 +0000[diff] [blame]820 <dt><tt>signext</tt></dt>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]821 <dd>This indicates that the parameter should be sign extended just before
822 a call to this function.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]823
Anton Korobeynikove8166852007-01-28 14:30:45 +0000[diff] [blame]824 <dt><tt>inreg</tt></dt>
825 <dd>This indicates that the parameter should be placed in register (if
Anton Korobeynikove93c6e82007-01-28 15:27:21 +0000[diff] [blame]826 possible) during assembling function call. Support for this attribute is
827 target-specific</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]828
829 <dt><tt>byval</tt></dt>
Chris Lattner352ab9b2008-01-15 04:34:22 +0000[diff] [blame]830 <dd>This indicates that the pointer parameter should really be passed by
831 value to the function. The attribute implies that a hidden copy of the
832 pointee is made between the caller and the callee, so the callee is unable
833 to modify the value in the callee. This attribute is only valid on llvm
834 pointer arguments. It is generally used to pass structs and arrays by
835 value, but is also valid on scalars (even though this is silly).</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]836
Anton Korobeynikove8166852007-01-28 14:30:45 +0000[diff] [blame]837 <dt><tt>sret</tt></dt>
Duncan Sandsfa4b6732008-02-18 04:19:38 +0000[diff] [blame]838 <dd>This indicates that the pointer parameter specifies the address of a
839 structure that is the return value of the function in the source program.
Duncan Sandsc572c1e2008-03-17 12:17:41 +0000[diff] [blame]840 Loads and stores to the structure are assumed not to trap.
Duncan Sandsfa4b6732008-02-18 04:19:38 +0000[diff] [blame]841 May only be applied to the first parameter.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]842
Zhou Sheng2444a9a2007-06-05 05:28:26 +0000[diff] [blame]843 <dt><tt>noalias</tt></dt>
Owen Anderson61101282008-02-18 04:09:01 +0000[diff] [blame]844 <dd>This indicates that the parameter does not alias any global or any other
845 parameter. The caller is responsible for ensuring that this is the case,
846 usually by placing the value in a stack allocation.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]847
Reid Spencer9d1700e2007-03-22 02:18:56 +0000[diff] [blame]848 <dt><tt>noreturn</tt></dt>
849 <dd>This function attribute indicates that the function never returns. This
850 indicates to LLVM that every call to this function should be treated as if
851 an <tt>unreachable</tt> instruction immediately followed the call.</dd>
Chris Lattner5cee13f2008-01-11 06:20:47 +0000[diff] [blame]852
Reid Spencer05dbb9d2007-03-22 02:02:11 +0000[diff] [blame]853 <dt><tt>nounwind</tt></dt>
Duncan Sandsc572c1e2008-03-17 12:17:41 +0000[diff] [blame]854 <dd>This function attribute indicates that no exceptions unwind out of the
855 function. Usually this is because the function makes no use of exceptions,
856 but it may also be that the function catches any exceptions thrown when
857 executing it.</dd>
858
Duncan Sands27e91592007-07-27 19:57:41 +0000[diff] [blame]859 <dt><tt>nest</tt></dt>
860 <dd>This indicates that the parameter can be excised using the
861 <a href="#int_trampoline">trampoline intrinsics</a>.</dd>
Duncan Sandsa89a1132007-11-22 20:23:04 +0000[diff] [blame]862 <dt><tt>readonly</tt></dt>
Duncan Sands730a3262007-11-14 21:14:02 +0000[diff] [blame]863 <dd>This function attribute indicates that the function has no side-effects
Duncan Sandsa89a1132007-11-22 20:23:04 +0000[diff] [blame]864 except for producing a return value or throwing an exception. The value
865 returned must only depend on the function arguments and/or global variables.
866 It may use values obtained by dereferencing pointers.</dd>
867 <dt><tt>readnone</tt></dt>
868 <dd>A <tt>readnone</tt> function has the same restrictions as a <tt>readonly</tt>
Duncan Sands730a3262007-11-14 21:14:02 +0000[diff] [blame]869 function, but in addition it is not allowed to dereference any pointer arguments
870 or global variables.
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]871 </dl>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]872
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]873</div>
874
875<!-- ======================================================================= -->
Chris Lattner91c15c42006-01-23 23:23:47 +0000[diff] [blame]876<div class="doc_subsection">
Gordon Henriksen71183b62007-12-10 03:18:06 +0000[diff] [blame]877 <a name="gc">Garbage Collector Names</a>
878</div>
879
880<div class="doc_text">
881<p>Each function may specify a garbage collector name, which is simply a
882string.</p>
883
884<div class="doc_code"><pre
885>define void @f() gc "name" { ...</pre></div>
886
887<p>The compiler declares the supported values of <i>name</i>. Specifying a
888collector which will cause the compiler to alter its output in order to support
889the named garbage collection algorithm.</p>
890</div>
891
892<!-- ======================================================================= -->
893<div class="doc_subsection">
Chris Lattner93564892006-04-08 04:40:53 +0000[diff] [blame]894 <a name="moduleasm">Module-Level Inline Assembly</a>
Chris Lattner91c15c42006-01-23 23:23:47 +0000[diff] [blame]895</div>
896
897<div class="doc_text">
898<p>
899Modules may contain "module-level inline asm" blocks, which corresponds to the
900GCC "file scope inline asm" blocks. These blocks are internally concatenated by
901LLVM and treated as a single unit, but may be separated in the .ll file if
902desired. The syntax is very simple:
903</p>
904
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]905<div class="doc_code">
906<pre>
907module asm "inline asm code goes here"
908module asm "more can go here"
909</pre>
910</div>
Chris Lattner91c15c42006-01-23 23:23:47 +0000[diff] [blame]911
912<p>The strings can contain any character by escaping non-printable characters.
913 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
914 for the number.
915</p>
916
917<p>
918 The inline asm code is simply printed to the machine code .s file when
919 assembly code is generated.
920</p>
921</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]922
Reid Spencer50c723a2007-02-19 23:54:10 +0000[diff] [blame]923<!-- ======================================================================= -->
924<div class="doc_subsection">
925 <a name="datalayout">Data Layout</a>
926</div>
927
928<div class="doc_text">
929<p>A module may specify a target specific data layout string that specifies how
Reid Spencer7972c472007-04-11 23:49:50 +0000[diff] [blame]930data is to be laid out in memory. The syntax for the data layout is simply:</p>
931<pre> target datalayout = "<i>layout specification</i>"</pre>
932<p>The <i>layout specification</i> consists of a list of specifications
933separated by the minus sign character ('-'). Each specification starts with a
934letter and may include other information after the letter to define some
935aspect of the data layout. The specifications accepted are as follows: </p>
Reid Spencer50c723a2007-02-19 23:54:10 +0000[diff] [blame]936<dl>
937 <dt><tt>E</tt></dt>
938 <dd>Specifies that the target lays out data in big-endian form. That is, the
939 bits with the most significance have the lowest address location.</dd>
940 <dt><tt>e</tt></dt>
941 <dd>Specifies that hte target lays out data in little-endian form. That is,
942 the bits with the least significance have the lowest address location.</dd>
943 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
944 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
945 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
946 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
947 too.</dd>
948 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
949 <dd>This specifies the alignment for an integer type of a given bit
950 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
951 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
952 <dd>This specifies the alignment for a vector type of a given bit
953 <i>size</i>.</dd>
954 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
955 <dd>This specifies the alignment for a floating point type of a given bit
956 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
957 (double).</dd>
958 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
959 <dd>This specifies the alignment for an aggregate type of a given bit
960 <i>size</i>.</dd>
961</dl>
962<p>When constructing the data layout for a given target, LLVM starts with a
963default set of specifications which are then (possibly) overriden by the
964specifications in the <tt>datalayout</tt> keyword. The default specifications
965are given in this list:</p>
966<ul>
967 <li><tt>E</tt> - big endian</li>
968 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
969 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
970 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
971 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
972 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
973 <li><tt>i64:32:64</tt> - i64 has abi alignment of 32-bits but preferred
974 alignment of 64-bits</li>
975 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
976 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
977 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
978 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
979 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
980</ul>
981<p>When llvm is determining the alignment for a given type, it uses the
982following rules:
983<ol>
984 <li>If the type sought is an exact match for one of the specifications, that
985 specification is used.</li>
986 <li>If no match is found, and the type sought is an integer type, then the
987 smallest integer type that is larger than the bitwidth of the sought type is
988 used. If none of the specifications are larger than the bitwidth then the the
989 largest integer type is used. For example, given the default specifications
990 above, the i7 type will use the alignment of i8 (next largest) while both
991 i65 and i256 will use the alignment of i64 (largest specified).</li>
992 <li>If no match is found, and the type sought is a vector type, then the
993 largest vector type that is smaller than the sought vector type will be used
994 as a fall back. This happens because <128 x double> can be implemented in
995 terms of 64 <2 x double>, for example.</li>
996</ol>
997</div>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]998
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]999<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1000<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1001<!-- *********************************************************************** -->
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]1002
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1003<div class="doc_text">
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]1004
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1005<p>The LLVM type system is one of the most important features of the
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1006intermediate representation. Being typed enables a number of
1007optimizations to be performed on the IR directly, without having to do
1008extra analyses on the side before the transformation. A strong type
1009system makes it easier to read the generated code and enables novel
1010analyses and transformations that are not feasible to perform on normal
1011three address code representations.</p>
Chris Lattner6af02f32004-12-09 16:11:40 +0000[diff] [blame]1012
1013</div>
1014
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1015<!-- ======================================================================= -->
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1016<div class="doc_subsection"> <a name="t_classifications">Type
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1017Classifications</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1018<div class="doc_text">
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1019<p>The types fall into a few useful
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1020classifications:</p>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]1021
1022<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1023 <tbody>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1024 <tr><th>Classification</th><th>Types</th></tr>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1025 <tr>
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1026 <td><a href="#t_integer">integer</a></td>
Reid Spencer138249b2007-05-16 18:44:01 +0000[diff] [blame]1027 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1028 </tr>
1029 <tr>
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1030 <td><a href="#t_floating">floating point</a></td>
1031 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1032 </tr>
1033 <tr>
1034 <td><a name="t_firstclass">first class</a></td>
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1035 <td><a href="#t_integer">integer</a>,
1036 <a href="#t_floating">floating point</a>,
1037 <a href="#t_pointer">pointer</a>,
1038 <a href="#t_vector">vector</a>
Dan Gohmanb9d66602008-05-12 23:51:09 +0000[diff] [blame^]1039 <a href="#t_struct">structure</a>,
1040 <a href="#t_array">array</a>,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1041 </td>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1042 </tr>
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1043 <tr>
1044 <td><a href="#t_primitive">primitive</a></td>
1045 <td><a href="#t_label">label</a>,
1046 <a href="#t_void">void</a>,
1047 <a href="#t_integer">integer</a>,
1048 <a href="#t_floating">floating point</a>.</td>
1049 </tr>
1050 <tr>
1051 <td><a href="#t_derived">derived</a></td>
1052 <td><a href="#t_integer">integer</a>,
1053 <a href="#t_array">array</a>,
1054 <a href="#t_function">function</a>,
1055 <a href="#t_pointer">pointer</a>,
1056 <a href="#t_struct">structure</a>,
1057 <a href="#t_pstruct">packed structure</a>,
1058 <a href="#t_vector">vector</a>,
1059 <a href="#t_opaque">opaque</a>.
1060 </tr>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1061 </tbody>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1062</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]1063
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1064<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1065most important. Values of these types are the only ones which can be
1066produced by instructions, passed as arguments, or used as operands to
1067instructions. This means that all structures and arrays must be
1068manipulated either by pointer or by component.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1069</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1070
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1071<!-- ======================================================================= -->
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1072<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner43542b32008-01-04 04:34:14 +0000[diff] [blame]1073
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1074<div class="doc_text">
1075<p>The primitive types are the fundamental building blocks of the LLVM
1076system.</p>
1077
Chris Lattner43542b32008-01-04 04:34:14 +0000[diff] [blame]1078</div>
1079
Chris Lattner7824d182008-01-04 04:32:38 +0000[diff] [blame]1080<!-- _______________________________________________________________________ -->
1081<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1082
1083<div class="doc_text">
1084 <table>
1085 <tbody>
1086 <tr><th>Type</th><th>Description</th></tr>
1087 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1088 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1089 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1090 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1091 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1092 </tbody>
1093 </table>
1094</div>
1095
1096<!-- _______________________________________________________________________ -->
1097<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1098
1099<div class="doc_text">
1100<h5>Overview:</h5>
1101<p>The void type does not represent any value and has no size.</p>
1102
1103<h5>Syntax:</h5>
1104
1105<pre>
1106 void
1107</pre>
1108</div>
1109
1110<!-- _______________________________________________________________________ -->
1111<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1112
1113<div class="doc_text">
1114<h5>Overview:</h5>
1115<p>The label type represents code labels.</p>
1116
1117<h5>Syntax:</h5>
1118
1119<pre>
1120 label
1121</pre>
1122</div>
1123
1124
1125<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1126<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1127
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1128<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1129
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1130<p>The real power in LLVM comes from the derived types in the system.
1131This is what allows a programmer to represent arrays, functions,
1132pointers, and other useful types. Note that these derived types may be
1133recursive: For example, it is possible to have a two dimensional array.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1134
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1135</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1136
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1137<!-- _______________________________________________________________________ -->
Reid Spencer138249b2007-05-16 18:44:01 +0000[diff] [blame]1138<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1139
1140<div class="doc_text">
1141
1142<h5>Overview:</h5>
1143<p>The integer type is a very simple derived type that simply specifies an
1144arbitrary bit width for the integer type desired. Any bit width from 1 bit to
11452^23-1 (about 8 million) can be specified.</p>
1146
1147<h5>Syntax:</h5>
1148
1149<pre>
1150 iN
1151</pre>
1152
1153<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1154value.</p>
1155
1156<h5>Examples:</h5>
1157<table class="layout">
Chris Lattner9a2e3cb2007-12-18 06:18:21 +0000[diff] [blame]1158 <tbody>
1159 <tr>
1160 <td><tt>i1</tt></td>
1161 <td>a single-bit integer.</td>
1162 </tr><tr>
1163 <td><tt>i32</tt></td>
1164 <td>a 32-bit integer.</td>
1165 </tr><tr>
1166 <td><tt>i1942652</tt></td>
1167 <td>a really big integer of over 1 million bits.</td>
Reid Spencer138249b2007-05-16 18:44:01 +0000[diff] [blame]1168 </tr>
Chris Lattner9a2e3cb2007-12-18 06:18:21 +0000[diff] [blame]1169 </tbody>
Reid Spencer138249b2007-05-16 18:44:01 +0000[diff] [blame]1170</table>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1171</div>
Reid Spencer138249b2007-05-16 18:44:01 +0000[diff] [blame]1172
1173<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1174<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1175
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1176<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1177
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1178<h5>Overview:</h5>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1179
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1180<p>The array type is a very simple derived type that arranges elements
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1181sequentially in memory. The array type requires a size (number of
1182elements) and an underlying data type.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1183
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1184<h5>Syntax:</h5>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1185
1186<pre>
1187 [&lt;# elements&gt; x &lt;elementtype&gt;]
1188</pre>
1189
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]1190<p>The number of elements is a constant integer value; elementtype may
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1191be any type with a size.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1192
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1193<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1194<table class="layout">
1195 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1196 <td class="left"><tt>[40 x i32]</tt></td>
1197 <td class="left">Array of 40 32-bit integer values.</td>
1198 </tr>
1199 <tr class="layout">
1200 <td class="left"><tt>[41 x i32]</tt></td>
1201 <td class="left">Array of 41 32-bit integer values.</td>
1202 </tr>
1203 <tr class="layout">
1204 <td class="left"><tt>[4 x i8]</tt></td>
1205 <td class="left">Array of 4 8-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1206 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1207</table>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1208<p>Here are some examples of multidimensional arrays:</p>
1209<table class="layout">
1210 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1211 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1212 <td class="left">3x4 array of 32-bit integer values.</td>
1213 </tr>
1214 <tr class="layout">
1215 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1216 <td class="left">12x10 array of single precision floating point values.</td>
1217 </tr>
1218 <tr class="layout">
1219 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1220 <td class="left">2x3x4 array of 16-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1221 </tr>
1222</table>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000[diff] [blame]1223
John Criswell4c0cf7f2005-10-24 16:17:18 +0000[diff] [blame]1224<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1225length array. Normally, accesses past the end of an array are undefined in
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000[diff] [blame]1226LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1227As a special case, however, zero length arrays are recognized to be variable
1228length. This allows implementation of 'pascal style arrays' with the LLVM
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1229type "{ i32, [0 x float]}", for example.</p>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000[diff] [blame]1230
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1231</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1232
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1233<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1234<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1235<div class="doc_text">
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1236
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1237<h5>Overview:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1238
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1239<p>The function type can be thought of as a function signature. It
Devang Patele3dfc1c2008-03-24 05:35:41 +0000[diff] [blame]1240consists of a return type and a list of formal parameter types. The
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1241return type of a function type is a scalar type, a void type, or a struct type.
Devang Patel9c1f8b12008-03-24 20:52:42 +0000[diff] [blame]1242If the return type is a struct type then all struct elements must be of first
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1243class types, and the struct must have at least one element.</p>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]1244
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1245<h5>Syntax:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1246
1247<pre>
1248 &lt;returntype list&gt; (&lt;parameter list&gt;)
1249</pre>
1250
John Criswell4c0cf7f2005-10-24 16:17:18 +0000[diff] [blame]1251<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Misha Brukman20f9a622004-08-12 20:16:08 +0000[diff] [blame]1252specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
Chris Lattner5ed60612003-09-03 00:41:47 +0000[diff] [blame]1253which indicates that the function takes a variable number of arguments.
1254Variable argument functions can access their arguments with the <a
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]1255 href="#int_varargs">variable argument handling intrinsic</a> functions.
1256'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1257<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1258
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1259<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1260<table class="layout">
1261 <tr class="layout">
Reid Spencer58c08712006-12-31 07:18:34 +0000[diff] [blame]1262 <td class="left"><tt>i32 (i32)</tt></td>
1263 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1264 </td>
Reid Spencer58c08712006-12-31 07:18:34 +0000[diff] [blame]1265 </tr><tr class="layout">
Reid Spencer314e1cb2007-07-19 23:13:04 +0000[diff] [blame]1266 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Reid Spencer655dcc62006-12-31 07:20:23 +0000[diff] [blame]1267 </tt></td>
Reid Spencer58c08712006-12-31 07:18:34 +0000[diff] [blame]1268 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1269 an <tt>i16</tt> that should be sign extended and a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1270 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
Reid Spencer58c08712006-12-31 07:18:34 +0000[diff] [blame]1271 <tt>float</tt>.
1272 </td>
1273 </tr><tr class="layout">
1274 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1275 <td class="left">A vararg function that takes at least one
Reid Spencer3e628eb92007-01-04 16:43:23 +0000[diff] [blame]1276 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
Reid Spencer58c08712006-12-31 07:18:34 +0000[diff] [blame]1277 which returns an integer. This is the signature for <tt>printf</tt> in
1278 LLVM.
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1279 </td>
Devang Patele3dfc1c2008-03-24 05:35:41 +0000[diff] [blame]1280 </tr><tr class="layout">
1281 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Devang Patel8dec6c22008-03-24 18:10:52 +0000[diff] [blame]1282 <td class="left">A function taking an <tt>i32></tt>, returning two
1283 <tt> i32 </tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Patele3dfc1c2008-03-24 05:35:41 +0000[diff] [blame]1284 </td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1285 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1286</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]1287
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1288</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1289<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1290<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1291<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1292<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1293<p>The structure type is used to represent a collection of data members
1294together in memory. The packing of the field types is defined to match
1295the ABI of the underlying processor. The elements of a structure may
1296be any type that has a size.</p>
1297<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1298and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1299field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1300instruction.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1301<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1302<pre> { &lt;type list&gt; }<br></pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1303<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1304<table class="layout">
1305 <tr class="layout">
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]1306 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1307 <td class="left">A triple of three <tt>i32</tt> values</td>
1308 </tr><tr class="layout">
1309 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1310 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1311 second element is a <a href="#t_pointer">pointer</a> to a
1312 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1313 an <tt>i32</tt>.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1314 </tr>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1315</table>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1316</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1317
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1318<!-- _______________________________________________________________________ -->
Andrew Lenharth8df88e22006-12-08 17:13:00 +0000[diff] [blame]1319<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1320</div>
1321<div class="doc_text">
1322<h5>Overview:</h5>
1323<p>The packed structure type is used to represent a collection of data members
1324together in memory. There is no padding between fields. Further, the alignment
1325of a packed structure is 1 byte. The elements of a packed structure may
1326be any type that has a size.</p>
1327<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1328and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1329field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1330instruction.</p>
1331<h5>Syntax:</h5>
1332<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1333<h5>Examples:</h5>
1334<table class="layout">
1335 <tr class="layout">
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]1336 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1337 <td class="left">A triple of three <tt>i32</tt> values</td>
1338 </tr><tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1339 <td class="left"><tt>&lt; { float, i32 (i32)* } &gt;</tt></td>
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]1340 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1341 second element is a <a href="#t_pointer">pointer</a> to a
1342 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1343 an <tt>i32</tt>.</td>
Andrew Lenharth8df88e22006-12-08 17:13:00 +0000[diff] [blame]1344 </tr>
1345</table>
1346</div>
1347
1348<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1349<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1350<div class="doc_text">
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1351<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1352<p>As in many languages, the pointer type represents a pointer or
Christopher Lamb308121c2007-12-11 09:31:00 +0000[diff] [blame]1353reference to another object, which must live in memory. Pointer types may have
1354an optional address space attribute defining the target-specific numbered
1355address space where the pointed-to object resides. The default address space is
1356zero.</p>
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1357<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1358<pre> &lt;type&gt; *<br></pre>
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1359<h5>Examples:</h5>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1360<table class="layout">
1361 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1362 <td class="left"><tt>[4x i32]*</tt></td>
1363 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1364 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1365 </tr>
1366 <tr class="layout">
1367 <td class="left"><tt>i32 (i32 *) *</tt></td>
1368 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1369 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1370 <tt>i32</tt>.</td>
1371 </tr>
1372 <tr class="layout">
1373 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1374 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1375 that resides in address space #5.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1376 </tr>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1377</table>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1378</div>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1379
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]1380<!-- _______________________________________________________________________ -->
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1381<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1382<div class="doc_text">
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1383
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]1384<h5>Overview:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1385
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1386<p>A vector type is a simple derived type that represents a vector
1387of elements. Vector types are used when multiple primitive data
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]1388are operated in parallel using a single instruction (SIMD).
