Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 55202eaf6299c8be2aaa97697c6dfccbbbfcc81c / . / docs / LangRef.html
blob: 32f143a93ae236ddca7c1e74b5eda46175af0b4f [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]8 <meta name="description"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
12
13<body>
14
15<div class="doc_title"> LLVM Language Reference Manual </div>
16<ol>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a>
24 <ol>
Bill Wendlinge2753242009-07-20 02:41:50 +0000[diff] [blame]25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
27 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
28 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
29 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
30 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
31 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
32 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
33 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner0fee5c22009-10-10 18:26:06 +0000[diff] [blame]34 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendlinge2753242009-07-20 02:41:50 +0000[diff] [blame]35 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
36 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
37 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]39 </ol>
40 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]41 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]42 <li><a href="#namedtypes">Named Types</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]43 <li><a href="#globalvars">Global Variables</a></li>
44 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]45 <li><a href="#aliasstructure">Aliases</a></li>
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]46 <li><a href="#namedmetadatastructure">Named Metadata</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]47 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]48 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]49 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]50 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
51 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]52 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]53 <li><a href="#volatile">Volatile Memory Accesses</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]54 </ol>
55 </li>
56 <li><a href="#typesystem">Type System</a>
57 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]58 <li><a href="#t_classifications">Type Classifications</a></li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]59 <li><a href="#t_primitive">Primitive Types</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]60 <ol>
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]61 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]62 <li><a href="#t_floating">Floating Point Types</a></li>
63 <li><a href="#t_void">Void Type</a></li>
64 <li><a href="#t_label">Label Type</a></li>
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]65 <li><a href="#t_metadata">Metadata Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]66 </ol>
67 </li>
68 <li><a href="#t_derived">Derived Types</a>
69 <ol>
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]70 <li><a href="#t_aggregate">Aggregate Types</a>
71 <ol>
72 <li><a href="#t_array">Array Type</a></li>
73 <li><a href="#t_struct">Structure Type</a></li>
74 <li><a href="#t_pstruct">Packed Structure Type</a></li>
75 <li><a href="#t_union">Union Type</a></li>
76 <li><a href="#t_vector">Vector Type</a></li>
77 </ol>
78 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]79 <li><a href="#t_function">Function Type</a></li>
80 <li><a href="#t_pointer">Pointer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]81 <li><a href="#t_opaque">Opaque Type</a></li>
82 </ol>
83 </li>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]84 <li><a href="#t_uprefs">Type Up-references</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]85 </ol>
86 </li>
87 <li><a href="#constants">Constants</a>
88 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]89 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]90 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]91 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
92 <li><a href="#undefvalues">Undefined Values</a></li>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]93 <li><a href="#trapvalues">Trap Values</a></li>
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]94 <li><a href="#blockaddress">Addresses of Basic Blocks</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]95 <li><a href="#constantexprs">Constant Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]96 </ol>
97 </li>
98 <li><a href="#othervalues">Other Values</a>
99 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]100 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]101 <li><a href="#metadata">Metadata Nodes and Metadata Strings</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]102 </ol>
103 </li>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]104 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
105 <ol>
106 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]107 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
108 Global Variable</a></li>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]109 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
110 Global Variable</a></li>
111 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
112 Global Variable</a></li>
113 </ol>
114 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]115 <li><a href="#instref">Instruction Reference</a>
116 <ol>
117 <li><a href="#terminators">Terminator Instructions</a>
118 <ol>
119 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
120 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
121 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]122 <li><a href="#i_indirectbr">'<tt>indirectbr</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]123 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
124 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
125 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
126 </ol>
127 </li>
128 <li><a href="#binaryops">Binary Operations</a>
129 <ol>
130 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]131 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]132 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]133 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]134 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]135 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]136 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
137 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
138 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
139 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
140 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
141 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
142 </ol>
143 </li>
144 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
145 <ol>
146 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
147 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
148 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
149 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
150 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
151 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
152 </ol>
153 </li>
154 <li><a href="#vectorops">Vector Operations</a>
155 <ol>
156 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
157 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
158 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
159 </ol>
160 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]161 <li><a href="#aggregateops">Aggregate Operations</a>
162 <ol>
163 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
164 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
165 </ol>
166 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]167 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
168 <ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]169 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
170 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
171 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
172 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
173 </ol>
174 </li>
175 <li><a href="#convertops">Conversion Operations</a>
176 <ol>
177 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
178 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
179 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
180 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
181 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
182 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
183 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
184 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
185 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
186 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
187 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
188 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
189 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]190 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]191 <li><a href="#otherops">Other Operations</a>
192 <ol>
193 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
194 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
195 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