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1389A vector type requires a size (number of
Chris Lattner330ce692005-11-10 01:44:22 +0000[diff] [blame]1390elements) and an underlying primitive data type. Vectors must have a power
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1391of two length (1, 2, 4, 8, 16 ...). Vector types are
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]1392considered <a href="#t_firstclass">first class</a>.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1393
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]1394<h5>Syntax:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1395
1396<pre>
1397 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1398</pre>
1399
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]1400<p>The number of elements is a constant integer value; elementtype may
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]1401be any integer or floating point type.</p>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1402
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]1403<h5>Examples:</h5>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1404
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1405<table class="layout">
1406 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1407 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1408 <td class="left">Vector of 4 32-bit integer values.</td>
1409 </tr>
1410 <tr class="layout">
1411 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1412 <td class="left">Vector of 8 32-bit floating-point values.</td>
1413 </tr>
1414 <tr class="layout">
1415 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1416 <td class="left">Vector of 2 64-bit integer values.</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]1417 </tr>
1418</table>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1419</div>
1420
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1421<!-- _______________________________________________________________________ -->
1422<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1423<div class="doc_text">
1424
1425<h5>Overview:</h5>
1426
1427<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksena699c4d2007-10-14 00:34:53 +0000[diff] [blame]1428corresponds (for example) to the C notion of a forward declared structure type.
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1429In LLVM, opaque types can eventually be resolved to any type (not just a
1430structure type).</p>
1431
1432<h5>Syntax:</h5>
1433
1434<pre>
1435 opaque
1436</pre>
1437
1438<h5>Examples:</h5>
1439
1440<table class="layout">
1441 <tr class="layout">
Chris Lattner747359f2007-12-19 05:04:11 +0000[diff] [blame]1442 <td class="left"><tt>opaque</tt></td>
1443 <td class="left">An opaque type.</td>
Chris Lattner37b6b092005-04-25 17:34:15 +0000[diff] [blame]1444 </tr>
1445</table>
1446</div>
1447
1448
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1449<!-- *********************************************************************** -->
1450<div class="doc_section"> <a name="constants">Constants</a> </div>
1451<!-- *********************************************************************** -->
1452
1453<div class="doc_text">
1454
1455<p>LLVM has several different basic types of constants. This section describes
1456them all and their syntax.</p>
1457
1458</div>
1459
1460<!-- ======================================================================= -->
Reid Spencer8f08d802004-12-09 18:02:53 +0000[diff] [blame]1461<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1462
1463<div class="doc_text">
1464
1465<dl>
1466 <dt><b>Boolean constants</b></dt>
1467
1468 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]1469 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1470 </dd>
1471
1472 <dt><b>Integer constants</b></dt>
1473
Reid Spencer8f08d802004-12-09 18:02:53 +0000[diff] [blame]1474 <dd>Standard integers (such as '4') are constants of the <a
Reid Spencer3e628eb92007-01-04 16:43:23 +0000[diff] [blame]1475 href="#t_integer">integer</a> type. Negative numbers may be used with
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1476 integer types.
1477 </dd>
1478
1479 <dt><b>Floating point constants</b></dt>
1480
1481 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1482 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattner1429e6f2008-04-01 18:45:27 +0000[diff] [blame]1483 notation (see below). The assembler requires the exact decimal value of
1484 a floating-point constant. For example, the assembler accepts 1.25 but
1485 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1486 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1487
1488 <dt><b>Null pointer constants</b></dt>
1489
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]1490 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1491 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1492
1493</dl>
1494
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]1495<p>The one non-intuitive notation for constants is the optional hexadecimal form
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1496of floating point constants. For example, the form '<tt>double
14970x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
14984.5e+15</tt>'. The only time hexadecimal floating point constants are required
Reid Spencer8f08d802004-12-09 18:02:53 +0000[diff] [blame]1499(and the only time that they are generated by the disassembler) is when a
1500floating point constant must be emitted but it cannot be represented as a
1501decimal floating point number. For example, NaN's, infinities, and other
1502special values are represented in their IEEE hexadecimal format so that
1503assembly and disassembly do not cause any bits to change in the constants.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1504
1505</div>
1506
1507<!-- ======================================================================= -->
1508<div class="doc_subsection"><a name="aggregateconstants">Aggregate Constants</a>
1509</div>
1510
1511<div class="doc_text">
Chris Lattner455fc8c2005-03-07 22:13:59 +0000[diff] [blame]1512<p>Aggregate constants arise from aggregation of simple constants
1513and smaller aggregate constants.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1514
1515<dl>
1516 <dt><b>Structure constants</b></dt>
1517
1518 <dd>Structure constants are represented with notation similar to structure
1519 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattnerbea11172007-12-25 20:34:52 +0000[diff] [blame]1520 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1521 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>". Structure constants
Chris Lattner455fc8c2005-03-07 22:13:59 +0000[diff] [blame]1522 must have <a href="#t_struct">structure type</a>, and the number and
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1523 types of elements must match those specified by the type.
1524 </dd>
1525
1526 <dt><b>Array constants</b></dt>
1527
1528 <dd>Array constants are represented with notation similar to array type
1529 definitions (a comma separated list of elements, surrounded by square brackets
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1530 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1531 constants must have <a href="#t_array">array type</a>, and the number and
1532 types of elements must match those specified by the type.
1533 </dd>
1534
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1535 <dt><b>Vector constants</b></dt>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1536
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1537 <dd>Vector constants are represented with notation similar to vector type
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1538 definitions (a comma separated list of elements, surrounded by
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1539 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]1540 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1541 href="#t_vector">vector type</a>, and the number and types of elements must
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1542 match those specified by the type.
1543 </dd>
1544
1545 <dt><b>Zero initialization</b></dt>
1546
1547 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1548 value to zero of <em>any</em> type, including scalar and aggregate types.
1549 This is often used to avoid having to print large zero initializers (e.g. for
John Criswell4c0cf7f2005-10-24 16:17:18 +0000[diff] [blame]1550 large arrays) and is always exactly equivalent to using explicit zero
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1551 initializers.
1552 </dd>
1553</dl>
1554
1555</div>
1556
1557<!-- ======================================================================= -->
1558<div class="doc_subsection">
1559 <a name="globalconstants">Global Variable and Function Addresses</a>
1560</div>
1561
1562<div class="doc_text">
1563
1564<p>The addresses of <a href="#globalvars">global variables</a> and <a
1565href="#functionstructure">functions</a> are always implicitly valid (link-time)
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]1566constants. These constants are explicitly referenced when the <a
1567href="#identifiers">identifier for the global</a> is used and always have <a
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1568href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1569file:</p>
1570
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1571<div class="doc_code">
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1572<pre>
Chris Lattner00538a12007-06-06 18:28:13 +0000[diff] [blame]1573@X = global i32 17
1574@Y = global i32 42
1575@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1576</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1577</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1578
1579</div>
1580
1581<!-- ======================================================================= -->
Reid Spencer641f5c92004-12-09 18:13:12 +0000[diff] [blame]1582<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1583<div class="doc_text">
Reid Spencer641f5c92004-12-09 18:13:12 +0000[diff] [blame]1584 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]1585 no specific value. Undefined values may be of any type and be used anywhere
Reid Spencer641f5c92004-12-09 18:13:12 +0000[diff] [blame]1586 a constant is permitted.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1587
Reid Spencer641f5c92004-12-09 18:13:12 +0000[diff] [blame]1588 <p>Undefined values indicate to the compiler that the program is well defined
1589 no matter what value is used, giving the compiler more freedom to optimize.
1590 </p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1591</div>
1592
1593<!-- ======================================================================= -->
1594<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1595</div>
1596
1597<div class="doc_text">
1598
1599<p>Constant expressions are used to allow expressions involving other constants
1600to be used as constants. Constant expressions may be of any <a
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]1601href="#t_firstclass">first class</a> type and may involve any LLVM operation
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1602that does not have side effects (e.g. load and call are not supported). The
1603following is the syntax for constant expressions:</p>
1604
1605<dl>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1606 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1607 <dd>Truncate a constant to another type. The bit size of CST must be larger
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]1608 than the bit size of TYPE. Both types must be integers.</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1609
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1610 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1611 <dd>Zero extend a constant to another type. The bit size of CST must be
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]1612 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1613
1614 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1615 <dd>Sign extend a constant to another type. The bit size of CST must be
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]1616 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1617
1618 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1619 <dd>Truncate a floating point constant to another floating point type. The
1620 size of CST must be larger than the size of TYPE. Both types must be
1621 floating point.</dd>
1622
1623 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1624 <dd>Floating point extend a constant to another type. The size of CST must be
1625 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1626
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]1627 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1628 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]1629 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1630 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1631 of the same number of elements. If the value won't fit in the integer type,
1632 the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1633
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]1634 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1635 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]1636 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1637 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1638 of the same number of elements. If the value won't fit in the integer type,
1639 the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1640
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]1641 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1642 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]1643 constant. TYPE must be a scalar or vector floating point type. CST must be of
1644 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1645 of the same number of elements. If the value won't fit in the floating point
1646 type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1647
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]1648 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1649 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]1650 constant. TYPE must be a scalar or vector floating point type. CST must be of
1651 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1652 of the same number of elements. If the value won't fit in the floating point
1653 type, the results are undefined.</dd>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1654
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]1655 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
1656 <dd>Convert a pointer typed constant to the corresponding integer constant
1657 TYPE must be an integer type. CST must be of pointer type. The CST value is
1658 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
1659
1660 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
1661 <dd>Convert a integer constant to a pointer constant. TYPE must be a
1662 pointer type. CST must be of integer type. The CST value is zero extended,
1663 truncated, or unchanged to make it fit in a pointer size. This one is
1664 <i>really</i> dangerous!</dd>
1665
1666 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1667 <dd>Convert a constant, CST, to another TYPE. The size of CST and TYPE must be
1668 identical (same number of bits). The conversion is done as if the CST value
1669 was stored to memory and read back as TYPE. In other words, no bits change
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]1670 with this operator, just the type. This can be used for conversion of
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]1671 vector types to any other type, as long as they have the same bit width. For
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]1672 pointers it is only valid to cast to another pointer type.
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]1673 </dd>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1674
1675 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
1676
1677 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
1678 constants. As with the <a href="#i_getelementptr">getelementptr</a>
1679 instruction, the index list may have zero or more indexes, which are required
1680 to make sense for the type of "CSTPTR".</dd>
1681
Robert Bocchino7e97a6d2006-01-10 19:31:34 +0000[diff] [blame]1682 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
1683
1684 <dd>Perform the <a href="#i_select">select operation</a> on
Reid Spencer9965ee72006-12-04 19:23:19 +0000[diff] [blame]1685 constants.</dd>
1686
1687 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
1688 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
1689
1690 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
1691 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
Robert Bocchino7e97a6d2006-01-10 19:31:34 +0000[diff] [blame]1692
Nate Begemand2195702008-05-12 19:01:56 +0000[diff] [blame]1693 <dt><b><tt>vicmp COND ( VAL1, VAL2 )</tt></b></dt>
1694 <dd>Performs the <a href="#i_vicmp">vicmp operation</a> on constants.</dd>
1695
1696 <dt><b><tt>vfcmp COND ( VAL1, VAL2 )</tt></b></dt>
1697 <dd>Performs the <a href="#i_vfcmp">vfcmp operation</a> on constants.</dd>
1698
Robert Bocchino7e97a6d2006-01-10 19:31:34 +0000[diff] [blame]1699 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
1700
1701 <dd>Perform the <a href="#i_extractelement">extractelement
1702 operation</a> on constants.
1703
Robert Bocchinof72fdfe2006-01-15 20:48:27 +0000[diff] [blame]1704 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
1705
1706 <dd>Perform the <a href="#i_insertelement">insertelement
Reid Spencer9965ee72006-12-04 19:23:19 +0000[diff] [blame]1707 operation</a> on constants.</dd>
Robert Bocchinof72fdfe2006-01-15 20:48:27 +0000[diff] [blame]1708
Chris Lattner016a0e52006-04-08 00:13:41 +0000[diff] [blame]1709
1710 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
1711
1712 <dd>Perform the <a href="#i_shufflevector">shufflevector
Reid Spencer9965ee72006-12-04 19:23:19 +0000[diff] [blame]1713 operation</a> on constants.</dd>
Chris Lattner016a0e52006-04-08 00:13:41 +0000[diff] [blame]1714
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1715 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
1716
Reid Spencer641f5c92004-12-09 18:13:12 +0000[diff] [blame]1717 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
1718 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1719 binary</a> operations. The constraints on operands are the same as those for
1720 the corresponding instruction (e.g. no bitwise operations on floating point
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]1721 values are allowed).</dd>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1722</dl>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1723</div>
Chris Lattnerb1652612004-03-08 16:49:10 +0000[diff] [blame]1724
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1725<!-- *********************************************************************** -->
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1726<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
1727<!-- *********************************************************************** -->
1728
1729<!-- ======================================================================= -->
1730<div class="doc_subsection">
1731<a name="inlineasm">Inline Assembler Expressions</a>
1732</div>
1733
1734<div class="doc_text">
1735
1736<p>
1737LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
1738Module-Level Inline Assembly</a>) through the use of a special value. This
1739value represents the inline assembler as a string (containing the instructions
1740to emit), a list of operand constraints (stored as a string), and a flag that
1741indicates whether or not the inline asm expression has side effects. An example
1742inline assembler expression is:
1743</p>
1744
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1745<div class="doc_code">
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1746<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1747i32 (i32) asm "bswap $0", "=r,r"
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1748</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1749</div>
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1750
1751<p>
1752Inline assembler expressions may <b>only</b> be used as the callee operand of
1753a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
1754</p>
1755
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1756<div class="doc_code">
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1757<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1758%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1759</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1760</div>
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1761
1762<p>
1763Inline asms with side effects not visible in the constraint list must be marked
1764as having side effects. This is done through the use of the
1765'<tt>sideeffect</tt>' keyword, like so:
1766</p>
1767
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1768<div class="doc_code">
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1769<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1770call void asm sideeffect "eieio", ""()
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1771</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]1772</div>
Chris Lattner98f013c2006-01-25 23:47:57 +0000[diff] [blame]1773
1774<p>TODO: The format of the asm and constraints string still need to be
1775documented here. Constraints on what can be done (e.g. duplication, moving, etc
1776need to be documented).
1777</p>
1778
1779</div>
1780
1781<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1782<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
1783<!-- *********************************************************************** -->
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1784
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1785<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1786
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1787<p>The LLVM instruction set consists of several different
1788classifications of instructions: <a href="#terminators">terminator
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]1789instructions</a>, <a href="#binaryops">binary instructions</a>,
1790<a href="#bitwiseops">bitwise binary instructions</a>, <a
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1791 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
1792instructions</a>.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1793
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1794</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1795
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1796<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1797<div class="doc_subsection"> <a name="terminators">Terminator
1798Instructions</a> </div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1799
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1800<div class="doc_text">
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1801
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1802<p>As mentioned <a href="#functionstructure">previously</a>, every
1803basic block in a program ends with a "Terminator" instruction, which
1804indicates which block should be executed after the current block is
1805finished. These terminator instructions typically yield a '<tt>void</tt>'
1806value: they produce control flow, not values (the one exception being
1807the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]1808<p>There are six different terminator instructions: the '<a
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1809 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
1810instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]1811the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
1812 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
1813 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1814
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1815</div>
Chris Lattner74d3f822004-12-09 17:30:23 +0000[diff] [blame]1816
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1817<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1818<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
1819Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1820<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1821<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1822<pre> ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1823 ret void <i>; Return from void function</i>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]1824 ret &lt;type&gt; &lt;value&gt;, &lt;type&gt; &lt;value&gt; <i>; Return two values from a non-void function </i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1825</pre>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1826
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1827<h5>Overview:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1828
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1829<p>The '<tt>ret</tt>' instruction is used to return control flow (and a
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]1830value) from a function back to the caller.</p>
John Criswell417228d2004-04-09 16:48:45 +0000[diff] [blame]1831<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1832returns value(s) and then causes control flow, and one that just causes
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1833control flow to occur.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1834
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1835<h5>Arguments:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1836
1837<p>The '<tt>ret</tt>' instruction may return zero, one or multiple values.
1838The type of each return value must be a '<a href="#t_firstclass">first
1839class</a>' type. Note that a function is not <a href="#wellformed">well
1840formed</a> if there exists a '<tt>ret</tt>' instruction inside of the
1841function that returns values that do not match the return type of the
1842function.</p>
1843
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1844<h5>Semantics:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1845
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1846<p>When the '<tt>ret</tt>' instruction is executed, control flow
1847returns back to the calling function's context. If the caller is a "<a
John Criswell40db33f2004-06-25 15:16:57 +0000[diff] [blame]1848 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1849the instruction after the call. If the caller was an "<a
1850 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]1851at the beginning of the "normal" destination block. If the instruction
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1852returns a value, that value shall set the call or invoke instruction's
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]1853return value. If the instruction returns multiple values then these
Devang Pateld0f47642008-03-11 05:51:59 +0000[diff] [blame]1854values can only be accessed through a '<a href="#i_getresult"><tt>getresult</tt>
1855</a>' instruction.</p>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1856
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1857<h5>Example:</h5>
Chris Lattnerda508ac2008-04-23 04:59:35 +0000[diff] [blame]1858
1859<pre>
1860 ret i32 5 <i>; Return an integer value of 5</i>
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]1861 ret void <i>; Return from a void function</i>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]1862 ret i32 4, i8 2 <i>; Return two values 4 and 2 </i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1863</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1864</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1865<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1866<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1867<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1868<h5>Syntax:</h5>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]1869<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1870</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1871<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1872<p>The '<tt>br</tt>' instruction is used to cause control flow to
1873transfer to a different basic block in the current function. There are
1874two forms of this instruction, corresponding to a conditional branch
1875and an unconditional branch.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1876<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1877<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]1878single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
Reid Spencer50c723a2007-02-19 23:54:10 +0000[diff] [blame]1879unconditional form of the '<tt>br</tt>' instruction takes a single
1880'<tt>label</tt>' value as a target.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1881<h5>Semantics:</h5>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]1882<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1883argument is evaluated. If the value is <tt>true</tt>, control flows
1884to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
1885control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1886<h5>Example:</h5>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]1887<pre>Test:<br> %cond = <a href="#i_icmp">icmp</a> eq, i32 %a, %b<br> br i1 %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1888 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1889</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1890<!-- _______________________________________________________________________ -->
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1891<div class="doc_subsubsection">
1892 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
1893</div>
1894
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1895<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1896<h5>Syntax:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1897
1898<pre>
1899 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
1900</pre>
1901
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1902<h5>Overview:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1903
1904<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
1905several different places. It is a generalization of the '<tt>br</tt>'
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1906instruction, allowing a branch to occur to one of many possible
1907destinations.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1908
1909
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1910<h5>Arguments:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1911
1912<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
1913comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
1914an array of pairs of comparison value constants and '<tt>label</tt>'s. The
1915table is not allowed to contain duplicate constant entries.</p>
1916
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1917<h5>Semantics:</h5>
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1918
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]1919<p>The <tt>switch</tt> instruction specifies a table of values and
1920destinations. When the '<tt>switch</tt>' instruction is executed, this
John Criswellbcbb18c2004-06-25 16:05:06 +0000[diff] [blame]1921table is searched for the given value. If the value is found, control flow is
1922transfered to the corresponding destination; otherwise, control flow is
1923transfered to the default destination.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1924
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1925<h5>Implementation:</h5>
1926
1927<p>Depending on properties of the target machine and the particular
1928<tt>switch</tt> instruction, this instruction may be code generated in different
John Criswellbcbb18c2004-06-25 16:05:06 +0000[diff] [blame]1929ways. For example, it could be generated as a series of chained conditional
1930branches or with a lookup table.</p>
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1931
1932<h5>Example:</h5>
1933
1934<pre>
1935 <i>; Emulate a conditional br instruction</i>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]1936 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1937 switch i32 %Val, label %truedest [i32 0, label %falsedest ]
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1938
1939 <i>; Emulate an unconditional br instruction</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1940 switch i32 0, label %dest [ ]
Chris Lattnercf96c6c2004-02-24 04:54:45 +0000[diff] [blame]1941
1942 <i>; Implement a jump table:</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]1943 switch i32 %val, label %otherwise [ i32 0, label %onzero
1944 i32 1, label %onone
1945 i32 2, label %ontwo ]
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1946</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1947</div>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1948
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1949<!-- _______________________________________________________________________ -->
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1950<div class="doc_subsubsection">
1951 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
1952</div>
1953
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1954<div class="doc_text">
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1955
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1956<h5>Syntax:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1957
1958<pre>
Nick Lewycky084ab472008-03-16 07:18:12 +0000[diff] [blame]1959 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;)
Chris Lattner6b7a0082006-05-14 18:23:06 +0000[diff] [blame]1960 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1961</pre>
1962
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]1963<h5>Overview:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1964
1965<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
1966function, with the possibility of control flow transfer to either the
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]1967'<tt>normal</tt>' label or the
1968'<tt>exception</tt>' label. If the callee function returns with the
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1969"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
1970"normal" label. If the callee (or any indirect callees) returns with the "<a
John Criswell02fdc6f2005-05-12 16:52:32 +0000[diff] [blame]1971href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]1972continued at the dynamically nearest "exception" label. If the callee function
Devang Pateld0f47642008-03-11 05:51:59 +0000[diff] [blame]1973returns multiple values then individual return values are only accessible through
1974a '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1975
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1976<h5>Arguments:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1977
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]1978<p>This instruction requires several arguments:</p>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1979
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]1980<ol>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1981 <li>
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]1982 The optional "cconv" marker indicates which <a href="#callingconv">calling
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]1983 convention</a> the call should use. If none is specified, the call defaults
1984 to using C calling conventions.
1985 </li>
1986 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
1987 function value being invoked. In most cases, this is a direct function
1988 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
1989 an arbitrary pointer to function value.
1990 </li>
1991
1992 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
1993 function to be invoked. </li>
1994
1995 <li>'<tt>function args</tt>': argument list whose types match the function
1996 signature argument types. If the function signature indicates the function
1997 accepts a variable number of arguments, the extra arguments can be
1998 specified. </li>
1999
2000 <li>'<tt>normal label</tt>': the label reached when the called function
2001 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2002
2003 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2004 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2005
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2006</ol>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]2007
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2008<h5>Semantics:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]2009
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2010<p>This instruction is designed to operate as a standard '<tt><a
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]2011href="#i_call">call</a></tt>' instruction in most regards. The primary
2012difference is that it establishes an association with a label, which is used by
2013the runtime library to unwind the stack.</p>
2014
2015<p>This instruction is used in languages with destructors to ensure that proper
2016cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2017exception. Additionally, this is important for implementation of
2018'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2019
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2020<h5>Example:</h5>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]2021<pre>
Nick Lewycky084ab472008-03-16 07:18:12 +0000[diff] [blame]2022 %retval = invoke i32 @Test(i32 15) to label %Continue
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2023 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewycky084ab472008-03-16 07:18:12 +0000[diff] [blame]2024 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2025 unwind label %TestCleanup <i>; {i32}:retval set</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2026</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2027</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2028
2029
Chris Lattner5ed60612003-09-03 00:41:47 +0000[diff] [blame]2030<!-- _______________________________________________________________________ -->
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2031
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2032<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2033Instruction</a> </div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2034
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2035<div class="doc_text">
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2036
Chris Lattner5ed60612003-09-03 00:41:47 +0000[diff] [blame]2037<h5>Syntax:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2038<pre>
2039 unwind
2040</pre>
2041
Chris Lattner5ed60612003-09-03 00:41:47 +0000[diff] [blame]2042<h5>Overview:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2043
2044<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2045at the first callee in the dynamic call stack which used an <a
2046href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2047primarily used to implement exception handling.</p>
2048
Chris Lattner5ed60612003-09-03 00:41:47 +0000[diff] [blame]2049<h5>Semantics:</h5>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2050
Chris Lattnerfe8519c2008-04-19 21:01:16 +0000[diff] [blame]2051<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2052immediately halt. The dynamic call stack is then searched for the first <a
2053href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2054execution continues at the "exceptional" destination block specified by the
2055<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2056dynamic call chain, undefined behavior results.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2057</div>
Chris Lattner08b7d5b2004-10-16 18:04:13 +0000[diff] [blame]2058
2059<!-- _______________________________________________________________________ -->
2060
2061<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2062Instruction</a> </div>
2063
2064<div class="doc_text">
2065
2066<h5>Syntax:</h5>
2067<pre>
2068 unreachable
2069</pre>
2070
2071<h5>Overview:</h5>
2072
2073<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2074instruction is used to inform the optimizer that a particular portion of the
2075code is not reachable. This can be used to indicate that the code after a
2076no-return function cannot be reached, and other facts.</p>
2077
2078<h5>Semantics:</h5>
2079
2080<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2081</div>
2082
2083
2084
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2085<!-- ======================================================================= -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2086<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2087<div class="doc_text">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2088<p>Binary operators are used to do most of the computation in a
Chris Lattner81f92972008-04-01 18:47:32 +0000[diff] [blame]2089program. They require two operands of the same type, execute an operation on them, and
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]2090produce a single value. The operands might represent
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2091multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
Chris Lattner81f92972008-04-01 18:47:32 +0000[diff] [blame]2092The result value has the same type as its operands.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2093<p>There are several different binary operators:</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2094</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2095<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2096<div class="doc_subsubsection"> <a name="i_add">'<tt>add</tt>'
2097Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2098<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2099<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2100<pre> &lt;result&gt; = add &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2101</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2102<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2103<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2104<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2105<p>The two arguments to the '<tt>add</tt>' instruction must be either <a
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]2106 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a> values.
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2107 This instruction can also take <a href="#t_vector">vector</a> versions of the values.
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]2108Both arguments must have identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2109<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2110<p>The value produced is the integer or floating point sum of the two
2111operands.</p>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2112<p>If an integer sum has unsigned overflow, the result returned is the
2113mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2114the result.</p>
2115<p>Because LLVM integers use a two's complement representation, this
2116instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2117<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2118<pre> &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2119</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2120</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2121<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2122<div class="doc_subsubsection"> <a name="i_sub">'<tt>sub</tt>'
2123Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2124<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2125<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2126<pre> &lt;result&gt; = sub &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2127</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2128<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2129<p>The '<tt>sub</tt>' instruction returns the difference of its two
2130operands.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2131<p>Note that the '<tt>sub</tt>' instruction is used to represent the '<tt>neg</tt>'
2132instruction present in most other intermediate representations.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2133<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2134<p>The two arguments to the '<tt>sub</tt>' instruction must be either <a
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2135 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]2136values.
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2137This instruction can also take <a href="#t_vector">vector</a> versions of the values.
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]2138Both arguments must have identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2139<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2140<p>The value produced is the integer or floating point difference of
2141the two operands.</p>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2142<p>If an integer difference has unsigned overflow, the result returned is the
2143mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2144the result.</p>
2145<p>Because LLVM integers use a two's complement representation, this
2146instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2147<h5>Example:</h5>
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000[diff] [blame]2148<pre>
2149 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2150 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2151</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2152</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2153<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2154<div class="doc_subsubsection"> <a name="i_mul">'<tt>mul</tt>'
2155Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2156<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2157<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2158<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2159</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2160<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2161<p>The '<tt>mul</tt>' instruction returns the product of its two
2162operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2163<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2164<p>The two arguments to the '<tt>mul</tt>' instruction must be either <a
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2165 href="#t_integer">integer</a> or <a href="#t_floating">floating point</a>
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]2166values.
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2167This instruction can also take <a href="#t_vector">vector</a> versions of the values.
Chris Lattnerc8cb6952004-08-12 19:12:28 +0000[diff] [blame]2168Both arguments must have identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2169<h5>Semantics:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2170<p>The value produced is the integer or floating point product of the
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2171two operands.</p>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2172<p>If the result of an integer multiplication has unsigned overflow,
2173the result returned is the mathematical result modulo
21742<sup>n</sup>, where n is the bit width of the result.</p>
2175<p>Because LLVM integers use a two's complement representation, and the
2176result is the same width as the operands, this instruction returns the
2177correct result for both signed and unsigned integers. If a full product
2178(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2179should be sign-extended or zero-extended as appropriate to the
2180width of the full product.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2181<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2182<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2183</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2184</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2185<!-- _______________________________________________________________________ -->
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2186<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2187</a></div>
2188<div class="doc_text">
2189<h5>Syntax:</h5>
2190<pre> &lt;result&gt; = udiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2191</pre>
2192<h5>Overview:</h5>
2193<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2194operands.</p>
2195<h5>Arguments:</h5>
2196<p>The two arguments to the '<tt>udiv</tt>' instruction must be
2197<a href="#t_integer">integer</a> values. Both arguments must have identical
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2198types. This instruction can also take <a href="#t_vector">vector</a> versions
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2199of the values in which case the elements must be integers.</p>
2200<h5>Semantics:</h5>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2201<p>The value produced is the unsigned integer quotient of the two operands.</p>
2202<p>Note that unsigned integer division and signed integer division are distinct
2203operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2204<p>Division by zero leads to undefined behavior.</p>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2205<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2206<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2207</pre>
2208</div>
2209<!-- _______________________________________________________________________ -->
2210<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2211</a> </div>
2212<div class="doc_text">
2213<h5>Syntax:</h5>
2214<pre> &lt;result&gt; = sdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2215</pre>
2216<h5>Overview:</h5>
2217<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2218operands.</p>
2219<h5>Arguments:</h5>
2220<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2221<a href="#t_integer">integer</a> values. Both arguments must have identical
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2222types. This instruction can also take <a href="#t_vector">vector</a> versions
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2223of the values in which case the elements must be integers.</p>
2224<h5>Semantics:</h5>
Chris Lattner1429e6f2008-04-01 18:45:27 +0000[diff] [blame]2225<p>The value produced is the signed integer quotient of the two operands rounded towards zero.</p>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2226<p>Note that signed integer division and unsigned integer division are distinct
2227operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2228<p>Division by zero leads to undefined behavior. Overflow also leads to
2229undefined behavior; this is a rare case, but can occur, for example,
2230by doing a 32-bit division of -2147483648 by -1.</p>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2231<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2232<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2233</pre>
2234</div>
2235<!-- _______________________________________________________________________ -->
2236<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2237Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2238<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2239<h5>Syntax:</h5>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2240<pre> &lt;result&gt; = fdiv &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2241</pre>
2242<h5>Overview:</h5>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2243<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2244operands.</p>
2245<h5>Arguments:</h5>
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2246<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2247<a href="#t_floating">floating point</a> values. Both arguments must have
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2248identical types. This instruction can also take <a href="#t_vector">vector</a>
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2249versions of floating point values.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2250<h5>Semantics:</h5>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2251<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2252<h5>Example:</h5>
Reid Spencer7e80b0b2006-10-26 06:15:43 +0000[diff] [blame]2253<pre> &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2254</pre>
2255</div>
2256<!-- _______________________________________________________________________ -->
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2257<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2258</div>
2259<div class="doc_text">
2260<h5>Syntax:</h5>
2261<pre> &lt;result&gt; = urem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2262</pre>
2263<h5>Overview:</h5>
2264<p>The '<tt>urem</tt>' instruction returns the remainder from the
2265unsigned division of its two arguments.</p>
2266<h5>Arguments:</h5>
2267<p>The two arguments to the '<tt>urem</tt>' instruction must be
2268<a href="#t_integer">integer</a> values. Both arguments must have identical
Dan Gohman08143e32007-11-05 23:35:22 +0000[diff] [blame]2269types. This instruction can also take <a href="#t_vector">vector</a> versions
2270of the values in which case the elements must be integers.</p>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2271<h5>Semantics:</h5>
2272<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattner1429e6f2008-04-01 18:45:27 +0000[diff] [blame]2273This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2274<p>Note that unsigned integer remainder and signed integer remainder are
2275distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2276<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2277<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2278<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2279</pre>
2280
2281</div>
2282<!-- _______________________________________________________________________ -->
2283<div class="doc_subsubsection"> <a name="i_srem">'<tt>srem</tt>'
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2284Instruction</a> </div>
2285<div class="doc_text">
2286<h5>Syntax:</h5>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2287<pre> &lt;result&gt; = srem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2288</pre>
2289<h5>Overview:</h5>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2290<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman08143e32007-11-05 23:35:22 +0000[diff] [blame]2291signed division of its two operands. This instruction can also take
2292<a href="#t_vector">vector</a> versions of the values in which case
2293the elements must be integers.</p>
Chris Lattnerb8f816e2008-01-04 04:33:49 +0000[diff] [blame]2294
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2295<h5>Arguments:</h5>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2296<p>The two arguments to the '<tt>srem</tt>' instruction must be
2297<a href="#t_integer">integer</a> values. Both arguments must have identical
2298types.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2299<h5>Semantics:</h5>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2300<p>This instruction returns the <i>remainder</i> of a division (where the result
Reid Spencer806ad6a2007-03-24 22:23:39 +0000[diff] [blame]2301has the same sign as the dividend, <tt>var1</tt>), not the <i>modulo</i>
2302operator (where the result has the same sign as the divisor, <tt>var2</tt>) of
2303a value. For more information about the difference, see <a
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2304 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
Reid Spencer806ad6a2007-03-24 22:23:39 +0000[diff] [blame]2305Math Forum</a>. For a table of how this is implemented in various languages,
Reid Spencerdb3b93b2007-03-24 22:40:44 +0000[diff] [blame]2306please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
Reid Spencer806ad6a2007-03-24 22:23:39 +0000[diff] [blame]2307Wikipedia: modulo operation</a>.</p>
Chris Lattner2f2427e2008-01-28 00:36:27 +0000[diff] [blame]2308<p>Note that signed integer remainder and unsigned integer remainder are
2309distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2310<p>Taking the remainder of a division by zero leads to undefined behavior.
2311Overflow also leads to undefined behavior; this is a rare case, but can occur,
2312for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2313(The remainder doesn't actually overflow, but this rule lets srem be
2314implemented using instructions that return both the result of the division
2315and the remainder.)</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2316<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2317<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2318</pre>
2319
2320</div>
2321<!-- _______________________________________________________________________ -->
2322<div class="doc_subsubsection"> <a name="i_frem">'<tt>frem</tt>'
2323Instruction</a> </div>
2324<div class="doc_text">
2325<h5>Syntax:</h5>
2326<pre> &lt;result&gt; = frem &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2327</pre>
2328<h5>Overview:</h5>
2329<p>The '<tt>frem</tt>' instruction returns the remainder from the
2330division of its two operands.</p>
2331<h5>Arguments:</h5>
2332<p>The two arguments to the '<tt>frem</tt>' instruction must be
2333<a href="#t_floating">floating point</a> values. Both arguments must have
Dan Gohman08143e32007-11-05 23:35:22 +0000[diff] [blame]2334identical types. This instruction can also take <a href="#t_vector">vector</a>
2335versions of floating point values.</p>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2336<h5>Semantics:</h5>
Chris Lattner1429e6f2008-04-01 18:45:27 +0000[diff] [blame]2337<p>This instruction returns the <i>remainder</i> of a division.
2338The remainder has the same sign as the dividend.</p>
Reid Spencer7eb55b32006-11-02 01:53:59 +0000[diff] [blame]2339<h5>Example:</h5>
2340<pre> &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2341</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2342</div>
Robert Bocchino820bc75b2006-02-17 21:18:08 +0000[diff] [blame]2343
Reid Spencer2ab01932007-02-02 13:57:07 +0000[diff] [blame]2344<!-- ======================================================================= -->
2345<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2346Operations</a> </div>
2347<div class="doc_text">
2348<p>Bitwise binary operators are used to do various forms of
2349bit-twiddling in a program. They are generally very efficient
2350instructions and can commonly be strength reduced from other
Chris Lattner1429e6f2008-04-01 18:45:27 +0000[diff] [blame]2351instructions. They require two operands of the same type, execute an operation on them,
2352and produce a single value. The resulting value is the same type as its operands.</p>
Reid Spencer2ab01932007-02-02 13:57:07 +0000[diff] [blame]2353</div>
2354
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2355<!-- _______________________________________________________________________ -->
2356<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2357Instruction</a> </div>
2358<div class="doc_text">
2359<h5>Syntax:</h5>
2360<pre> &lt;result&gt; = shl &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2361</pre>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2362
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2363<h5>Overview:</h5>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2364
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2365<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2366the left a specified number of bits.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2367
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2368<h5>Arguments:</h5>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2369
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2370<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Chris Lattnerfe8519c2008-04-19 21:01:16 +0000[diff] [blame]2371 href="#t_integer">integer</a> type. '<tt>var2</tt>' is treated as an
2372unsigned value.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2373
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2374<h5>Semantics:</h5>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2375
Chris Lattner1429e6f2008-04-01 18:45:27 +0000[diff] [blame]2376<p>The value produced is <tt>var1</tt> * 2<sup><tt>var2</tt></sup> mod 2<sup>n</sup>,
2377where n is the width of the result. If <tt>var2</tt> is (statically or dynamically) negative or
2378equal to or larger than the number of bits in <tt>var1</tt>, the result is undefined.</p>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2379
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2380<h5>Example:</h5><pre>
2381 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2382 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2383 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2384 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2385</pre>
2386</div>
2387<!-- _______________________________________________________________________ -->
2388<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2389Instruction</a> </div>
2390<div class="doc_text">
2391<h5>Syntax:</h5>
2392<pre> &lt;result&gt; = lshr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2393</pre>
2394
2395<h5>Overview:</h5>
2396<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2397operand shifted to the right a specified number of bits with zero fill.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2398
2399<h5>Arguments:</h5>
2400<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Chris Lattnerfe8519c2008-04-19 21:01:16 +0000[diff] [blame]2401<a href="#t_integer">integer</a> type. '<tt>var2</tt>' is treated as an
2402unsigned value.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2403
2404<h5>Semantics:</h5>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2405
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2406<p>This instruction always performs a logical shift right operation. The most
2407significant bits of the result will be filled with zero bits after the
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2408shift. If <tt>var2</tt> is (statically or dynamically) equal to or larger than
2409the number of bits in <tt>var1</tt>, the result is undefined.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2410
2411<h5>Example:</h5>
2412<pre>
2413 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2414 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2415 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2416 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2417 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2418</pre>
2419</div>
2420
Reid Spencer2ab01932007-02-02 13:57:07 +0000[diff] [blame]2421<!-- _______________________________________________________________________ -->
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2422<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2423Instruction</a> </div>
2424<div class="doc_text">
2425
2426<h5>Syntax:</h5>
2427<pre> &lt;result&gt; = ashr &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
2428</pre>
2429
2430<h5>Overview:</h5>
2431<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2432operand shifted to the right a specified number of bits with sign extension.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2433
2434<h5>Arguments:</h5>
2435<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Chris Lattnerfe8519c2008-04-19 21:01:16 +0000[diff] [blame]2436<a href="#t_integer">integer</a> type. '<tt>var2</tt>' is treated as an
2437unsigned value.</p>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2438
2439<h5>Semantics:</h5>
2440<p>This instruction always performs an arithmetic shift right operation,
2441The most significant bits of the result will be filled with the sign bit
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2442of <tt>var1</tt>. If <tt>var2</tt> is (statically or dynamically) equal to or
2443larger than the number of bits in <tt>var1</tt>, the result is undefined.
2444</p>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2445
2446<h5>Example:</h5>
2447<pre>
2448 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
2449 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
2450 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
2451 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerf0e50112007-10-03 21:01:14 +0000[diff] [blame]2452 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Reid Spencer04e259b2007-01-31 21:39:12 +0000[diff] [blame]2453</pre>
2454</div>
2455
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2456<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2457<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
2458Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2459<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2460<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2461<pre> &lt;result&gt; = and &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2462</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2463<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2464<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
2465its two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2466<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2467<p>The two arguments to the '<tt>and</tt>' instruction must be <a
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]2468 href="#t_integer">integer</a> values. Both arguments must have
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2469identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2470<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2471<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2472<p> </p>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]2473<div style="align: center">
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2474<table border="1" cellspacing="0" cellpadding="4">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2475 <tbody>
2476 <tr>
2477 <td>In0</td>
2478 <td>In1</td>
2479 <td>Out</td>
2480 </tr>
2481 <tr>
2482 <td>0</td>
2483 <td>0</td>
2484 <td>0</td>
2485 </tr>
2486 <tr>
2487 <td>0</td>
2488 <td>1</td>
2489 <td>0</td>
2490 </tr>
2491 <tr>
2492 <td>1</td>
2493 <td>0</td>
2494 <td>0</td>
2495 </tr>
2496 <tr>
2497 <td>1</td>
2498 <td>1</td>
2499 <td>1</td>
2500 </tr>
2501 </tbody>
2502</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]2503</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2504<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2505<pre> &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
2506 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
2507 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2508</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2509</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2510<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2511<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2512<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2513<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2514<pre> &lt;result&gt; = or &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2515</pre>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2516<h5>Overview:</h5>
2517<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
2518or of its two operands.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2519<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2520<p>The two arguments to the '<tt>or</tt>' instruction must be <a
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]2521 href="#t_integer">integer</a> values. Both arguments must have
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2522identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2523<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2524<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2525<p> </p>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]2526<div style="align: center">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2527<table border="1" cellspacing="0" cellpadding="4">
2528 <tbody>
2529 <tr>
2530 <td>In0</td>
2531 <td>In1</td>
2532 <td>Out</td>
2533 </tr>
2534 <tr>
2535 <td>0</td>
2536 <td>0</td>
2537 <td>0</td>
2538 </tr>
2539 <tr>
2540 <td>0</td>
2541 <td>1</td>
2542 <td>1</td>
2543 </tr>
2544 <tr>
2545 <td>1</td>
2546 <td>0</td>
2547 <td>1</td>
2548 </tr>
2549 <tr>
2550 <td>1</td>
2551 <td>1</td>
2552 <td>1</td>
2553 </tr>
2554 </tbody>
2555</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]2556</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2557<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2558<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
2559 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
2560 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2561</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2562</div>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2563<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2564<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
2565Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2566<div class="doc_text">
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2567<h5>Syntax:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2568<pre> &lt;result&gt; = xor &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2569</pre>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2570<h5>Overview:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2571<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
2572or of its two operands. The <tt>xor</tt> is used to implement the
2573"one's complement" operation, which is the "~" operator in C.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2574<h5>Arguments:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2575<p>The two arguments to the '<tt>xor</tt>' instruction must be <a
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]2576 href="#t_integer">integer</a> values. Both arguments must have
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2577identical types.</p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2578<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2579<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2580<p> </p>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]2581<div style="align: center">
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2582<table border="1" cellspacing="0" cellpadding="4">
2583 <tbody>
2584 <tr>
2585 <td>In0</td>
2586 <td>In1</td>
2587 <td>Out</td>
2588 </tr>
2589 <tr>
2590 <td>0</td>
2591 <td>0</td>
2592 <td>0</td>
2593 </tr>
2594 <tr>
2595 <td>0</td>
2596 <td>1</td>
2597 <td>1</td>
2598 </tr>
2599 <tr>
2600 <td>1</td>
2601 <td>0</td>
2602 <td>1</td>
2603 </tr>
2604 <tr>
2605 <td>1</td>
2606 <td>1</td>
2607 <td>0</td>
2608 </tr>
2609 </tbody>
2610</table>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]2611</div>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2612<p> </p>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2613<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2614<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
2615 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
2616 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
2617 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2618</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2619</div>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2620
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2621<!-- ======================================================================= -->
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2622<div class="doc_subsection">
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2623 <a name="vectorops">Vector Operations</a>
2624</div>
2625
2626<div class="doc_text">
2627
2628<p>LLVM supports several instructions to represent vector operations in a
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2629target-independent manner. These instructions cover the element-access and
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2630vector-specific operations needed to process vectors effectively. While LLVM
2631does directly support these vector operations, many sophisticated algorithms
2632will want to use target-specific intrinsics to take full advantage of a specific
2633target.</p>
2634
2635</div>
2636
2637<!-- _______________________________________________________________________ -->
2638<div class="doc_subsubsection">
2639 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
2640</div>
2641
2642<div class="doc_text">
2643
2644<h5>Syntax:</h5>
2645
2646<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2647 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2648</pre>
2649
2650<h5>Overview:</h5>
2651
2652<p>
2653The '<tt>extractelement</tt>' instruction extracts a single scalar
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2654element from a vector at a specified index.
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2655</p>
2656
2657
2658<h5>Arguments:</h5>
2659
2660<p>
2661The first operand of an '<tt>extractelement</tt>' instruction is a
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2662value of <a href="#t_vector">vector</a> type. The second operand is
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2663an index indicating the position from which to extract the element.
2664The index may be a variable.</p>
2665
2666<h5>Semantics:</h5>
2667
2668<p>
2669The result is a scalar of the same type as the element type of
2670<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
2671<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
2672results are undefined.
2673</p>
2674
2675<h5>Example:</h5>
2676
2677<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2678 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2679</pre>
2680</div>
2681
2682
2683<!-- _______________________________________________________________________ -->
2684<div class="doc_subsubsection">
2685 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
2686</div>
2687
2688<div class="doc_text">
2689
2690<h5>Syntax:</h5>
2691
2692<pre>
Dan Gohman43ba0672008-05-12 23:38:42 +0000[diff] [blame]2693 &lt;result&gt; = insertelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, i32 &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2694</pre>
2695
2696<h5>Overview:</h5>
2697
2698<p>
2699The '<tt>insertelement</tt>' instruction inserts a scalar
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2700element into a vector at a specified index.
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2701</p>
2702
2703
2704<h5>Arguments:</h5>
2705
2706<p>
2707The first operand of an '<tt>insertelement</tt>' instruction is a
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2708value of <a href="#t_vector">vector</a> type. The second operand is a
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2709scalar value whose type must equal the element type of the first
2710operand. The third operand is an index indicating the position at
2711which to insert the value. The index may be a variable.</p>
2712
2713<h5>Semantics:</h5>
2714
2715<p>
Reid Spencer404a3252007-02-15 03:07:05 +0000[diff] [blame]2716The result is a vector of the same type as <tt>val</tt>. Its
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2717element values are those of <tt>val</tt> except at position
2718<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
2719exceeds the length of <tt>val</tt>, the results are undefined.
2720</p>
2721
2722<h5>Example:</h5>
2723
2724<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2725 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2726</pre>
2727</div>
2728
2729<!-- _______________________________________________________________________ -->
2730<div class="doc_subsubsection">
2731 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
2732</div>
2733
2734<div class="doc_text">
2735
2736<h5>Syntax:</h5>
2737
2738<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2739 &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;n x i32&gt; &lt;mask&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2740</pre>
2741
2742<h5>Overview:</h5>
2743
2744<p>
2745The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
2746from two input vectors, returning a vector of the same type.
2747</p>
2748
2749<h5>Arguments:</h5>
2750
2751<p>
2752The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
2753with types that match each other and types that match the result of the
2754instruction. The third argument is a shuffle mask, which has the same number
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2755of elements as the other vector type, but whose element type is always 'i32'.
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2756</p>
2757
2758<p>
2759The shuffle mask operand is required to be a constant vector with either
2760constant integer or undef values.
2761</p>
2762
2763<h5>Semantics:</h5>
2764
2765<p>
2766The elements of the two input vectors are numbered from left to right across
2767both of the vectors. The shuffle mask operand specifies, for each element of
2768the result vector, which element of the two input registers the result element
2769gets. The element selector may be undef (meaning "don't care") and the second
2770operand may be undef if performing a shuffle from only one vector.
2771</p>
2772
2773<h5>Example:</h5>
2774
2775<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2776 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]2777 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2778 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
2779 &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.
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2780</pre>
2781</div>
2782
Tanya Lattnerb138bbe2006-04-14 19:24:33 +0000[diff] [blame]2783
Chris Lattnerce83bff2006-04-08 23:07:04 +0000[diff] [blame]2784<!-- ======================================================================= -->
2785<div class="doc_subsection">
Dan Gohmanb9d66602008-05-12 23:51:09 +0000[diff] [blame^]2786 <a name="aggregateops">Aggregate Operations</a>
2787</div>
2788
2789<div class="doc_text">
2790
2791<p>LLVM supports several instructions for working with aggregate values.
2792</p>
2793
2794</div>
2795
2796<!-- _______________________________________________________________________ -->
2797<div class="doc_subsubsection">
2798 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
2799</div>
2800
2801<div class="doc_text">
2802
2803<h5>Syntax:</h5>
2804
2805<pre>
2806 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
2807</pre>
2808
2809<h5>Overview:</h5>
2810
2811<p>
2812The '<tt>extractvalue</tt>' instruction extracts a value
2813from an aggregate value.
2814</p>
2815
2816
2817<h5>Arguments:</h5>
2818
2819<p>
2820The first operand of an '<tt>extractvalue</tt>' instruction is a
2821value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
2822type. The operands are constant indicies to specify which value to extract
2823in the same manner as indicies in a
2824'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2825</p>
2826
2827<h5>Semantics:</h5>
2828
2829<p>
2830The result is the value at the position in the aggregate specified by
2831the index operands.
2832</p>
2833
2834<h5>Example:</h5>
2835
2836<pre>
2837 %result = extractvalue {i32, float} %agg, i32 0 <i>; yields i32</i>
2838</pre>
2839</div>
2840
2841
2842<!-- _______________________________________________________________________ -->
2843<div class="doc_subsubsection">
2844 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
2845</div>
2846
2847<div class="doc_text">
2848
2849<h5>Syntax:</h5>
2850
2851<pre>
2852 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
2853</pre>
2854
2855<h5>Overview:</h5>
2856
2857<p>
2858The '<tt>insertvalue</tt>' instruction inserts a value
2859into a aggregate.
2860</p>
2861
2862
2863<h5>Arguments:</h5>
2864
2865<p>
2866The first operand of an '<tt>insertvalue</tt>' instruction is a
2867value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
2868The second operand is a first-class value to insert.
2869type of the first operand. The following operands are constant indicies
2870indicating the position at which to insert the value in the same manner as
2871indicies in a
2872'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
2873The value to insert must have the same type as the value identified
2874by the indicies.
2875
2876<h5>Semantics:</h5>
2877
2878<p>
2879The result is an aggregate of the same type as <tt>val</tt>. Its
2880value is that of <tt>val</tt> except that the value at the position
2881specified by the indicies is that of <tt>elt</tt>.
2882</p>
2883
2884<h5>Example:</h5>
2885
2886<pre>
2887 %result = insertvalue {i32, float} %agg, i32 1, i32 0 <i>; yields {i32, float}</i>
2888</pre>
2889</div>
2890
2891
2892<!-- ======================================================================= -->
2893<div class="doc_subsection">
Chris Lattner6ab66722006-08-15 00:45:58 +0000[diff] [blame]2894 <a name="memoryops">Memory Access and Addressing Operations</a>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2895</div>
2896
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2897<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2898
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2899<p>A key design point of an SSA-based representation is how it
2900represents memory. In LLVM, no memory locations are in SSA form, which
2901makes things very simple. This section describes how to read, write,
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]2902allocate, and free memory in LLVM.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2903
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2904</div>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2905
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2906<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2907<div class="doc_subsubsection">
2908 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
2909</div>
2910
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2911<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2912
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2913<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2914
2915<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2916 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2917</pre>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2918
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2919<h5>Overview:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2920
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2921<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lamb55c6d4f2007-12-17 01:00:21 +0000[diff] [blame]2922heap and returns a pointer to it. The object is always allocated in the generic
2923address space (address space zero).</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2924
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2925<h5>Arguments:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2926
2927<p>The '<tt>malloc</tt>' instruction allocates
2928<tt>sizeof(&lt;type&gt;)*NumElements</tt>
John Criswella92e5862004-02-24 16:13:56 +0000[diff] [blame]2929bytes of memory from the operating system and returns a pointer of the
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2930appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greifdd1fc982008-02-09 22:24:34 +0000[diff] [blame]2931number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]2932If a constant alignment is specified, the value result of the allocation is guaranteed to
Gabor Greifdd1fc982008-02-09 22:24:34 +0000[diff] [blame]2933be aligned to at least that boundary. If not specified, or if zero, the target can
2934choose to align the allocation on any convenient boundary.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2935
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2936<p>'<tt>type</tt>' must be a sized type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2937
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2938<h5>Semantics:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2939
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2940<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Chris Lattnerfe8519c2008-04-19 21:01:16 +0000[diff] [blame]2941a pointer is returned. The result of a zero byte allocattion is undefined. The
2942result is null if there is insufficient memory available.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2943
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2944<h5>Example:</h5>
2945
2946<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2947 %array = malloc [4 x i8 ] <i>; yields {[%4 x i8]*}:array</i>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2948
Bill Wendling2d8b9a82007-05-29 09:42:13 +0000[diff] [blame]2949 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
2950 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
2951 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
2952 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
2953 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2954</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2955</div>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2956
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2957<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2958<div class="doc_subsubsection">
2959 <a name="i_free">'<tt>free</tt>' Instruction</a>
2960</div>
2961
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2962<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2963
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2964<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2965
2966<pre>
2967 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2968</pre>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2969
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2970<h5>Overview:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2971
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2972<p>The '<tt>free</tt>' instruction returns memory back to the unused
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]2973memory heap to be reallocated in the future.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2974
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2975<h5>Arguments:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2976
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]2977<p>'<tt>value</tt>' shall be a pointer value that points to a value
2978that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
2979instruction.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2980
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2981<h5>Semantics:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2982
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]2983<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner0f103e12008-04-19 22:41:32 +0000[diff] [blame]2984after this instruction executes. If the pointer is null, the operation
2985is a noop.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2986
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2987<h5>Example:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2988
2989<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]2990 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
2991 free [4 x i8]* %array
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2992</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]2993</div>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2994
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]2995<!-- _______________________________________________________________________ -->
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]2996<div class="doc_subsubsection">
2997 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
2998</div>
2999
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3000<div class="doc_text">
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3001
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3002<h5>Syntax:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3003
3004<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3005 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3006</pre>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3007
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3008<h5>Overview:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3009
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]3010<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3011currently executing function, to be automatically released when this function
Christopher Lamb55c6d4f2007-12-17 01:00:21 +0000[diff] [blame]3012returns to its caller. The object is always allocated in the generic address
3013space (address space zero).</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3014
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3015<h5>Arguments:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3016
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]3017<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3018bytes of memory on the runtime stack, returning a pointer of the
Gabor Greifdd1fc982008-02-09 22:24:34 +0000[diff] [blame]3019appropriate type to the program. If "NumElements" is specified, it is the
3020number of elements allocated, otherwise "NumElements" is defaulted to be one.
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3021If a constant alignment is specified, the value result of the allocation is guaranteed
Gabor Greifdd1fc982008-02-09 22:24:34 +0000[diff] [blame]3022to be aligned to at least that boundary. If not specified, or if zero, the target
3023can choose to align the allocation on any convenient boundary.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3024
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3025<p>'<tt>type</tt>' may be any sized type.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3026
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3027<h5>Semantics:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3028
Chris Lattnerfe8519c2008-04-19 21:01:16 +0000[diff] [blame]3029<p>Memory is allocated; a pointer is returned. The operation is undefiend if
3030there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3031memory is automatically released when the function returns. The '<tt>alloca</tt>'
3032instruction is commonly used to represent automatic variables that must
3033have an address available. When the function returns (either with the <tt><a
John Criswellc932bef2005-05-12 16:55:34 +0000[diff] [blame]3034 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3035instructions), the memory is reclaimed. Allocating zero bytes
3036is legal, but the result is undefined.</p>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3037
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3038<h5>Example:</h5>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3039
3040<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3041 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3042 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3043 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3044 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3045</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3046</div>
Chris Lattner54611b42005-11-06 08:02:57 +0000[diff] [blame]3047
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3048<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3049<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3050Instruction</a> </div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3051<div class="doc_text">
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3052<h5>Syntax:</h5>
Christopher Lambbff50202007-04-21 08:16:25 +0000[diff] [blame]3053<pre> &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]<br> &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]<br></pre>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3054<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3055<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3056<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3057<p>The argument to the '<tt>load</tt>' instruction specifies the memory
John Criswell4c0cf7f2005-10-24 16:17:18 +0000[diff] [blame]3058address from which to load. The pointer must point to a <a
Chris Lattner10ee9652004-06-03 22:57:15 +0000[diff] [blame]3059 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
John Criswell4c0cf7f2005-10-24 16:17:18 +0000[diff] [blame]3060marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3061the number or order of execution of this <tt>load</tt> with other
3062volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3063instructions. </p>
Chris Lattner2a1993f2008-01-06 21:04:43 +0000[diff] [blame]3064<p>
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3065The optional constant "align" argument specifies the alignment of the operation
Chris Lattner2a1993f2008-01-06 21:04:43 +0000[diff] [blame]3066(that is, the alignment of the memory address). A value of 0 or an
3067omitted "align" argument means that the operation has the preferential
3068alignment for the target. It is the responsibility of the code emitter
3069to ensure that the alignment information is correct. Overestimating
3070the alignment results in an undefined behavior. Underestimating the
3071alignment may produce less efficient code. An alignment of 1 is always
3072safe.
3073</p>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3074<h5>Semantics:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3075<p>The location of memory pointed to is loaded.</p>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3076<h5>Examples:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3077<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3078 <a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3079 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3080 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3081</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3082</div>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3083<!-- _______________________________________________________________________ -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3084<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3085Instruction</a> </div>
Reid Spencera89fb182006-11-09 21:18:01 +0000[diff] [blame]3086<div class="doc_text">
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3087<h5>Syntax:</h5>
Christopher Lambbff50202007-04-21 08:16:25 +0000[diff] [blame]3088<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3089 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3090</pre>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3091<h5>Overview:</h5>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3092<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3093<h5>Arguments:</h5>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3094<p>There are two arguments to the '<tt>store</tt>' instruction: a value
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]3095to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3096operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3097of the '<tt>&lt;value&gt;</tt>'
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]3098operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3099optimizer is not allowed to modify the number or order of execution of
3100this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3101 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner2a1993f2008-01-06 21:04:43 +0000[diff] [blame]3102<p>
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3103The optional constant "align" argument specifies the alignment of the operation
Chris Lattner2a1993f2008-01-06 21:04:43 +0000[diff] [blame]3104(that is, the alignment of the memory address). A value of 0 or an
3105omitted "align" argument means that the operation has the preferential
3106alignment for the target. It is the responsibility of the code emitter
3107to ensure that the alignment information is correct. Overestimating
3108the alignment results in an undefined behavior. Underestimating the
3109alignment may produce less efficient code. An alignment of 1 is always
3110safe.
3111</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3112<h5>Semantics:</h5>
3113<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
3114at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.</p>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3115<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3116<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling8830ffe2007-10-22 05:10:05 +0000[diff] [blame]3117 store i32 3, i32* %ptr <i>; yields {void}</i>
3118 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3119</pre>
Reid Spencer443460a2006-11-09 21:15:49 +0000[diff] [blame]3120</div>
3121
Chris Lattner095735d2002-05-06 03:03:22 +0000[diff] [blame]3122<!-- _______________________________________________________________________ -->
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3123<div class="doc_subsubsection">
3124 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3125</div>
3126
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3127<div class="doc_text">
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]3128<h5>Syntax:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3129<pre>
3130 &lt;result&gt; = getelementptr &lt;ty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
3131</pre>
3132
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]3133<h5>Overview:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3134
3135<p>
3136The '<tt>getelementptr</tt>' instruction is used to get the address of a
3137subelement of an aggregate data structure.</p>
3138
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]3139<h5>Arguments:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3140
Reid Spencercee005c2006-12-04 21:29:24 +0000[diff] [blame]3141<p>This instruction takes a list of integer operands that indicate what
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3142elements of the aggregate object to index to. The actual types of the arguments
3143provided depend on the type of the first pointer argument. The
3144'<tt>getelementptr</tt>' instruction is used to index down through the type
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]3145levels of a structure or to a specific index in an array. When indexing into a
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3146structure, only <tt>i32</tt> integer constants are allowed. When indexing
Chris Lattner851b7712008-04-24 05:59:56 +0000[diff] [blame]3147into an array or pointer, only integers of 32 or 64 bits are allowed; 32-bit
3148values will be sign extended to 64-bits if required.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3149
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3150<p>For example, let's consider a C code fragment and how it gets
3151compiled to LLVM:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3152
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3153<div class="doc_code">
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3154<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3155struct RT {
3156 char A;
Chris Lattnera446f1b2007-05-29 15:43:56 +0000[diff] [blame]3157 int B[10][20];
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3158 char C;
3159};
3160struct ST {
Chris Lattnera446f1b2007-05-29 15:43:56 +0000[diff] [blame]3161 int X;
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3162 double Y;
3163 struct RT Z;
3164};
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3165
Chris Lattnera446f1b2007-05-29 15:43:56 +0000[diff] [blame]3166int *foo(struct ST *s) {
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3167 return &amp;s[1].Z.B[5][13];
3168}
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3169</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3170</div>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3171
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3172<p>The LLVM code generated by the GCC frontend is:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3173
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3174<div class="doc_code">
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3175<pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3176%RT = type { i8 , [10 x [20 x i32]], i8 }
3177%ST = type { i32, double, %RT }
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3178
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3179define i32* %foo(%ST* %s) {
3180entry:
3181 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3182 ret i32* %reg
3183}
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3184</pre>
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]3185</div>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3186
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]3187<h5>Semantics:</h5>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3188
3189<p>The index types specified for the '<tt>getelementptr</tt>' instruction depend
John Criswell4a3327e2005-05-13 22:25:59 +0000[diff] [blame]3190on the pointer type that is being indexed into. <a href="#t_pointer">Pointer</a>
Reid Spencercee005c2006-12-04 21:29:24 +0000[diff] [blame]3191and <a href="#t_array">array</a> types can use a 32-bit or 64-bit
Reid Spencerc0312692006-12-03 16:53:48 +0000[diff] [blame]3192<a href="#t_integer">integer</a> type but the value will always be sign extended
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3193to 64-bits. <a href="#t_struct">Structure</a> and <a href="#t_pstruct">packed
3194structure</a> types require <tt>i32</tt> <b>constants</b>.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3195
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3196<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3197type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3198}</tt>' type, a structure. The second index indexes into the third element of
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3199the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3200i8 }</tt>' type, another structure. The third index indexes into the second
3201element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3202array. The two dimensions of the array are subscripted into, yielding an
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3203'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3204to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3205
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]3206<p>Note that it is perfectly legal to index partially through a
3207structure, returning a pointer to an inner element. Because of this,
3208the LLVM code for the given testcase is equivalent to:</p>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3209
3210<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3211 define i32* %foo(%ST* %s) {
3212 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
Jeff Cohen5819f182007-04-22 01:17:39 +0000[diff] [blame]3213 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3214 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3215 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3216 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3217 ret i32* %t5
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3218 }
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]3219</pre>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000[diff] [blame]3220
3221<p>Note that it is undefined to access an array out of bounds: array and
3222pointer indexes must always be within the defined bounds of the array type.
Chris Lattner851b7712008-04-24 05:59:56 +0000[diff] [blame]3223The one exception for this rule is zero length arrays. These arrays are
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000[diff] [blame]3224defined to be accessible as variable length arrays, which requires access
3225beyond the zero'th element.</p>
3226
Chris Lattner6ab66722006-08-15 00:45:58 +0000[diff] [blame]3227<p>The getelementptr instruction is often confusing. For some more insight
3228into how it works, see <a href="GetElementPtr.html">the getelementptr
3229FAQ</a>.</p>
3230
Chris Lattner590645f2002-04-14 06:13:44 +0000[diff] [blame]3231<h5>Example:</h5>
Chris Lattnerc0ad71e2005-06-24 17:22:57 +0000[diff] [blame]3232
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3233<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3234 <i>; yields [12 x i8]*:aptr</i>
3235 %aptr = getelementptr {i32, [12 x i8]}* %sptr, i64 0, i32 1
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3236</pre>
Chris Lattner33fd7022004-04-05 01:30:49 +0000[diff] [blame]3237</div>
Reid Spencer443460a2006-11-09 21:15:49 +0000[diff] [blame]3238
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]3239<!-- ======================================================================= -->
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3240<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3241</div>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3242<div class="doc_text">
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3243<p>The instructions in this category are the conversion instructions (casting)
3244which all take a single operand and a type. They perform various bit conversions
3245on the operand.</p>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3246</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3247
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]3248<!-- _______________________________________________________________________ -->
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3249<div class="doc_subsubsection">
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3250 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3251</div>
3252<div class="doc_text">
3253
3254<h5>Syntax:</h5>
3255<pre>
3256 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3257</pre>
3258
3259<h5>Overview:</h5>
3260<p>
3261The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3262</p>
3263
3264<h5>Arguments:</h5>
3265<p>
3266The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3267be an <a href="#t_integer">integer</a> type, and a type that specifies the size
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]3268and type of the result, which must be an <a href="#t_integer">integer</a>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3269type. The bit size of <tt>value</tt> must be larger than the bit size of
3270<tt>ty2</tt>. Equal sized types are not allowed.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3271
3272<h5>Semantics:</h5>
3273<p>
3274The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3275and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3276larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3277It will always truncate bits.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3278
3279<h5>Example:</h5>
3280<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3281 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3282 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3283 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3284</pre>
3285</div>
3286
3287<!-- _______________________________________________________________________ -->
3288<div class="doc_subsubsection">
3289 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3290</div>
3291<div class="doc_text">
3292
3293<h5>Syntax:</h5>
3294<pre>
3295 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3296</pre>
3297
3298<h5>Overview:</h5>
3299<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3300<tt>ty2</tt>.</p>
3301
3302
3303<h5>Arguments:</h5>
3304<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]3305<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3306also be of <a href="#t_integer">integer</a> type. The bit size of the
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3307<tt>value</tt> must be smaller than the bit size of the destination type,
3308<tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3309
3310<h5>Semantics:</h5>
3311<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Chris Lattnerc87f3df2007-05-24 19:13:27 +0000[diff] [blame]3312bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3313
Reid Spencer07c9c682007-01-12 15:46:11 +0000[diff] [blame]3314<p>When zero extending from i1, the result will always be either 0 or 1.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3315
3316<h5>Example:</h5>
3317<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3318 %X = zext i32 257 to i64 <i>; yields i64:257</i>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3319 %Y = zext i1 true to i32 <i>; yields i32:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3320</pre>
3321</div>
3322
3323<!-- _______________________________________________________________________ -->
3324<div class="doc_subsubsection">
3325 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3326</div>
3327<div class="doc_text">
3328
3329<h5>Syntax:</h5>
3330<pre>
3331 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3332</pre>
3333
3334<h5>Overview:</h5>
3335<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3336
3337<h5>Arguments:</h5>
3338<p>
3339The '<tt>sext</tt>' instruction takes a value to cast, which must be of
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]3340<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3341also be of <a href="#t_integer">integer</a> type. The bit size of the
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3342<tt>value</tt> must be smaller than the bit size of the destination type,
3343<tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3344
3345<h5>Semantics:</h5>
3346<p>
3347The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3348bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
Chris Lattnerc87f3df2007-05-24 19:13:27 +0000[diff] [blame]3349the type <tt>ty2</tt>.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3350
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3351<p>When sign extending from i1, the extension always results in -1 or 0.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3352
3353<h5>Example:</h5>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3354<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3355 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3356 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3357</pre>
3358</div>
3359
3360<!-- _______________________________________________________________________ -->
3361<div class="doc_subsubsection">
Reid Spencer2e2740d2006-11-09 21:48:10 +0000[diff] [blame]3362 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3363</div>
3364
3365<div class="doc_text">
3366
3367<h5>Syntax:</h5>
3368
3369<pre>
3370 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3371</pre>
3372
3373<h5>Overview:</h5>
3374<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3375<tt>ty2</tt>.</p>
3376
3377
3378<h5>Arguments:</h5>
3379<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
3380 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
3381cast it to. The size of <tt>value</tt> must be larger than the size of
3382<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
3383<i>no-op cast</i>.</p>
3384
3385<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3386<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
3387<a href="#t_floating">floating point</a> type to a smaller
3388<a href="#t_floating">floating point</a> type. If the value cannot fit within
3389the destination type, <tt>ty2</tt>, then the results are undefined.</p>
Reid Spencer2e2740d2006-11-09 21:48:10 +0000[diff] [blame]3390
3391<h5>Example:</h5>
3392<pre>
3393 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
3394 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
3395</pre>
3396</div>
3397
3398<!-- _______________________________________________________________________ -->
3399<div class="doc_subsubsection">
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3400 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
3401</div>
3402<div class="doc_text">
3403
3404<h5>Syntax:</h5>
3405<pre>
3406 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3407</pre>
3408
3409<h5>Overview:</h5>
3410<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
3411floating point value.</p>
3412
3413<h5>Arguments:</h5>
3414<p>The '<tt>fpext</tt>' instruction takes a
3415<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3416and a <a href="#t_floating">floating point</a> type to cast it to. The source
3417type must be smaller than the destination type.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3418
3419<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3420<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]3421<a href="#t_floating">floating point</a> type to a larger
3422<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3423used to make a <i>no-op cast</i> because it always changes bits. Use
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]3424<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3425
3426<h5>Example:</h5>
3427<pre>
3428 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
3429 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
3430</pre>
3431</div>
3432
3433<!-- _______________________________________________________________________ -->
3434<div class="doc_subsubsection">
Reid Spencer2eadb532007-01-21 00:29:26 +0000[diff] [blame]3435 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3436</div>
3437<div class="doc_text">
3438
3439<h5>Syntax:</h5>
3440<pre>
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]3441 &lt;result&gt; = fptoui &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3442</pre>
3443
3444<h5>Overview:</h5>
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]3445<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3446unsigned integer equivalent of type <tt>ty2</tt>.
3447</p>
3448
3449<h5>Arguments:</h5>
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]3450<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]3451scalar or vector <a href="#t_floating">floating point</a> value, and a type
3452to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3453type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3454vector integer type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3455
3456<h5>Semantics:</h5>
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]3457<p> The '<tt>fptoui</tt>' instruction converts its
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3458<a href="#t_floating">floating point</a> operand into the nearest (rounding
3459towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
3460the results are undefined.</p>
3461
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3462<h5>Example:</h5>
3463<pre>
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]3464 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner5b95a172007-09-22 03:17:52 +0000[diff] [blame]3465 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencer753163d2007-07-31 14:40:14 +0000[diff] [blame]3466 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3467</pre>
3468</div>
3469
3470<!-- _______________________________________________________________________ -->
3471<div class="doc_subsubsection">
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3472 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3473</div>
3474<div class="doc_text">
3475
3476<h5>Syntax:</h5>
3477<pre>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3478 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3479</pre>
3480
3481<h5>Overview:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3482<p>The '<tt>fptosi</tt>' instruction converts
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3483<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3484</p>
3485
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]3486<h5>Arguments:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3487<p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]3488scalar or vector <a href="#t_floating">floating point</a> value, and a type
3489to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
3490type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
3491vector integer type with the same number of elements as <tt>ty</tt></p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3492
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]3493<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3494<p>The '<tt>fptosi</tt>' instruction converts its
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3495<a href="#t_floating">floating point</a> operand into the nearest (rounding
3496towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
3497the results are undefined.</p>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3498
Chris Lattner70de6632001-07-09 00:26:23 +0000[diff] [blame]3499<h5>Example:</h5>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3500<pre>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3501 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner5b95a172007-09-22 03:17:52 +0000[diff] [blame]3502 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3503 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3504</pre>
3505</div>
3506
3507<!-- _______________________________________________________________________ -->
3508<div class="doc_subsubsection">
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3509 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3510</div>
3511<div class="doc_text">
3512
3513<h5>Syntax:</h5>
3514<pre>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3515 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3516</pre>
3517
3518<h5>Overview:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3519<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3520integer and converts that value to the <tt>ty2</tt> type.</p>
3521
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3522<h5>Arguments:</h5>
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]3523<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
3524scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3525to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3526type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3527floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3528
3529<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3530<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3531integer quantity and converts it to the corresponding floating point value. If
Jeff Cohenbeccb742007-04-22 14:56:37 +0000[diff] [blame]3532the value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3533
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3534<h5>Example:</h5>
3535<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3536 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3537 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3538</pre>
3539</div>
3540
3541<!-- _______________________________________________________________________ -->
3542<div class="doc_subsubsection">
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3543 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3544</div>
3545<div class="doc_text">
3546
3547<h5>Syntax:</h5>
3548<pre>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3549 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3550</pre>
3551
3552<h5>Overview:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3553<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3554integer and converts that value to the <tt>ty2</tt> type.</p>
3555
3556<h5>Arguments:</h5>
Nate Begemand4d45c22007-11-17 03:58:34 +0000[diff] [blame]3557<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
3558scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
3559to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
3560type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
3561floating point type with the same number of elements as <tt>ty</tt></p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3562
3563<h5>Semantics:</h5>
Reid Spencer51b07252006-11-09 23:03:26 +0000[diff] [blame]3564<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3565integer quantity and converts it to the corresponding floating point value. If
Jeff Cohenbeccb742007-04-22 14:56:37 +0000[diff] [blame]3566the value cannot fit in the floating point value, the results are undefined.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3567
3568<h5>Example:</h5>
3569<pre>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3570 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3571 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3572</pre>
3573</div>
3574
3575<!-- _______________________________________________________________________ -->
3576<div class="doc_subsubsection">
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3577 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
3578</div>
3579<div class="doc_text">
3580
3581<h5>Syntax:</h5>
3582<pre>
3583 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3584</pre>
3585
3586<h5>Overview:</h5>
3587<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
3588the integer type <tt>ty2</tt>.</p>
3589
3590<h5>Arguments:</h5>
3591<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]3592must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3593<tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.
3594
3595<h5>Semantics:</h5>
3596<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
3597<tt>ty2</tt> by interpreting the pointer value as an integer and either
3598truncating or zero extending that value to the size of the integer type. If
3599<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
3600<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3601are the same size, then nothing is done (<i>no-op cast</i>) other than a type
3602change.</p>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3603
3604<h5>Example:</h5>
3605<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3606 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
3607 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3608</pre>
3609</div>
3610
3611<!-- _______________________________________________________________________ -->
3612<div class="doc_subsubsection">
3613 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
3614</div>
3615<div class="doc_text">
3616
3617<h5>Syntax:</h5>
3618<pre>
3619 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3620</pre>
3621
3622<h5>Overview:</h5>
3623<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
3624a pointer type, <tt>ty2</tt>.</p>
3625
3626<h5>Arguments:</h5>
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]3627<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3628value to cast, and a type to cast it to, which must be a
Anton Korobeynikova0554d92007-01-12 19:20:47 +0000[diff] [blame]3629<a href="#t_pointer">pointer</a> type.
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3630
3631<h5>Semantics:</h5>
3632<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
3633<tt>ty2</tt> by applying either a zero extension or a truncation depending on
3634the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
3635size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
3636the size of a pointer then a zero extension is done. If they are the same size,
3637nothing is done (<i>no-op cast</i>).</p>
3638
3639<h5>Example:</h5>
3640<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3641 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
3642 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
3643 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3644</pre>
3645</div>
3646
3647<!-- _______________________________________________________________________ -->
3648<div class="doc_subsubsection">
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]3649 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3650</div>
3651<div class="doc_text">
3652
3653<h5>Syntax:</h5>
3654<pre>
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]3655 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3656</pre>
3657
3658<h5>Overview:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]3659<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3660<tt>ty2</tt> without changing any bits.</p>
3661
3662<h5>Arguments:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]3663<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3664a first class value, and a type to cast it to, which must also be a <a
3665 href="#t_firstclass">first class</a> type. The bit sizes of <tt>value</tt>
Reid Spencere3db84c2007-01-09 20:08:58 +0000[diff] [blame]3666and the destination type, <tt>ty2</tt>, must be identical. If the source
3667type is a pointer, the destination type must also be a pointer.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3668
3669<h5>Semantics:</h5>
Reid Spencer5b950642006-11-11 23:08:07 +0000[diff] [blame]3670<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Reid Spencerb7344ff2006-11-11 21:00:47 +0000[diff] [blame]3671<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
3672this conversion. The conversion is done as if the <tt>value</tt> had been
3673stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
3674converted to other pointer types with this instruction. To convert pointers to
3675other types, use the <a href="#i_inttoptr">inttoptr</a> or
3676<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Reid Spencer59b6b7d2006-11-08 01:11:31 +0000[diff] [blame]3677
3678<h5>Example:</h5>
3679<pre>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3680 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3681 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
3682 %Z = bitcast <2xint> %V to i64; <i>; yields i64: %V</i>
Chris Lattner70de6632001-07-09 00:26:23 +0000[diff] [blame]3683</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]3684</div>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3685
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3686<!-- ======================================================================= -->
3687<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
3688<div class="doc_text">
3689<p>The instructions in this category are the "miscellaneous"
3690instructions, which defy better classification.</p>
3691</div>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3692
3693<!-- _______________________________________________________________________ -->
3694<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
3695</div>
3696<div class="doc_text">
3697<h5>Syntax:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3698<pre> &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3699</pre>
3700<h5>Overview:</h5>
3701<p>The '<tt>icmp</tt>' instruction returns a boolean value based on comparison
Chris Lattner1f17cce2008-04-02 00:38:26 +0000[diff] [blame]3702of its two integer or pointer operands.</p>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3703<h5>Arguments:</h5>
3704<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3705the condition code indicating the kind of comparison to perform. It is not
3706a value, just a keyword. The possible condition code are:
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3707<ol>
3708 <li><tt>eq</tt>: equal</li>
3709 <li><tt>ne</tt>: not equal </li>
3710 <li><tt>ugt</tt>: unsigned greater than</li>
3711 <li><tt>uge</tt>: unsigned greater or equal</li>
3712 <li><tt>ult</tt>: unsigned less than</li>
3713 <li><tt>ule</tt>: unsigned less or equal</li>
3714 <li><tt>sgt</tt>: signed greater than</li>
3715 <li><tt>sge</tt>: signed greater or equal</li>
3716 <li><tt>slt</tt>: signed less than</li>
3717 <li><tt>sle</tt>: signed less or equal</li>
3718</ol>
Chris Lattnerc0f423a2007-01-15 01:54:13 +0000[diff] [blame]3719<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Reid Spencer784ef792007-01-04 05:19:58 +0000[diff] [blame]3720<a href="#t_pointer">pointer</a> typed. They must also be identical types.</p>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3721<h5>Semantics:</h5>
3722<p>The '<tt>icmp</tt>' compares <tt>var1</tt> and <tt>var2</tt> according to
3723the condition code given as <tt>cond</tt>. The comparison performed always
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3724yields a <a href="#t_primitive">i1</a> result, as follows:
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3725<ol>
3726 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
3727 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3728 </li>
3729 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
3730 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
3731 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
3732 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3733 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
3734 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3735 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
3736 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3737 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
3738 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
3739 <li><tt>sgt</tt>: interprets the operands as signed values and yields
3740 <tt>true</tt> if <tt>var1</tt> is greater than <tt>var2</tt>.</li>
3741 <li><tt>sge</tt>: interprets the operands as signed values and yields
3742 <tt>true</tt> if <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
3743 <li><tt>slt</tt>: interprets the operands as signed values and yields
3744 <tt>true</tt> if <tt>var1</tt> is less than <tt>var2</tt>.</li>
3745 <li><tt>sle</tt>: interprets the operands as signed values and yields
3746 <tt>true</tt> if <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3747</ol>
3748<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3749values are compared as if they were integers.</p>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3750
3751<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3752<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
3753 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
3754 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
3755 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
3756 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
3757 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3758</pre>
3759</div>
3760
3761<!-- _______________________________________________________________________ -->
3762<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
3763</div>
3764<div class="doc_text">
3765<h5>Syntax:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3766<pre> &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {i1}:result</i>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3767</pre>
3768<h5>Overview:</h5>
3769<p>The '<tt>fcmp</tt>' instruction returns a boolean value based on comparison
3770of its floating point operands.</p>
3771<h5>Arguments:</h5>
3772<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3773the condition code indicating the kind of comparison to perform. It is not
3774a value, just a keyword. The possible condition code are:
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3775<ol>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3776 <li><tt>false</tt>: no comparison, always returns false</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3777 <li><tt>oeq</tt>: ordered and equal</li>
3778 <li><tt>ogt</tt>: ordered and greater than </li>
3779 <li><tt>oge</tt>: ordered and greater than or equal</li>
3780 <li><tt>olt</tt>: ordered and less than </li>
3781 <li><tt>ole</tt>: ordered and less than or equal</li>
3782 <li><tt>one</tt>: ordered and not equal</li>
3783 <li><tt>ord</tt>: ordered (no nans)</li>
3784 <li><tt>ueq</tt>: unordered or equal</li>
3785 <li><tt>ugt</tt>: unordered or greater than </li>
3786 <li><tt>uge</tt>: unordered or greater than or equal</li>
3787 <li><tt>ult</tt>: unordered or less than </li>
3788 <li><tt>ule</tt>: unordered or less than or equal</li>
3789 <li><tt>une</tt>: unordered or not equal</li>
3790 <li><tt>uno</tt>: unordered (either nans)</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3791 <li><tt>true</tt>: no comparison, always returns true</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3792</ol>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3793<p><i>Ordered</i> means that neither operand is a QNAN while
Reid Spencer02e0d1d2006-12-06 07:08:07 +0000[diff] [blame]3794<i>unordered</i> means that either operand may be a QNAN.</p>
Reid Spencer784ef792007-01-04 05:19:58 +0000[diff] [blame]3795<p>The <tt>val1</tt> and <tt>val2</tt> arguments must be
3796<a href="#t_floating">floating point</a> typed. They must have identical
3797types.</p>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3798<h5>Semantics:</h5>
Nate Begemand2195702008-05-12 19:01:56 +0000[diff] [blame]3799<p>The '<tt>fcmp</tt>' instruction compares <tt>var1</tt> and <tt>var2</tt>
3800according to the condition code given as <tt>cond</tt>. The comparison performed
3801always yields a <a href="#t_primitive">i1</a> result, as follows:
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3802<ol>
3803 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3804 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3805 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3806 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3807 <tt>var1</tt> is greather than <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3808 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3809 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3810 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3811 <tt>var1</tt> is less than <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3812 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3813 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3814 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3815 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3816 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
3817 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3818 <tt>var1</tt> is equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3819 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3820 <tt>var1</tt> is greater than <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3821 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3822 <tt>var1</tt> is greater than or equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3823 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3824 <tt>var1</tt> is less than <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3825 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3826 <tt>var1</tt> is less than or equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3827 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3828 <tt>var1</tt> is not equal to <tt>var2</tt>.</li>
Reid Spencerf69acf32006-11-19 03:00:14 +0000[diff] [blame]3829 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3830 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
3831</ol>
Reid Spencerc828a0e2006-11-18 21:50:54 +0000[diff] [blame]3832
3833<h5>Example:</h5>
3834<pre> &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
3835 &lt;result&gt; = icmp one float 4.0, 5.0 <i>; yields: result=true</i>
3836 &lt;result&gt; = icmp olt float 4.0, 5.0 <i>; yields: result=true</i>
3837 &lt;result&gt; = icmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
3838</pre>
3839</div>
3840
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3841<!-- _______________________________________________________________________ -->
Nate Begemand2195702008-05-12 19:01:56 +0000[diff] [blame]3842<div class="doc_subsubsection">
3843 <a name="i_vicmp">'<tt>vicmp</tt>' Instruction</a>
3844</div>
3845<div class="doc_text">
3846<h5>Syntax:</h5>
3847<pre> &lt;result&gt; = vicmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt; <i>; yields {ty}:result</i>
3848</pre>
3849<h5>Overview:</h5>
3850<p>The '<tt>vicmp</tt>' instruction returns an integer vector value based on
3851element-wise comparison of its two integer vector operands.</p>
3852<h5>Arguments:</h5>
3853<p>The '<tt>vicmp</tt>' instruction takes three operands. The first operand is
3854the condition code indicating the kind of comparison to perform. It is not
3855a value, just a keyword. The possible condition code are:
3856<ol>
3857 <li><tt>eq</tt>: equal</li>
3858 <li><tt>ne</tt>: not equal </li>
3859 <li><tt>ugt</tt>: unsigned greater than</li>
3860 <li><tt>uge</tt>: unsigned greater or equal</li>
3861 <li><tt>ult</tt>: unsigned less than</li>
3862 <li><tt>ule</tt>: unsigned less or equal</li>
3863 <li><tt>sgt</tt>: signed greater than</li>
3864 <li><tt>sge</tt>: signed greater or equal</li>
3865 <li><tt>slt</tt>: signed less than</li>
3866 <li><tt>sle</tt>: signed less or equal</li>
3867</ol>
3868<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
3869<a href="#t_integer">integer</a> typed. They must also be identical types.</p>
3870<h5>Semantics:</h5>
3871<p>The '<tt>vicmp</tt>' instruction compares <tt>var1</tt> and <tt>var2</tt>
3872according to the condition code given as <tt>cond</tt>. The comparison yields a
3873<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, of
3874identical type as the values being compared. The most significant bit in each
3875element is 1 if the element-wise comparison evaluates to true, and is 0
3876otherwise. All other bits of the result are undefined. The condition codes
3877are evaluated identically to the <a href="#i_icmp">'<tt>icmp</tt>'
3878instruction</a>.
3879
3880<h5>Example:</h5>
3881<pre>
3882 &lt;result&gt; = vicmp eq <2 x i32> < i32 4, i32 0 >, < i32 5, i32 0 > <i>; yields: result=<2 x i32> < i32 0, i32 -1 ></i>
3883 &lt;result&gt; = vicmp ult <2 x i8> < i8 1, i8 2 >, < i8 2, i8 2> <i>; yields: result=<2 x i8> < i8 -1, i8 0 ></i>
3884</pre>
3885</div>
3886
3887<!-- _______________________________________________________________________ -->
3888<div class="doc_subsubsection">
3889 <a name="i_vfcmp">'<tt>vfcmp</tt>' Instruction</a>
3890</div>
3891<div class="doc_text">
3892<h5>Syntax:</h5>
3893<pre> &lt;result&gt; = vfcmp &lt;cond&gt; &lt;ty&gt; &lt;var1&gt;, &lt;var2&gt;</pre>
3894<h5>Overview:</h5>
3895<p>The '<tt>vfcmp</tt>' instruction returns an integer vector value based on
3896element-wise comparison of its two floating point vector operands. The output
3897elements have the same width as the input elements.</p>
3898<h5>Arguments:</h5>
3899<p>The '<tt>vfcmp</tt>' instruction takes three operands. The first operand is
3900the condition code indicating the kind of comparison to perform. It is not
3901a value, just a keyword. The possible condition code are:
3902<ol>
3903 <li><tt>false</tt>: no comparison, always returns false</li>
3904 <li><tt>oeq</tt>: ordered and equal</li>
3905 <li><tt>ogt</tt>: ordered and greater than </li>
3906 <li><tt>oge</tt>: ordered and greater than or equal</li>
3907 <li><tt>olt</tt>: ordered and less than </li>
3908 <li><tt>ole</tt>: ordered and less than or equal</li>
3909 <li><tt>one</tt>: ordered and not equal</li>
3910 <li><tt>ord</tt>: ordered (no nans)</li>
3911 <li><tt>ueq</tt>: unordered or equal</li>
3912 <li><tt>ugt</tt>: unordered or greater than </li>
3913 <li><tt>uge</tt>: unordered or greater than or equal</li>
3914 <li><tt>ult</tt>: unordered or less than </li>
3915 <li><tt>ule</tt>: unordered or less than or equal</li>
3916 <li><tt>une</tt>: unordered or not equal</li>
3917 <li><tt>uno</tt>: unordered (either nans)</li>
3918 <li><tt>true</tt>: no comparison, always returns true</li>
3919</ol>
3920<p>The remaining two arguments must be <a href="#t_vector">vector</a> of
3921<a href="#t_floating">floating point</a> typed. They must also be identical
3922types.</p>
3923<h5>Semantics:</h5>
3924<p>The '<tt>vfcmp</tt>' instruction compares <tt>var1</tt> and <tt>var2</tt>
3925according to the condition code given as <tt>cond</tt>. The comparison yields a
3926<a href="#t_vector">vector</a> of <a href="#t_integer">integer</a> result, with
3927an identical number of elements as the values being compared, and each element
3928having identical with to the width of the floating point elements. The most
3929significant bit in each element is 1 if the element-wise comparison evaluates to
3930true, and is 0 otherwise. All other bits of the result are undefined. The
3931condition codes are evaluated identically to the
3932<a href="#i_fcmp">'<tt>fcmp</tt>' instruction</a>.
3933
3934<h5>Example:</h5>
3935<pre>
3936 &lt;result&gt; = vfcmp oeq <2 x float> < float 4, float 0 >, < float 5, float 0 > <i>; yields: result=<2 x i32> < i32 0, i32 -1 ></i>
3937 &lt;result&gt; = vfcmp ult <2 x double> < double 1, double 2 >, < double 2, double 2> <i>; yields: result=<2 x i64> < i64 -1, i64 0 ></i>
3938</pre>
3939</div>
3940
3941<!-- _______________________________________________________________________ -->
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3942<div class="doc_subsubsection"> <a name="i_phi">'<tt>phi</tt>'
3943Instruction</a> </div>
3944<div class="doc_text">
3945<h5>Syntax:</h5>
3946<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
3947<h5>Overview:</h5>
3948<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
3949the SSA graph representing the function.</p>
3950<h5>Arguments:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3951<p>The type of the incoming values is specified with the first type
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3952field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
3953as arguments, with one pair for each predecessor basic block of the
3954current block. Only values of <a href="#t_firstclass">first class</a>
3955type may be used as the value arguments to the PHI node. Only labels
3956may be used as the label arguments.</p>
3957<p>There must be no non-phi instructions between the start of a basic
3958block and the PHI instructions: i.e. PHI instructions must be first in
3959a basic block.</p>
3960<h5>Semantics:</h5>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]3961<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
3962specified by the pair corresponding to the predecessor basic block that executed
3963just prior to the current block.</p>
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3964<h5>Example:</h5>
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]3965<pre>Loop: ; Infinite loop that counts from 0 on up...<br> %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]<br> %nextindvar = add i32 %indvar, 1<br> br label %Loop<br></pre>
Reid Spencer97c5fa42006-11-08 01:18:52 +0000[diff] [blame]3966</div>
3967
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3968<!-- _______________________________________________________________________ -->
3969<div class="doc_subsubsection">
3970 <a name="i_select">'<tt>select</tt>' Instruction</a>
3971</div>
3972
3973<div class="doc_text">
3974
3975<h5>Syntax:</h5>
3976
3977<pre>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]3978 &lt;result&gt; = select i1 &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]3979</pre>
3980
3981<h5>Overview:</h5>
3982
3983<p>
3984The '<tt>select</tt>' instruction is used to choose one value based on a
3985condition, without branching.
3986</p>
3987
3988
3989<h5>Arguments:</h5>
3990
3991<p>
3992The '<tt>select</tt>' instruction requires a boolean value indicating the condition, and two values of the same <a href="#t_firstclass">first class</a> type.
3993</p>
3994
3995<h5>Semantics:</h5>
3996
3997<p>
3998If the boolean condition evaluates to true, the instruction returns the first
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]3999value argument; otherwise, it returns the second value argument.
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]4000</p>
4001
4002<h5>Example:</h5>
4003
4004<pre>
Reid Spencer36a15422007-01-12 03:35:51 +0000[diff] [blame]4005 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
Chris Lattnerb53c28d2004-03-12 05:50:16 +0000[diff] [blame]4006</pre>
4007</div>
4008
Robert Bocchinof72fdfe2006-01-15 20:48:27 +0000[diff] [blame]4009
4010<!-- _______________________________________________________________________ -->
4011<div class="doc_subsubsection">
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4012 <a name="i_call">'<tt>call</tt>' Instruction</a>
4013</div>
4014
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4015<div class="doc_text">
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4016
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]4017<h5>Syntax:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4018<pre>
Nick Lewyckya9b13d52007-09-08 13:57:50 +0000[diff] [blame]4019 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;param list&gt;)
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4020</pre>
4021
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]4022<h5>Overview:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4023
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4024<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4025
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]4026<h5>Arguments:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4027
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4028<p>This instruction requires several arguments:</p>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4029
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]4030<ol>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4031 <li>
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]4032 <p>The optional "tail" marker indicates whether the callee function accesses
4033 any allocas or varargs in the caller. If the "tail" marker is present, the
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4034 function call is eligible for tail call optimization. Note that calls may
4035 be marked "tail" even if they do not occur before a <a
4036 href="#i_ret"><tt>ret</tt></a> instruction.
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4037 </li>
4038 <li>
Duncan Sands16f122e2007-03-30 12:22:09 +0000[diff] [blame]4039 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
Chris Lattner0132aff2005-05-06 22:57:40 +0000[diff] [blame]4040 convention</a> the call should use. If none is specified, the call defaults
4041 to using C calling conventions.
4042 </li>
4043 <li>
Nick Lewyckya9b13d52007-09-08 13:57:50 +0000[diff] [blame]4044 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4045 the type of the return value. Functions that return no value are marked
4046 <tt><a href="#t_void">void</a></tt>.</p>
4047 </li>
4048 <li>
4049 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4050 value being invoked. The argument types must match the types implied by
4051 this signature. This type can be omitted if the function is not varargs
4052 and if the function type does not return a pointer to a function.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4053 </li>
4054 <li>
4055 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4056 be invoked. In most cases, this is a direct function invocation, but
4057 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]4058 to function value.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4059 </li>
4060 <li>
4061 <p>'<tt>function args</tt>': argument list whose types match the
Reid Spencerd845d162005-05-01 22:22:57 +0000[diff] [blame]4062 function signature argument types. All arguments must be of
4063 <a href="#t_firstclass">first class</a> type. If the function signature
4064 indicates the function accepts a variable number of arguments, the extra
4065 arguments can be specified.</p>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4066 </li>
Chris Lattnera8292f32002-05-06 22:08:29 +0000[diff] [blame]4067</ol>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4068
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]4069<h5>Semantics:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4070
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4071<p>The '<tt>call</tt>' instruction is used to cause control flow to
4072transfer to a specified function, with its incoming arguments bound to
4073the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4074instruction in the called function, control flow continues with the
4075instruction after the function call, and the return value of the
Chris Lattnerfb7c88d2008-03-21 17:24:17 +0000[diff] [blame]4076function is bound to the result argument. If the callee returns multiple
4077values then the return values of the function are only accessible through
4078the '<tt><a href="#i_getresult">getresult</a></tt>' instruction.</p>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4079
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]4080<h5>Example:</h5>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4081
4082<pre>
Nick Lewyckya9b13d52007-09-08 13:57:50 +0000[diff] [blame]4083 %retval = call i32 @test(i32 %argc)
Chris Lattnerfb7c88d2008-03-21 17:24:17 +0000[diff] [blame]4084 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4085 %X = tail call i32 @foo() <i>; yields i32</i>
4086 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4087 call void %foo(i8 97 signext)
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4088
4089 %struct.A = type { i32, i8 }
Chris Lattnerfb7c88d2008-03-21 17:24:17 +0000[diff] [blame]4090 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
4091 %gr = getresult %struct.A %r, 0 <i>; yields i32</i>
4092 %gr1 = getresult %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnere23c1392005-05-06 05:47:36 +0000[diff] [blame]4093</pre>
4094
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4095</div>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4096
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4097<!-- _______________________________________________________________________ -->
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4098<div class="doc_subsubsection">
Chris Lattner33337472006-01-13 23:26:01 +0000[diff] [blame]4099 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4100</div>
4101
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4102<div class="doc_text">
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4103
Chris Lattner26ca62e2003-10-18 05:51:36 +0000[diff] [blame]4104<h5>Syntax:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4105
4106<pre>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4107 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4108</pre>
4109
Chris Lattner26ca62e2003-10-18 05:51:36 +0000[diff] [blame]4110<h5>Overview:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4111
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4112<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4113the "variable argument" area of a function call. It is used to implement the
4114<tt>va_arg</tt> macro in C.</p>
4115
Chris Lattner26ca62e2003-10-18 05:51:36 +0000[diff] [blame]4116<h5>Arguments:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4117
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4118<p>This instruction takes a <tt>va_list*</tt> value and the type of
4119the argument. It returns a value of the specified argument type and
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4120increments the <tt>va_list</tt> to point to the next argument. The
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4121actual type of <tt>va_list</tt> is target specific.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4122
Chris Lattner26ca62e2003-10-18 05:51:36 +0000[diff] [blame]4123<h5>Semantics:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4124
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4125<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4126type from the specified <tt>va_list</tt> and causes the
4127<tt>va_list</tt> to point to the next argument. For more information,
4128see the variable argument handling <a href="#int_varargs">Intrinsic
4129Functions</a>.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4130
4131<p>It is legal for this instruction to be called in a function which does not
4132take a variable number of arguments, for example, the <tt>vfprintf</tt>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4133function.</p>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4134
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4135<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]4136href="#intrinsics">intrinsic function</a> because it takes a type as an
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4137argument.</p>
4138
Chris Lattner26ca62e2003-10-18 05:51:36 +0000[diff] [blame]4139<h5>Example:</h5>
Chris Lattner6a4a0492004-09-27 21:51:25 +0000[diff] [blame]4140
4141<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4142
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4143</div>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4144
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4145<!-- _______________________________________________________________________ -->
4146<div class="doc_subsubsection">
4147 <a name="i_getresult">'<tt>getresult</tt>' Instruction</a>
4148</div>
4149
4150<div class="doc_text">
4151
4152<h5>Syntax:</h5>
4153<pre>
Chris Lattner141b6132008-03-21 17:20:51 +0000[diff] [blame]4154 &lt;resultval&gt; = getresult &lt;type&gt; &lt;retval&gt;, &lt;index&gt;
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4155</pre>
Chris Lattner141b6132008-03-21 17:20:51 +0000[diff] [blame]4156
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4157<h5>Overview:</h5>
4158
4159<p> The '<tt>getresult</tt>' instruction is used to extract individual values
Chris Lattner141b6132008-03-21 17:20:51 +0000[diff] [blame]4160from a '<tt><a href="#i_call">call</a></tt>'
4161or '<tt><a href="#i_invoke">invoke</a></tt>' instruction that returns multiple
4162results.</p>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4163
4164<h5>Arguments:</h5>
4165
Chris Lattner141b6132008-03-21 17:20:51 +0000[diff] [blame]4166<p>The '<tt>getresult</tt>' instruction takes a call or invoke value as its
Chris Lattner1a640a62008-04-23 04:06:52 +0000[diff] [blame]4167first argument, or an undef value. The value must have <a
4168href="#t_struct">structure type</a>. The second argument is a constant
4169unsigned index value which must be in range for the number of values returned
4170by the call.</p>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4171
4172<h5>Semantics:</h5>
4173
Chris Lattner141b6132008-03-21 17:20:51 +0000[diff] [blame]4174<p>The '<tt>getresult</tt>' instruction extracts the element identified by
4175'<tt>index</tt>' from the aggregate value.</p>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4176
4177<h5>Example:</h5>
4178
4179<pre>
4180 %struct.A = type { i32, i8 }
4181
4182 %r = call %struct.A @foo()
Chris Lattner141b6132008-03-21 17:20:51 +0000[diff] [blame]4183 %gr = getresult %struct.A %r, 0 <i>; yields i32:%gr</i>
4184 %gr1 = getresult %struct.A %r, 1 <i>; yields i8:%gr1</i>
Devang Pateld6cff512008-03-10 20:49:15 +0000[diff] [blame]4185 add i32 %gr, 42
4186 add i8 %gr1, 41
4187</pre>
4188
4189</div>
4190
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4191<!-- *********************************************************************** -->
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4192<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4193<!-- *********************************************************************** -->
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4194
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4195<div class="doc_text">
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4196
4197<p>LLVM supports the notion of an "intrinsic function". These functions have
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]4198well known names and semantics and are required to follow certain restrictions.
4199Overall, these intrinsics represent an extension mechanism for the LLVM
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4200language that does not require changing all of the transformations in LLVM when
Gabor Greifa54634a2007-07-06 22:07:22 +0000[diff] [blame]4201adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4202
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]4203<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4204prefix is reserved in LLVM for intrinsic names; thus, function names may not
4205begin with this prefix. Intrinsic functions must always be external functions:
4206you cannot define the body of intrinsic functions. Intrinsic functions may
4207only be used in call or invoke instructions: it is illegal to take the address
4208of an intrinsic function. Additionally, because intrinsic functions are part
4209of the LLVM language, it is required if any are added that they be documented
4210here.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4211
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]4212<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4213a family of functions that perform the same operation but on different data
4214types. Because LLVM can represent over 8 million different integer types,
4215overloading is used commonly to allow an intrinsic function to operate on any
4216integer type. One or more of the argument types or the result type can be
4217overloaded to accept any integer type. Argument types may also be defined as
4218exactly matching a previous argument's type or the result type. This allows an
4219intrinsic function which accepts multiple arguments, but needs all of them to
4220be of the same type, to only be overloaded with respect to a single argument or
4221the result.</p>
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4222
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]4223<p>Overloaded intrinsics will have the names of its overloaded argument types
4224encoded into its function name, each preceded by a period. Only those types
4225which are overloaded result in a name suffix. Arguments whose type is matched
4226against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4227take an integer of any width and returns an integer of exactly the same integer
4228width. This leads to a family of functions such as
4229<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4230Only one type, the return type, is overloaded, and only one type suffix is
4231required. Because the argument's type is matched against the return type, it
4232does not require its own name suffix.</p>
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]4233
4234<p>To learn how to add an intrinsic function, please see the
4235<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4236</p>
4237
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4238</div>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4239
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4240<!-- ======================================================================= -->
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4241<div class="doc_subsection">
4242 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4243</div>
4244
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4245<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4246
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4247<p>Variable argument support is defined in LLVM with the <a
Chris Lattner33337472006-01-13 23:26:01 +0000[diff] [blame]4248 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4249intrinsic functions. These functions are related to the similarly
4250named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4251
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4252<p>All of these functions operate on arguments that use a
4253target-specific value type "<tt>va_list</tt>". The LLVM assembly
4254language reference manual does not define what this type is, so all
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4255transformations should be prepared to handle these functions regardless of
4256the type used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4257
Chris Lattner30b868d2006-05-15 17:26:46 +0000[diff] [blame]4258<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4259instruction and the variable argument handling intrinsic functions are
4260used.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4261
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]4262<div class="doc_code">
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4263<pre>
Anton Korobeynikov640bbe02007-03-21 23:58:04 +0000[diff] [blame]4264define i32 @test(i32 %X, ...) {
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4265 ; Initialize variable argument processing
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4266 %ap = alloca i8*
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4267 %ap2 = bitcast i8** %ap to i8*
Anton Korobeynikov640bbe02007-03-21 23:58:04 +0000[diff] [blame]4268 call void @llvm.va_start(i8* %ap2)
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4269
4270 ; Read a single integer argument
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4271 %tmp = va_arg i8** %ap, i32
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4272
4273 ; Demonstrate usage of llvm.va_copy and llvm.va_end
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4274 %aq = alloca i8*
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4275 %aq2 = bitcast i8** %aq to i8*
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4276 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
Anton Korobeynikov640bbe02007-03-21 23:58:04 +0000[diff] [blame]4277 call void @llvm.va_end(i8* %aq2)
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4278
4279 ; Stop processing of arguments.
Anton Korobeynikov640bbe02007-03-21 23:58:04 +0000[diff] [blame]4280 call void @llvm.va_end(i8* %ap2)
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4281 ret i32 %tmp
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4282}
Anton Korobeynikov640bbe02007-03-21 23:58:04 +0000[diff] [blame]4283
4284declare void @llvm.va_start(i8*)
4285declare void @llvm.va_copy(i8*, i8*)
4286declare void @llvm.va_end(i8*)
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4287</pre>
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4288</div>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4289
Bill Wendling3716c5d2007-05-29 09:04:49 +0000[diff] [blame]4290</div>
4291
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4292<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4293<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4294 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4295</div>
4296
4297
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4298<div class="doc_text">
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4299<h5>Syntax:</h5>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4300<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4301<h5>Overview:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4302<P>The '<tt>llvm.va_start</tt>' intrinsic initializes
4303<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4304href="#i_va_arg">va_arg</a></tt>.</p>
4305
4306<h5>Arguments:</h5>
4307
4308<P>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
4309
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4310<h5>Semantics:</h5>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4311
4312<P>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
4313macro available in C. In a target-dependent way, it initializes the
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4314<tt>va_list</tt> element to which the argument points, so that the next call to
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4315<tt>va_arg</tt> will produce the first variable argument passed to the function.
4316Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4317last argument of the function as the compiler can figure that out.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4318
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4319</div>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4320
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4321<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4322<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4323 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4324</div>
4325
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4326<div class="doc_text">
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4327<h5>Syntax:</h5>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4328<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4329<h5>Overview:</h5>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4330
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4331<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4332which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
Chris Lattner48b383b02003-11-25 01:02:51 +0000[diff] [blame]4333or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4334
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4335<h5>Arguments:</h5>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4336
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4337<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4338
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4339<h5>Semantics:</h5>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4340
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4341<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4342macro available in C. In a target-dependent way, it destroys the
4343<tt>va_list</tt> element to which the argument points. Calls to <a
4344href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4345<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4346<tt>llvm.va_end</tt>.</p>
Chris Lattnerdb0790c2007-01-08 07:55:15 +0000[diff] [blame]4347
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4348</div>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4349
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4350<!-- _______________________________________________________________________ -->
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4351<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4352 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4353</div>
4354
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4355<div class="doc_text">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4356
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4357<h5>Syntax:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4358
4359<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4360 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4361</pre>
4362
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4363<h5>Overview:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4364
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4365<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4366from the source argument list to the destination argument list.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4367
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4368<h5>Arguments:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4369
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4370<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Andrew Lenharth5305ea52005-06-22 20:38:11 +0000[diff] [blame]4371The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
Andrew Lenharth5fb787c2005-06-18 18:28:17 +0000[diff] [blame]4372
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4373
Chris Lattnerbd64b4e2003-05-08 04:57:36 +0000[diff] [blame]4374<h5>Semantics:</h5>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4375
Jeff Cohen222a8a42007-04-29 01:07:00 +0000[diff] [blame]4376<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4377macro available in C. In a target-dependent way, it copies the source
4378<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4379intrinsic is necessary because the <tt><a href="#int_va_start">
4380llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4381example, memory allocation.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4382
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]4383</div>
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4384
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4385<!-- ======================================================================= -->
4386<div class="doc_subsection">
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4387 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4388</div>
4389
4390<div class="doc_text">
4391
4392<p>
4393LLVM support for <a href="GarbageCollection.html">Accurate Garbage
4394Collection</a> requires the implementation and generation of these intrinsics.
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4395These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4396stack</a>, as well as garbage collector implementations that require <a
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4397href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4398Front-ends for type-safe garbage collected languages should generate these
4399intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4400href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4401</p>
Christopher Lamb55c6d4f2007-12-17 01:00:21 +0000[diff] [blame]4402
4403<p>The garbage collection intrinsics only operate on objects in the generic
4404 address space (address space zero).</p>
4405
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4406</div>
4407
4408<!-- _______________________________________________________________________ -->
4409<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4410 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4411</div>
4412
4413<div class="doc_text">
4414
4415<h5>Syntax:</h5>
4416
4417<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4418 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4419</pre>
4420
4421<h5>Overview:</h5>
4422
John Criswelldfe6a862004-12-10 15:51:16 +0000[diff] [blame]4423<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4424the code generator, and allows some metadata to be associated with it.</p>
4425
4426<h5>Arguments:</h5>
4427
4428<p>The first argument specifies the address of a stack object that contains the
4429root pointer. The second pointer (which must be either a constant or a global
4430value address) contains the meta-data to be associated with the root.</p>
4431
4432<h5>Semantics:</h5>
4433
Chris Lattner851b7712008-04-24 05:59:56 +0000[diff] [blame]4434<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4435location. At compile-time, the code generator generates information to allow
Gordon Henriksenfb56bde2007-12-25 02:31:26 +0000[diff] [blame]4436the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
4437intrinsic may only be used in a function which <a href="#gc">specifies a GC
4438algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4439
4440</div>
4441
4442
4443<!-- _______________________________________________________________________ -->
4444<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4445 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4446</div>
4447
4448<div class="doc_text">
4449
4450<h5>Syntax:</h5>
4451
4452<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4453 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4454</pre>
4455
4456<h5>Overview:</h5>
4457
4458<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
4459locations, allowing garbage collector implementations that require read
4460barriers.</p>
4461
4462<h5>Arguments:</h5>
4463
Chris Lattnerf9228072006-03-14 20:02:51 +0000[diff] [blame]4464<p>The second argument is the address to read from, which should be an address
4465allocated from the garbage collector. The first object is a pointer to the
4466start of the referenced object, if needed by the language runtime (otherwise
4467null).</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4468
4469<h5>Semantics:</h5>
4470
4471<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
4472instruction, but may be replaced with substantially more complex code by the
Gordon Henriksenfb56bde2007-12-25 02:31:26 +0000[diff] [blame]4473garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
4474may only be used in a function which <a href="#gc">specifies a GC
4475algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4476
4477</div>
4478
4479
4480<!-- _______________________________________________________________________ -->
4481<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4482 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4483</div>
4484
4485<div class="doc_text">
4486
4487<h5>Syntax:</h5>
4488
4489<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4490 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4491</pre>
4492
4493<h5>Overview:</h5>
4494
4495<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
4496locations, allowing garbage collector implementations that require write
4497barriers (such as generational or reference counting collectors).</p>
4498
4499<h5>Arguments:</h5>
4500
Chris Lattnerf9228072006-03-14 20:02:51 +0000[diff] [blame]4501<p>The first argument is the reference to store, the second is the start of the
4502object to store it to, and the third is the address of the field of Obj to
4503store to. If the runtime does not require a pointer to the object, Obj may be
4504null.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4505
4506<h5>Semantics:</h5>
4507
4508<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
4509instruction, but may be replaced with substantially more complex code by the
Gordon Henriksenfb56bde2007-12-25 02:31:26 +0000[diff] [blame]4510garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
4511may only be used in a function which <a href="#gc">specifies a GC
4512algorithm</a>.</p>
Chris Lattner757528b0b2004-05-23 21:06:01 +0000[diff] [blame]4513
4514</div>
4515
4516
4517
4518<!-- ======================================================================= -->
4519<div class="doc_subsection">
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4520 <a name="int_codegen">Code Generator Intrinsics</a>
4521</div>
4522
4523<div class="doc_text">
4524<p>
4525These intrinsics are provided by LLVM to expose special features that may only
4526be implemented with code generator support.
4527</p>
4528
4529</div>
4530
4531<!-- _______________________________________________________________________ -->
4532<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4533 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4534</div>
4535
4536<div class="doc_text">
4537
4538<h5>Syntax:</h5>
4539<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4540 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4541</pre>
4542
4543<h5>Overview:</h5>
4544
4545<p>
Chris Lattnerc1fb4262006-10-15 20:05:59 +0000[diff] [blame]4546The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
4547target-specific value indicating the return address of the current function
4548or one of its callers.
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4549</p>
4550
4551<h5>Arguments:</h5>
4552
4553<p>
4554The argument to this intrinsic indicates which function to return the address
4555for. Zero indicates the calling function, one indicates its caller, etc. The
4556argument is <b>required</b> to be a constant integer value.
4557</p>
4558
4559<h5>Semantics:</h5>
4560
4561<p>
4562The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
4563the return address of the specified call frame, or zero if it cannot be
4564identified. The value returned by this intrinsic is likely to be incorrect or 0
4565for arguments other than zero, so it should only be used for debugging purposes.
4566</p>
4567
4568<p>
4569Note that calling this intrinsic does not prevent function inlining or other
Chris Lattner2e6eb5f2005-03-07 20:30:51 +0000[diff] [blame]4570aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4571source-language caller.
4572</p>
4573</div>
4574
4575
4576<!-- _______________________________________________________________________ -->
4577<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4578 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4579</div>
4580
4581<div class="doc_text">
4582
4583<h5>Syntax:</h5>
4584<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4585 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4586</pre>
4587
4588<h5>Overview:</h5>
4589
4590<p>
Chris Lattnerc1fb4262006-10-15 20:05:59 +0000[diff] [blame]4591The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
4592target-specific frame pointer value for the specified stack frame.
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4593</p>
4594
4595<h5>Arguments:</h5>
4596
4597<p>
4598The argument to this intrinsic indicates which function to return the frame
4599pointer for. Zero indicates the calling function, one indicates its caller,
4600etc. The argument is <b>required</b> to be a constant integer value.
4601</p>
4602
4603<h5>Semantics:</h5>
4604
4605<p>
4606The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
4607the frame address of the specified call frame, or zero if it cannot be
4608identified. The value returned by this intrinsic is likely to be incorrect or 0
4609for arguments other than zero, so it should only be used for debugging purposes.
4610</p>
4611
4612<p>
4613Note that calling this intrinsic does not prevent function inlining or other
Chris Lattner2e6eb5f2005-03-07 20:30:51 +0000[diff] [blame]4614aggressive transformations, so the value returned may not be that of the obvious
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4615source-language caller.
4616</p>
4617</div>
4618
Chris Lattnerc8a2c222005-02-28 19:24:19 +0000[diff] [blame]4619<!-- _______________________________________________________________________ -->
4620<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4621 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4622</div>
4623
4624<div class="doc_text">
4625
4626<h5>Syntax:</h5>
4627<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4628 declare i8 *@llvm.stacksave()
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4629</pre>
4630
4631<h5>Overview:</h5>
4632
4633<p>
4634The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4635the function stack, for use with <a href="#int_stackrestore">
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4636<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
4637features like scoped automatic variable sized arrays in C99.
4638</p>
4639
4640<h5>Semantics:</h5>
4641
4642<p>
4643This intrinsic returns a opaque pointer value that can be passed to <a
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4644href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4645<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
4646<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
4647state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
4648practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
4649that were allocated after the <tt>llvm.stacksave</tt> was executed.
4650</p>
4651
4652</div>
4653
4654<!-- _______________________________________________________________________ -->
4655<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4656 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4657</div>
4658
4659<div class="doc_text">
4660
4661<h5>Syntax:</h5>
4662<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4663 declare void @llvm.stackrestore(i8 * %ptr)
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4664</pre>
4665
4666<h5>Overview:</h5>
4667
4668<p>
4669The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
4670the function stack to the state it was in when the corresponding <a
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4671href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4672useful for implementing language features like scoped automatic variable sized
4673arrays in C99.
4674</p>
4675
4676<h5>Semantics:</h5>
4677
4678<p>
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4679See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
Chris Lattner2f0f0012006-01-13 02:03:13 +0000[diff] [blame]4680</p>
4681
4682</div>
4683
4684
4685<!-- _______________________________________________________________________ -->
4686<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4687 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
Chris Lattnerc8a2c222005-02-28 19:24:19 +0000[diff] [blame]4688</div>
4689
4690<div class="doc_text">
4691
4692<h5>Syntax:</h5>
4693<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4694 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Chris Lattnerc8a2c222005-02-28 19:24:19 +0000[diff] [blame]4695</pre>
4696
4697<h5>Overview:</h5>
4698
4699
4700<p>
4701The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]4702a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
4703no
4704effect on the behavior of the program but can change its performance
Chris Lattnerff851072005-02-28 19:47:14 +0000[diff] [blame]4705characteristics.
Chris Lattnerc8a2c222005-02-28 19:24:19 +0000[diff] [blame]4706</p>
4707
4708<h5>Arguments:</h5>
4709
4710<p>
4711<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
4712determining if the fetch should be for a read (0) or write (1), and
4713<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
Chris Lattnerd3e641c2005-03-07 20:31:38 +0000[diff] [blame]4714locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
Chris Lattnerc8a2c222005-02-28 19:24:19 +0000[diff] [blame]4715<tt>locality</tt> arguments must be constant integers.
4716</p>
4717
4718<h5>Semantics:</h5>
4719
4720<p>
4721This intrinsic does not modify the behavior of the program. In particular,
4722prefetches cannot trap and do not produce a value. On targets that support this
4723intrinsic, the prefetch can provide hints to the processor cache for better
4724performance.
4725</p>
4726
4727</div>
4728
Andrew Lenharthb4427912005-03-28 20:05:49 +0000[diff] [blame]4729<!-- _______________________________________________________________________ -->
4730<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4731 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
Andrew Lenharthb4427912005-03-28 20:05:49 +0000[diff] [blame]4732</div>
4733
4734<div class="doc_text">
4735
4736<h5>Syntax:</h5>
4737<pre>
Chris Lattner12477732007-09-21 17:30:40 +0000[diff] [blame]4738 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Andrew Lenharthb4427912005-03-28 20:05:49 +0000[diff] [blame]4739</pre>
4740
4741<h5>Overview:</h5>
4742
4743
4744<p>
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]4745The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
4746(PC) in a region of
Andrew Lenharthb4427912005-03-28 20:05:49 +0000[diff] [blame]4747code to simulators and other tools. The method is target specific, but it is
4748expected that the marker will use exported symbols to transmit the PC of the marker.
Jeff Cohendc6bfea2005-11-11 02:15:27 +0000[diff] [blame]4749The marker makes no guarantees that it will remain with any specific instruction
Chris Lattnere64d41d2005-11-15 06:07:55 +0000[diff] [blame]4750after optimizations. It is possible that the presence of a marker will inhibit
Chris Lattnerb40261e2006-03-24 07:16:10 +0000[diff] [blame]4751optimizations. The intended use is to be inserted after optimizations to allow
John Criswell88190562005-05-16 16:17:45 +0000[diff] [blame]4752correlations of simulation runs.
Andrew Lenharthb4427912005-03-28 20:05:49 +0000[diff] [blame]4753</p>
4754
4755<h5>Arguments:</h5>
4756
4757<p>
4758<tt>id</tt> is a numerical id identifying the marker.
4759</p>
4760
4761<h5>Semantics:</h5>
4762
4763<p>
4764This intrinsic does not modify the behavior of the program. Backends that do not
4765support this intrinisic may ignore it.
4766</p>
4767
4768</div>
4769
Andrew Lenharth01aa5632005-11-11 16:47:30 +0000[diff] [blame]4770<!-- _______________________________________________________________________ -->
4771<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4772 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
Andrew Lenharth01aa5632005-11-11 16:47:30 +0000[diff] [blame]4773</div>
4774
4775<div class="doc_text">
4776
4777<h5>Syntax:</h5>
4778<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4779 declare i64 @llvm.readcyclecounter( )
Andrew Lenharth01aa5632005-11-11 16:47:30 +0000[diff] [blame]4780</pre>
4781
4782<h5>Overview:</h5>
4783
4784
4785<p>
4786The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
4787counter register (or similar low latency, high accuracy clocks) on those targets
4788that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
4789As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
4790should only be used for small timings.
4791</p>
4792
4793<h5>Semantics:</h5>
4794
4795<p>
4796When directly supported, reading the cycle counter should not modify any memory.
4797Implementations are allowed to either return a application specific value or a
4798system wide value. On backends without support, this is lowered to a constant 0.
4799</p>
4800
4801</div>
4802
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4803<!-- ======================================================================= -->
4804<div class="doc_subsection">
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4805 <a name="int_libc">Standard C Library Intrinsics</a>
4806</div>
4807
4808<div class="doc_text">
4809<p>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4810LLVM provides intrinsics for a few important standard C library functions.
4811These intrinsics allow source-language front-ends to pass information about the
4812alignment of the pointer arguments to the code generator, providing opportunity
4813for more efficient code generation.
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4814</p>
4815
4816</div>
4817
4818<!-- _______________________________________________________________________ -->
4819<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4820 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4821</div>
4822
4823<div class="doc_text">
4824
4825<h5>Syntax:</h5>
4826<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4827 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4828 i32 &lt;len&gt;, i32 &lt;align&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4829 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4830 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4831</pre>
4832
4833<h5>Overview:</h5>
4834
4835<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4836The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4837location to the destination location.
4838</p>
4839
4840<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4841Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
4842intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4843</p>
4844
4845<h5>Arguments:</h5>
4846
4847<p>
4848The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4849the source. The third argument is an integer argument
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4850specifying the number of bytes to copy, and the fourth argument is the alignment
4851of the source and destination locations.
4852</p>
4853
Chris Lattner4c67c482004-02-12 21:18:15 +0000[diff] [blame]4854<p>
4855If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattner5316e5d2006-03-04 00:02:10 +0000[diff] [blame]4856the caller guarantees that both the source and destination pointers are aligned
4857to that boundary.
Chris Lattner4c67c482004-02-12 21:18:15 +0000[diff] [blame]4858</p>
4859
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4860<h5>Semantics:</h5>
4861
4862<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4863The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
Chris Lattnerfee11462004-02-12 17:01:32 +0000[diff] [blame]4864location to the destination location, which are not allowed to overlap. It
4865copies "len" bytes of memory over. If the argument is known to be aligned to
4866some boundary, this can be specified as the fourth argument, otherwise it should
4867be set to 0 or 1.
4868</p>
4869</div>
4870
4871
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4872<!-- _______________________________________________________________________ -->
4873<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4874 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4875</div>
4876
4877<div class="doc_text">
4878
4879<h5>Syntax:</h5>
4880<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4881 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4882 i32 &lt;len&gt;, i32 &lt;align&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4883 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4884 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4885</pre>
4886
4887<h5>Overview:</h5>
4888
4889<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4890The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
4891location to the destination location. It is similar to the
Chris Lattnerec564022008-01-06 19:51:52 +0000[diff] [blame]4892'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4893</p>
4894
4895<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4896Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
4897intrinsics do not return a value, and takes an extra alignment argument.
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4898</p>
4899
4900<h5>Arguments:</h5>
4901
4902<p>
4903The first argument is a pointer to the destination, the second is a pointer to
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4904the source. The third argument is an integer argument
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4905specifying the number of bytes to copy, and the fourth argument is the alignment
4906of the source and destination locations.
4907</p>
4908
Chris Lattner4c67c482004-02-12 21:18:15 +0000[diff] [blame]4909<p>
4910If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattner5316e5d2006-03-04 00:02:10 +0000[diff] [blame]4911the caller guarantees that the source and destination pointers are aligned to
4912that boundary.
Chris Lattner4c67c482004-02-12 21:18:15 +0000[diff] [blame]4913</p>
4914
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4915<h5>Semantics:</h5>
4916
4917<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4918The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
Chris Lattnerf30152e2004-02-12 18:10:10 +0000[diff] [blame]4919location to the destination location, which may overlap. It
4920copies "len" bytes of memory over. If the argument is known to be aligned to
4921some boundary, this can be specified as the fourth argument, otherwise it should
4922be set to 0 or 1.
4923</p>
4924</div>
4925
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]4926
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4927<!-- _______________________________________________________________________ -->
4928<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4929 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4930</div>
4931
4932<div class="doc_text">
4933
4934<h5>Syntax:</h5>
4935<pre>
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4936 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4937 i32 &lt;len&gt;, i32 &lt;align&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]4938 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]4939 i64 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4940</pre>
4941
4942<h5>Overview:</h5>
4943
4944<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4945The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4946byte value.
4947</p>
4948
4949<p>
4950Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
4951does not return a value, and takes an extra alignment argument.
4952</p>
4953
4954<h5>Arguments:</h5>
4955
4956<p>
4957The first argument is a pointer to the destination to fill, the second is the
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4958byte value to fill it with, the third argument is an integer
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4959argument specifying the number of bytes to fill, and the fourth argument is the
4960known alignment of destination location.
4961</p>
4962
4963<p>
4964If the call to this intrinisic has an alignment value that is not 0 or 1, then
Chris Lattner5316e5d2006-03-04 00:02:10 +0000[diff] [blame]4965the caller guarantees that the destination pointer is aligned to that boundary.
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4966</p>
4967
4968<h5>Semantics:</h5>
4969
4970<p>
Chris Lattner0c8b2592006-03-03 00:07:20 +0000[diff] [blame]4971The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
4972the
Chris Lattner3649c3a2004-02-14 04:08:35 +0000[diff] [blame]4973destination location. If the argument is known to be aligned to some boundary,
4974this can be specified as the fourth argument, otherwise it should be set to 0 or
49751.
4976</p>
4977</div>
4978
4979
Chris Lattner3b4f4372004-06-11 02:28:03 +0000[diff] [blame]4980<!-- _______________________________________________________________________ -->
4981<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]4982 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
Chris Lattner8a8f2e52005-07-21 01:29:16 +0000[diff] [blame]4983</div>
4984
4985<div class="doc_text">
4986
4987<h5>Syntax:</h5>
Dale Johannesendd89d272007-10-02 17:47:38 +0000[diff] [blame]4988<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]4989floating point or vector of floating point type. Not all targets support all
4990types however.
Chris Lattner8a8f2e52005-07-21 01:29:16 +0000[diff] [blame]4991<pre>
Dale Johannesendd89d272007-10-02 17:47:38 +0000[diff] [blame]4992 declare float @llvm.sqrt.f32(float %Val)
4993 declare double @llvm.sqrt.f64(double %Val)
4994 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
4995 declare fp128 @llvm.sqrt.f128(fp128 %Val)
4996 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Chris Lattner8a8f2e52005-07-21 01:29:16 +0000[diff] [blame]4997</pre>
4998
4999<h5>Overview:</h5>
5000
5001<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +0000[diff] [blame]5002The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5003returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Chris Lattner8a8f2e52005-07-21 01:29:16 +0000[diff] [blame]5004<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattner00d7cb92008-01-29 07:00:44 +0000[diff] [blame]5005negative numbers other than -0.0 (which allows for better optimization, because
5006there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5007defined to return -0.0 like IEEE sqrt.
Chris Lattner8a8f2e52005-07-21 01:29:16 +0000[diff] [blame]5008</p>
5009
5010<h5>Arguments:</h5>
5011
5012<p>
5013The argument and return value are floating point numbers of the same type.
5014</p>
5015
5016<h5>Semantics:</h5>
5017
5018<p>
Dan Gohman33988db2007-07-16 14:37:41 +0000[diff] [blame]5019This function returns the sqrt of the specified operand if it is a nonnegative
Chris Lattner8a8f2e52005-07-21 01:29:16 +0000[diff] [blame]5020floating point number.
5021</p>
5022</div>
5023
Chris Lattner33b73f92006-09-08 06:34:02 +0000[diff] [blame]5024<!-- _______________________________________________________________________ -->
5025<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]5026 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
Chris Lattner33b73f92006-09-08 06:34:02 +0000[diff] [blame]5027</div>
5028
5029<div class="doc_text">
5030
5031<h5>Syntax:</h5>
Dale Johannesendd89d272007-10-02 17:47:38 +0000[diff] [blame]5032<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5033floating point or vector of floating point type. Not all targets support all
5034types however.
Chris Lattner33b73f92006-09-08 06:34:02 +0000[diff] [blame]5035<pre>
Dale Johannesendd89d272007-10-02 17:47:38 +0000[diff] [blame]5036 declare float @llvm.powi.f32(float %Val, i32 %power)
5037 declare double @llvm.powi.f64(double %Val, i32 %power)
5038 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5039 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5040 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Chris Lattner33b73f92006-09-08 06:34:02 +0000[diff] [blame]5041</pre>
5042
5043<h5>Overview:</h5>
5044
5045<p>
5046The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5047specified (positive or negative) power. The order of evaluation of
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5048multiplications is not defined. When a vector of floating point type is
5049used, the second argument remains a scalar integer value.
Chris Lattner33b73f92006-09-08 06:34:02 +0000[diff] [blame]5050</p>
5051
5052<h5>Arguments:</h5>
5053
5054<p>
5055The second argument is an integer power, and the first is a value to raise to
5056that power.
5057</p>
5058
5059<h5>Semantics:</h5>
5060
5061<p>
5062This function returns the first value raised to the second power with an
5063unspecified sequence of rounding operations.</p>
5064</div>
5065
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5066<!-- _______________________________________________________________________ -->
5067<div class="doc_subsubsection">
5068 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5069</div>
5070
5071<div class="doc_text">
5072
5073<h5>Syntax:</h5>
5074<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5075floating point or vector of floating point type. Not all targets support all
5076types however.
5077<pre>
5078 declare float @llvm.sin.f32(float %Val)
5079 declare double @llvm.sin.f64(double %Val)
5080 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5081 declare fp128 @llvm.sin.f128(fp128 %Val)
5082 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5083</pre>
5084
5085<h5>Overview:</h5>
5086
5087<p>
5088The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5089</p>
5090
5091<h5>Arguments:</h5>
5092
5093<p>
5094The argument and return value are floating point numbers of the same type.
5095</p>
5096
5097<h5>Semantics:</h5>
5098
5099<p>
5100This function returns the sine of the specified operand, returning the
5101same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmand0806a02007-10-17 18:05:13 +0000[diff] [blame]5102conditions in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5103</div>
5104
5105<!-- _______________________________________________________________________ -->
5106<div class="doc_subsubsection">
5107 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5108</div>
5109
5110<div class="doc_text">
5111
5112<h5>Syntax:</h5>
5113<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5114floating point or vector of floating point type. Not all targets support all
5115types however.
5116<pre>
5117 declare float @llvm.cos.f32(float %Val)
5118 declare double @llvm.cos.f64(double %Val)
5119 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5120 declare fp128 @llvm.cos.f128(fp128 %Val)
5121 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5122</pre>
5123
5124<h5>Overview:</h5>
5125
5126<p>
5127The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5128</p>
5129
5130<h5>Arguments:</h5>
5131
5132<p>
5133The argument and return value are floating point numbers of the same type.
5134</p>
5135
5136<h5>Semantics:</h5>
5137
5138<p>
5139This function returns the cosine of the specified operand, returning the
5140same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmand0806a02007-10-17 18:05:13 +0000[diff] [blame]5141conditions in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5142</div>
5143
5144<!-- _______________________________________________________________________ -->
5145<div class="doc_subsubsection">
5146 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5147</div>
5148
5149<div class="doc_text">
5150
5151<h5>Syntax:</h5>
5152<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5153floating point or vector of floating point type. Not all targets support all
5154types however.
5155<pre>
5156 declare float @llvm.pow.f32(float %Val, float %Power)
5157 declare double @llvm.pow.f64(double %Val, double %Power)
5158 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5159 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5160 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5161</pre>
5162
5163<h5>Overview:</h5>
5164
5165<p>
5166The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5167specified (positive or negative) power.
5168</p>
5169
5170<h5>Arguments:</h5>
5171
5172<p>
5173The second argument is a floating point power, and the first is a value to
5174raise to that power.
5175</p>
5176
5177<h5>Semantics:</h5>
5178
5179<p>
5180This function returns the first value raised to the second power,
5181returning the
5182same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmand0806a02007-10-17 18:05:13 +0000[diff] [blame]5183conditions in the same way.</p>
Dan Gohmanb6324c12007-10-15 20:30:11 +0000[diff] [blame]5184</div>
5185
Chris Lattner33b73f92006-09-08 06:34:02 +0000[diff] [blame]5186
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5187<!-- ======================================================================= -->
5188<div class="doc_subsection">
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5189 <a name="int_manip">Bit Manipulation Intrinsics</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5190</div>
5191
5192<div class="doc_text">
5193<p>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5194LLVM provides intrinsics for a few important bit manipulation operations.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5195These allow efficient code generation for some algorithms.
5196</p>
5197
5198</div>
5199
5200<!-- _______________________________________________________________________ -->
5201<div class="doc_subsubsection">
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]5202 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5203</div>
5204
5205<div class="doc_text">
5206
5207<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]5208<p>This is an overloaded intrinsic function. You can use bswap on any integer
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5209type that is an even number of bytes (i.e. BitWidth % 16 == 0).
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5210<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5211 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5212 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5213 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5214</pre>
5215
5216<h5>Overview:</h5>
5217
5218<p>
Reid Spencerf361c4f2007-04-02 02:25:19 +0000[diff] [blame]5219The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]5220values with an even number of bytes (positive multiple of 16 bits). These are
5221useful for performing operations on data that is not in the target's native
5222byte order.
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5223</p>
5224
5225<h5>Semantics:</h5>
5226
5227<p>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5228The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]5229and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5230intrinsic returns an i32 value that has the four bytes of the input i32
5231swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5232i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5233<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]5234additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
Nate Begeman0f223bb2006-01-13 23:26:38 +0000[diff] [blame]5235</p>
5236
5237</div>
5238
5239<!-- _______________________________________________________________________ -->
5240<div class="doc_subsubsection">
Reid Spencerb4f9a6f2006-01-16 21:12:35 +0000[diff] [blame]5241 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5242</div>
5243
5244<div class="doc_text">
5245
5246<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]5247<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
5248width. Not all targets support all bit widths however.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5249<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5250 declare i8 @llvm.ctpop.i8 (i8 &lt;src&gt;)
5251 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]5252 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5253 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5254 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5255</pre>
5256
5257<h5>Overview:</h5>
5258
5259<p>
Chris Lattner069b5bd2006-01-16 22:38:59 +0000[diff] [blame]5260The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5261value.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5262</p>
5263
5264<h5>Arguments:</h5>
5265
5266<p>
Chris Lattner573f64e2005-05-07 01:46:40 +0000[diff] [blame]5267The only argument is the value to be counted. The argument may be of any
Reid Spencer3e628eb92007-01-04 16:43:23 +0000[diff] [blame]5268integer type. The return type must match the argument type.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5269</p>
5270
5271<h5>Semantics:</h5>
5272
5273<p>
5274The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5275</p>
5276</div>
5277
5278<!-- _______________________________________________________________________ -->
5279<div class="doc_subsubsection">
Chris Lattnerb748c672006-01-16 22:34:14 +0000[diff] [blame]5280 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5281</div>
5282
5283<div class="doc_text">
5284
5285<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]5286<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
5287integer bit width. Not all targets support all bit widths however.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5288<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5289 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5290 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]5291 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5292 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5293 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5294</pre>
5295
5296<h5>Overview:</h5>
5297
5298<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +0000[diff] [blame]5299The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5300leading zeros in a variable.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5301</p>
5302
5303<h5>Arguments:</h5>
5304
5305<p>
Chris Lattner573f64e2005-05-07 01:46:40 +0000[diff] [blame]5306The only argument is the value to be counted. The argument may be of any
Reid Spencer3e628eb92007-01-04 16:43:23 +0000[diff] [blame]5307integer type. The return type must match the argument type.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5308</p>
5309
5310<h5>Semantics:</h5>
5311
5312<p>
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5313The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5314in a variable. If the src == 0 then the result is the size in bits of the type
Reid Spencerb5ebf3d2006-12-31 07:07:53 +0000[diff] [blame]5315of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
Andrew Lenharth1d463522005-05-03 18:01:48 +0000[diff] [blame]5316</p>
5317</div>
Chris Lattner3b4f4372004-06-11 02:28:03 +0000[diff] [blame]5318
5319
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5320
5321<!-- _______________________________________________________________________ -->
5322<div class="doc_subsubsection">
Chris Lattnerb748c672006-01-16 22:34:14 +0000[diff] [blame]5323 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5324</div>
5325
5326<div class="doc_text">
5327
5328<h5>Syntax:</h5>
Reid Spencer4eefaab2007-04-01 08:04:23 +0000[diff] [blame]5329<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
5330integer bit width. Not all targets support all bit widths however.
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5331<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5332 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5333 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Anton Korobeynikovbe9c93c2007-03-22 00:02:17 +0000[diff] [blame]5334 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5335 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5336 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5337</pre>
5338
5339<h5>Overview:</h5>
5340
5341<p>
Reid Spencerb4f9a6f2006-01-16 21:12:35 +0000[diff] [blame]5342The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5343trailing zeros.
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5344</p>
5345
5346<h5>Arguments:</h5>
5347
5348<p>
5349The only argument is the value to be counted. The argument may be of any
Reid Spencer3e628eb92007-01-04 16:43:23 +0000[diff] [blame]5350integer type. The return type must match the argument type.
Chris Lattnerefa20fa2005-05-15 19:39:26 +0000[diff] [blame]5351</p>
5352
5353<h5>Semantics:</h5>
5354
5355<p>
5356The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5357in a variable. If the src == 0 then the result is the size in bits of the type
5358of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5359</p>
5360</div>
5361
Reid Spencer8a5799f2007-04-01 08:27:01 +0000[diff] [blame]5362<!-- _______________________________________________________________________ -->
5363<div class="doc_subsubsection">
Reid Spencerea2945e2007-04-10 02:51:31 +0000[diff] [blame]5364 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5365</div>
5366
5367<div class="doc_text">
5368
5369<h5>Syntax:</h5>
Reid Spencerea2945e2007-04-10 02:51:31 +0000[diff] [blame]5370<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5371on any integer bit width.
5372<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5373 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5374 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5375</pre>
5376
5377<h5>Overview:</h5>
Reid Spencerea2945e2007-04-10 02:51:31 +0000[diff] [blame]5378<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5379range of bits from an integer value and returns them in the same bit width as
5380the original value.</p>
5381
5382<h5>Arguments:</h5>
5383<p>The first argument, <tt>%val</tt> and the result may be integer types of
5384any bit width but they must have the same bit width. The second and third
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]5385arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5386
5387<h5>Semantics:</h5>
Reid Spencerea2945e2007-04-10 02:51:31 +0000[diff] [blame]5388<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
Reid Spencer96a5f022007-04-04 02:42:35 +0000[diff] [blame]5389of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5390<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5391operates in forward mode.</p>
5392<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5393right by <tt>%loBit</tt> bits and then ANDing it with a mask with
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5394only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5395<ol>
5396 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5397 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5398 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5399 to determine the number of bits to retain.</li>
5400 <li>A mask of the retained bits is created by shifting a -1 value.</li>
5401 <li>The mask is ANDed with <tt>%val</tt> to produce the result.
5402</ol>
Reid Spencer70845c02007-05-14 16:14:57 +0000[diff] [blame]5403<p>In reverse mode, a similar computation is made except that the bits are
5404returned in the reverse order. So, for example, if <tt>X</tt> has the value
5405<tt>i16 0x0ACF (101011001111)</tt> and we apply
5406<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5407<tt>i16 0x0026 (000000100110)</tt>.</p>
Reid Spencer8bc7d952007-04-01 19:00:37 +0000[diff] [blame]5408</div>
5409
Reid Spencer5bf54c82007-04-11 23:23:49 +0000[diff] [blame]5410<div class="doc_subsubsection">
5411 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5412</div>
5413
5414<div class="doc_text">
5415
5416<h5>Syntax:</h5>
5417<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
5418on any integer bit width.
5419<pre>
Chandler Carruth7132e002007-08-04 01:51:18 +0000[diff] [blame]5420 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
5421 declare i29 @llvm.part.set.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
Reid Spencer5bf54c82007-04-11 23:23:49 +0000[diff] [blame]5422</pre>
5423
5424<h5>Overview:</h5>
5425<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
5426of bits in an integer value with another integer value. It returns the integer
5427with the replaced bits.</p>
5428
5429<h5>Arguments:</h5>
5430<p>The first argument, <tt>%val</tt> and the result may be integer types of
5431any bit width but they must have the same bit width. <tt>%val</tt> is the value
5432whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
5433integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
5434type since they specify only a bit index.</p>
5435
5436<h5>Semantics:</h5>
5437<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
5438of operation: forwards and reverse. If <tt>%lo</tt> is greater than
5439<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
5440operates in forward mode.</p>
5441<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
5442truncating it down to the size of the replacement area or zero extending it
5443up to that size.</p>
5444<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
5445are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
5446in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
5447to the <tt>%hi</tt>th bit.
Reid Spencer146281c2007-05-14 16:50:20 +0000[diff] [blame]5448<p>In reverse mode, a similar computation is made except that the bits are
5449reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
5450<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.
Reid Spencer5bf54c82007-04-11 23:23:49 +0000[diff] [blame]5451<h5>Examples:</h5>
5452<pre>
Reid Spencerc70afc32007-04-12 01:03:03 +0000[diff] [blame]5453 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
Reid Spencer146281c2007-05-14 16:50:20 +0000[diff] [blame]5454 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
5455 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
5456 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
Reid Spencerc70afc32007-04-12 01:03:03 +0000[diff] [blame]5457 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
Reid Spencer7972c472007-04-11 23:49:50 +0000[diff] [blame]5458</pre>
Reid Spencer5bf54c82007-04-11 23:23:49 +0000[diff] [blame]5459</div>
5460
Chris Lattner941515c2004-01-06 05:31:32 +0000[diff] [blame]5461<!-- ======================================================================= -->
5462<div class="doc_subsection">
5463 <a name="int_debugger">Debugger Intrinsics</a>
5464</div>
5465
5466<div class="doc_text">
5467<p>
5468The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
5469are described in the <a
5470href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
5471Debugging</a> document.
5472</p>
5473</div>
5474
5475
Jim Laskey2211f492007-03-14 19:31:19 +0000[diff] [blame]5476<!-- ======================================================================= -->
5477<div class="doc_subsection">
5478 <a name="int_eh">Exception Handling Intrinsics</a>
5479</div>
5480
5481<div class="doc_text">
5482<p> The LLVM exception handling intrinsics (which all start with
5483<tt>llvm.eh.</tt> prefix), are described in the <a
5484href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
5485Handling</a> document. </p>
5486</div>
5487
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]5488<!-- ======================================================================= -->
5489<div class="doc_subsection">
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5490 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5491</div>
5492
5493<div class="doc_text">
5494<p>
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5495 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5496 the <tt>nest</tt> attribute, from a function. The result is a callable
5497 function pointer lacking the nest parameter - the caller does not need
5498 to provide a value for it. Instead, the value to use is stored in
5499 advance in a "trampoline", a block of memory usually allocated
5500 on the stack, which also contains code to splice the nest value into the
5501 argument list. This is used to implement the GCC nested function address
5502 extension.
5503</p>
5504<p>
5505 For example, if the function is
5506 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
Bill Wendling252570f2007-09-22 09:23:55 +0000[diff] [blame]5507 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as follows:</p>
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5508<pre>
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5509 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
5510 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
5511 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
5512 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5513</pre>
Bill Wendling252570f2007-09-22 09:23:55 +0000[diff] [blame]5514 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
5515 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5516</div>
5517
5518<!-- _______________________________________________________________________ -->
5519<div class="doc_subsubsection">
5520 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
5521</div>
5522<div class="doc_text">
5523<h5>Syntax:</h5>
5524<pre>
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5525declare i8* @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5526</pre>
5527<h5>Overview:</h5>
5528<p>
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5529 This fills the memory pointed to by <tt>tramp</tt> with code
5530 and returns a function pointer suitable for executing it.
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5531</p>
5532<h5>Arguments:</h5>
5533<p>
5534 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
5535 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
5536 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sandsf2bcd372007-08-22 23:39:54 +0000[diff] [blame]5537 intrinsic. Note that the size and the alignment are target-specific - LLVM
5538 currently provides no portable way of determining them, so a front-end that
5539 generates this intrinsic needs to have some target-specific knowledge.
5540 The <tt>func</tt> argument must hold a function bitcast to an <tt>i8*</tt>.
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5541</p>
5542<h5>Semantics:</h5>
5543<p>
5544 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5545 dependent code, turning it into a function. A pointer to this function is
5546 returned, but needs to be bitcast to an
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5547 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands86e01192007-09-11 14:10:23 +0000[diff] [blame]5548 before being called. The new function's signature is the same as that of
5549 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
5550 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
5551 of pointer type. Calling the new function is equivalent to calling
5552 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
5553 missing <tt>nest</tt> argument. If, after calling
5554 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
5555 modified, then the effect of any later call to the returned function pointer is
5556 undefined.
Duncan Sands644f9172007-07-27 12:58:54 +0000[diff] [blame]5557</p>
5558</div>
5559
5560<!-- ======================================================================= -->
5561<div class="doc_subsection">
Andrew Lenharth9b254ee2008-02-16 01:24:58 +0000[diff] [blame]5562 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
5563</div>
5564
5565<div class="doc_text">
5566<p>
5567 These intrinsic functions expand the "universal IR" of LLVM to represent
5568 hardware constructs for atomic operations and memory synchronization. This
5569 provides an interface to the hardware, not an interface to the programmer. It
5570 is aimed at a low enough level to allow any programming models or APIs which
5571 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
5572 hardware behavior. Just as hardware provides a "universal IR" for source
5573 languages, it also provides a starting point for developing a "universal"
5574 atomic operation and synchronization IR.
5575</p>
5576<p>
5577 These do <em>not</em> form an API such as high-level threading libraries,
5578 software transaction memory systems, atomic primitives, and intrinsic
5579 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
5580 application libraries. The hardware interface provided by LLVM should allow
5581 a clean implementation of all of these APIs and parallel programming models.
5582 No one model or paradigm should be selected above others unless the hardware
5583 itself ubiquitously does so.
5584
5585</p>
5586</div>
5587
5588<!-- _______________________________________________________________________ -->
5589<div class="doc_subsubsection">
5590 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
5591</div>
5592<div class="doc_text">
5593<h5>Syntax:</h5>
5594<pre>
5595declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
5596i1 &lt;device&gt; )
5597
5598</pre>
5599<h5>Overview:</h5>
5600<p>
5601 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
5602 specific pairs of memory access types.
5603</p>
5604<h5>Arguments:</h5>
5605<p>
5606 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
5607 The first four arguments enables a specific barrier as listed below. The fith
5608 argument specifies that the barrier applies to io or device or uncached memory.
5609
5610</p>
5611 <ul>
5612 <li><tt>ll</tt>: load-load barrier</li>
5613 <li><tt>ls</tt>: load-store barrier</li>
5614 <li><tt>sl</tt>: store-load barrier</li>
5615 <li><tt>ss</tt>: store-store barrier</li>
5616 <li><tt>device</tt>: barrier applies to device and uncached memory also.
5617 </ul>
5618<h5>Semantics:</h5>
5619<p>
5620 This intrinsic causes the system to enforce some ordering constraints upon
5621 the loads and stores of the program. This barrier does not indicate
5622 <em>when</em> any events will occur, it only enforces an <em>order</em> in
5623 which they occur. For any of the specified pairs of load and store operations
5624 (f.ex. load-load, or store-load), all of the first operations preceding the
5625 barrier will complete before any of the second operations succeeding the
5626 barrier begin. Specifically the semantics for each pairing is as follows:
5627</p>
5628 <ul>
5629 <li><tt>ll</tt>: All loads before the barrier must complete before any load
5630 after the barrier begins.</li>
5631
5632 <li><tt>ls</tt>: All loads before the barrier must complete before any
5633 store after the barrier begins.</li>
5634 <li><tt>ss</tt>: All stores before the barrier must complete before any
5635 store after the barrier begins.</li>
5636 <li><tt>sl</tt>: All stores before the barrier must complete before any
5637 load after the barrier begins.</li>
5638 </ul>
5639<p>
5640 These semantics are applied with a logical "and" behavior when more than one
5641 is enabled in a single memory barrier intrinsic.
5642</p>
5643<p>
5644 Backends may implement stronger barriers than those requested when they do not
5645 support as fine grained a barrier as requested. Some architectures do not
5646 need all types of barriers and on such architectures, these become noops.
5647</p>
5648<h5>Example:</h5>
5649<pre>
5650%ptr = malloc i32
5651 store i32 4, %ptr
5652
5653%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
5654 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
5655 <i>; guarantee the above finishes</i>
5656 store i32 8, %ptr <i>; before this begins</i>
5657</pre>
5658</div>
5659
Andrew Lenharth95528942008-02-21 06:45:13 +0000[diff] [blame]5660<!-- _______________________________________________________________________ -->
5661<div class="doc_subsubsection">
5662 <a name="int_atomic_lcs">'<tt>llvm.atomic.lcs.*</tt>' Intrinsic</a>
5663</div>
5664<div class="doc_text">
5665<h5>Syntax:</h5>
5666<p>
5667 This is an overloaded intrinsic. You can use <tt>llvm.atomic.lcs</tt> on any
5668 integer bit width. Not all targets support all bit widths however.</p>
5669
5670<pre>
5671declare i8 @llvm.atomic.lcs.i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
5672declare i16 @llvm.atomic.lcs.i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
5673declare i32 @llvm.atomic.lcs.i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
5674declare i64 @llvm.atomic.lcs.i64( i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt; )
5675
5676</pre>
5677<h5>Overview:</h5>
5678<p>
5679 This loads a value in memory and compares it to a given value. If they are
5680 equal, it stores a new value into the memory.
5681</p>
5682<h5>Arguments:</h5>
5683<p>
5684 The <tt>llvm.atomic.lcs</tt> intrinsic takes three arguments. The result as
5685 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
5686 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
5687 this integer type. While any bit width integer may be used, targets may only
5688 lower representations they support in hardware.
5689
5690</p>
5691<h5>Semantics:</h5>
5692<p>
5693 This entire intrinsic must be executed atomically. It first loads the value
5694 in memory pointed to by <tt>ptr</tt> and compares it with the value
5695 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
5696 loaded value is yielded in all cases. This provides the equivalent of an
5697 atomic compare-and-swap operation within the SSA framework.
5698</p>
5699<h5>Examples:</h5>
5700
5701<pre>
5702%ptr = malloc i32
5703 store i32 4, %ptr
5704
5705%val1 = add i32 4, 4
5706%result1 = call i32 @llvm.atomic.lcs.i32( i32* %ptr, i32 4, %val1 )
5707 <i>; yields {i32}:result1 = 4</i>
5708%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5709%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5710
5711%val2 = add i32 1, 1
5712%result2 = call i32 @llvm.atomic.lcs.i32( i32* %ptr, i32 5, %val2 )
5713 <i>; yields {i32}:result2 = 8</i>
5714%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
5715
5716%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
5717</pre>
5718</div>
5719
5720<!-- _______________________________________________________________________ -->
5721<div class="doc_subsubsection">
5722 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
5723</div>
5724<div class="doc_text">
5725<h5>Syntax:</h5>
5726
5727<p>
5728 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
5729 integer bit width. Not all targets support all bit widths however.</p>
5730<pre>
5731declare i8 @llvm.atomic.swap.i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
5732declare i16 @llvm.atomic.swap.i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
5733declare i32 @llvm.atomic.swap.i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
5734declare i64 @llvm.atomic.swap.i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
5735
5736</pre>
5737<h5>Overview:</h5>
5738<p>
5739 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
5740 the value from memory. It then stores the value in <tt>val</tt> in the memory
5741 at <tt>ptr</tt>.
5742</p>
5743<h5>Arguments:</h5>
5744
5745<p>
5746 The <tt>llvm.atomic.ls</tt> intrinsic takes two arguments. Both the
5747 <tt>val</tt> argument and the result must be integers of the same bit width.
5748 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
5749 integer type. The targets may only lower integer representations they
5750 support.
5751</p>
5752<h5>Semantics:</h5>
5753<p>
5754 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
5755 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
5756 equivalent of an atomic swap operation within the SSA framework.
5757
5758</p>
5759<h5>Examples:</h5>
5760<pre>
5761%ptr = malloc i32
5762 store i32 4, %ptr
5763
5764%val1 = add i32 4, 4
5765%result1 = call i32 @llvm.atomic.swap.i32( i32* %ptr, i32 %val1 )
5766 <i>; yields {i32}:result1 = 4</i>
5767%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
5768%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
5769
5770%val2 = add i32 1, 1
5771%result2 = call i32 @llvm.atomic.swap.i32( i32* %ptr, i32 %val2 )
5772 <i>; yields {i32}:result2 = 8</i>
5773
5774%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
5775%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
5776</pre>
5777</div>
5778
5779<!-- _______________________________________________________________________ -->
5780<div class="doc_subsubsection">
5781 <a name="int_atomic_las">'<tt>llvm.atomic.las.*</tt>' Intrinsic</a>
5782
5783</div>
5784<div class="doc_text">
5785<h5>Syntax:</h5>
5786<p>
5787 This is an overloaded intrinsic. You can use <tt>llvm.atomic.las</tt> on any
5788 integer bit width. Not all targets support all bit widths however.</p>
5789<pre>
5790declare i8 @llvm.atomic.las.i8.( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
5791declare i16 @llvm.atomic.las.i16.( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
5792declare i32 @llvm.atomic.las.i32.( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
5793declare i64 @llvm.atomic.las.i64.( i64* &lt;ptr&gt;, i64 &lt;delta&gt; )
5794
5795</pre>
5796<h5>Overview:</h5>
5797<p>
5798 This intrinsic adds <tt>delta</tt> to the value stored in memory at
5799 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
5800</p>
5801<h5>Arguments:</h5>
5802<p>
5803
5804 The intrinsic takes two arguments, the first a pointer to an integer value
5805 and the second an integer value. The result is also an integer value. These
5806 integer types can have any bit width, but they must all have the same bit
5807 width. The targets may only lower integer representations they support.
5808</p>
5809<h5>Semantics:</h5>
5810<p>
5811 This intrinsic does a series of operations atomically. It first loads the
5812 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
5813 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
5814</p>
5815
5816<h5>Examples:</h5>
5817<pre>
5818%ptr = malloc i32
5819 store i32 4, %ptr
5820%result1 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 4 )
5821 <i>; yields {i32}:result1 = 4</i>
5822%result2 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 2 )
5823 <i>; yields {i32}:result2 = 8</i>
5824%result3 = call i32 @llvm.atomic.las.i32( i32* %ptr, i32 5 )
5825 <i>; yields {i32}:result3 = 10</i>
5826%memval = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
5827</pre>
5828</div>
5829
Andrew Lenharth9b254ee2008-02-16 01:24:58 +0000[diff] [blame]5830
5831<!-- ======================================================================= -->
5832<div class="doc_subsection">
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]5833 <a name="int_general">General Intrinsics</a>
5834</div>
5835
5836<div class="doc_text">
5837<p> This class of intrinsics is designed to be generic and has
5838no specific purpose. </p>
5839</div>
5840
5841<!-- _______________________________________________________________________ -->
5842<div class="doc_subsubsection">
5843 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
5844</div>
5845
5846<div class="doc_text">
5847
5848<h5>Syntax:</h5>
5849<pre>
Tanya Lattnerbed1d4d2007-06-18 23:42:37 +0000[diff] [blame]5850 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]5851</pre>
5852
5853<h5>Overview:</h5>
5854
5855<p>
5856The '<tt>llvm.var.annotation</tt>' intrinsic
5857</p>
5858
5859<h5>Arguments:</h5>
5860
5861<p>
Tanya Lattnerbed1d4d2007-06-18 23:42:37 +0000[diff] [blame]5862The first argument is a pointer to a value, the second is a pointer to a
5863global string, the third is a pointer to a global string which is the source
5864file name, and the last argument is the line number.
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]5865</p>
5866
5867<h5>Semantics:</h5>
5868
5869<p>
Anton Korobeynikov06cbb652008-01-15 22:31:34 +0000[diff] [blame]5870This intrinsic allows annotation of local variables with arbitrary strings.
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]5871This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikov06cbb652008-01-15 22:31:34 +0000[diff] [blame]5872annotations. These have no other defined use, they are ignored by code
5873generation and optimization.
5874</p>
Tanya Lattnercb1b9602007-06-15 20:50:54 +0000[diff] [blame]5875</div>
5876
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]5877<!-- _______________________________________________________________________ -->
5878<div class="doc_subsubsection">
Tanya Lattner0186a652007-09-21 23:57:59 +0000[diff] [blame]5879 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]5880</div>
5881
5882<div class="doc_text">
5883
5884<h5>Syntax:</h5>
Tanya Lattner23dbd572007-09-21 23:56:27 +0000[diff] [blame]5885<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
5886any integer bit width.
5887</p>
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]5888<pre>
Tanya Lattnercf3e26f2007-09-22 00:03:01 +0000[diff] [blame]5889 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5890 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5891 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5892 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
5893 declare i256 @llvm.annotation.i256(i256 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]5894</pre>
5895
5896<h5>Overview:</h5>
Tanya Lattner23dbd572007-09-21 23:56:27 +0000[diff] [blame]5897
5898<p>
5899The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]5900</p>
5901
5902<h5>Arguments:</h5>
5903
5904<p>
5905The first argument is an integer value (result of some expression),
5906the second is a pointer to a global string, the third is a pointer to a global
5907string which is the source file name, and the last argument is the line number.
Tanya Lattner23dbd572007-09-21 23:56:27 +0000[diff] [blame]5908It returns the value of the first argument.
Tanya Lattner293c0372007-09-21 22:59:12 +0000[diff] [blame]5909</p>
5910
5911<h5>Semantics:</h5>
5912
5913<p>
5914This intrinsic allows annotations to be put on arbitrary expressions
5915with arbitrary strings. This can be useful for special purpose optimizations
5916that want to look for these annotations. These have no other defined use, they
5917are ignored by code generation and optimization.
5918</div>
Jim Laskey2211f492007-03-14 19:31:19 +0000[diff] [blame]5919
Anton Korobeynikov06cbb652008-01-15 22:31:34 +0000[diff] [blame]5920<!-- _______________________________________________________________________ -->
5921<div class="doc_subsubsection">
5922 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
5923</div>
5924
5925<div class="doc_text">
5926
5927<h5>Syntax:</h5>
5928<pre>
5929 declare void @llvm.trap()
5930</pre>
5931
5932<h5>Overview:</h5>
5933
5934<p>
5935The '<tt>llvm.trap</tt>' intrinsic
5936</p>
5937
5938<h5>Arguments:</h5>
5939
5940<p>
5941None
5942</p>
5943
5944<h5>Semantics:</h5>
5945
5946<p>
5947This intrinsics is lowered to the target dependent trap instruction. If the
5948target does not have a trap instruction, this intrinsic will be lowered to the
5949call of the abort() function.
5950</p>
5951</div>
5952
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]5953<!-- *********************************************************************** -->
Chris Lattner2f7c9632001-06-06 20:29:01 +0000[diff] [blame]5954<hr>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]5955<address>
5956 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
5957 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
5958 <a href="http://validator.w3.org/check/referer"><img
Chris Lattnerb8f816e2008-01-04 04:33:49 +0000[diff] [blame]5959 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]5960
5961 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencerca058542006-03-14 05:39:39 +0000[diff] [blame]5962 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
Misha Brukmanc501f552004-03-01 17:47:27 +0000[diff] [blame]5963 Last modified: $Date$
5964</address>
Chris Lattnerb8f816e2008-01-04 04:33:49 +0000[diff] [blame]5965
Misha Brukman76307852003-11-08 01:05:38 +0000[diff] [blame]5966</body>
5967</html>