196 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
197 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
198 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
199 </ol>
200 </li>
201 </ol>
202 </li>
203 <li><a href="#intrinsics">Intrinsic Functions</a>
204 <ol>
205 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
206 <ol>
207 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
208 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
209 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
210 </ol>
211 </li>
212 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
213 <ol>
214 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
215 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
216 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
217 </ol>
218 </li>
219 <li><a href="#int_codegen">Code Generator Intrinsics</a>
220 <ol>
221 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
222 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
223 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
224 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
225 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
226 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
Dan Gohmanf15a6b22010-05-26 21:56:15 +0000[diff] [blame]227 <li><a href="#int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]228 </ol>
229 </li>
230 <li><a href="#int_libc">Standard C Library Intrinsics</a>
231 <ol>
232 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
233 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
234 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
235 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
236 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]237 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
238 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
239 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]240 </ol>
241 </li>
242 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
243 <ol>
244 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
245 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
246 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
247 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]248 </ol>
249 </li>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]250 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
251 <ol>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]252 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
253 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
254 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
255 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
256 <li><a href="#int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]257 <li><a href="#int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt> Intrinsics</a></li>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]258 </ol>
259 </li>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]260 <li><a href="#int_fp16">Half Precision Floating Point Intrinsics</a>
261 <ol>
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]262 <li><a href="#int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a></li>
263 <li><a href="#int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a></li>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]264 </ol>
265 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]266 <li><a href="#int_debugger">Debugger intrinsics</a></li>
267 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]268 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]269 <ol>
270 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]271 </ol>
272 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]273 <li><a href="#int_atomics">Atomic intrinsics</a>
274 <ol>
275 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
276 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
277 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
278 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
279 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
280 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
281 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
282 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
283 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
284 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
285 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
286 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
287 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
288 </ol>
289 </li>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]290 <li><a href="#int_memorymarkers">Memory Use Markers</a>
291 <ol>
292 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
293 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
294 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
295 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
296 </ol>
297 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]298 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]299 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]300 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]301 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]302 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]303 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]304 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]305 '<tt>llvm.trap</tt>' Intrinsic</a></li>
306 <li><a href="#int_stackprotector">
307 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]308 <li><a href="#int_objectsize">
309 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]310 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]311 </li>
312 </ol>
313 </li>
314</ol>
315
316<div class="doc_author">
317 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
318 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
319</div>
320
321<!-- *********************************************************************** -->
322<div class="doc_section"> <a name="abstract">Abstract </a></div>
323<!-- *********************************************************************** -->
324
325<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]326
327<p>This document is a reference manual for the LLVM assembly language. LLVM is
328 a Static Single Assignment (SSA) based representation that provides type
329 safety, low-level operations, flexibility, and the capability of representing
330 'all' high-level languages cleanly. It is the common code representation
331 used throughout all phases of the LLVM compilation strategy.</p>
332
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]333</div>
334
335<!-- *********************************************************************** -->
336<div class="doc_section"> <a name="introduction">Introduction</a> </div>
337<!-- *********************************************************************** -->
338
339<div class="doc_text">
340
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]341<p>The LLVM code representation is designed to be used in three different forms:
342 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
343 for fast loading by a Just-In-Time compiler), and as a human readable
344 assembly language representation. This allows LLVM to provide a powerful
345 intermediate representation for efficient compiler transformations and
346 analysis, while providing a natural means to debug and visualize the
347 transformations. The three different forms of LLVM are all equivalent. This
348 document describes the human readable representation and notation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]349
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]350<p>The LLVM representation aims to be light-weight and low-level while being
351 expressive, typed, and extensible at the same time. It aims to be a
352 "universal IR" of sorts, by being at a low enough level that high-level ideas
353 may be cleanly mapped to it (similar to how microprocessors are "universal
354 IR's", allowing many source languages to be mapped to them). By providing
355 type information, LLVM can be used as the target of optimizations: for
356 example, through pointer analysis, it can be proven that a C automatic
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]357 variable is never accessed outside of the current function, allowing it to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]358 be promoted to a simple SSA value instead of a memory location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]359
360</div>
361
362<!-- _______________________________________________________________________ -->
363<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
364
365<div class="doc_text">
366
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]367<p>It is important to note that this document describes 'well formed' LLVM
368 assembly language. There is a difference between what the parser accepts and
369 what is considered 'well formed'. For example, the following instruction is
370 syntactically okay, but not well formed:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]371
372<div class="doc_code">
373<pre>
374%x = <a href="#i_add">add</a> i32 1, %x
375</pre>
376</div>
377
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]378<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
379 LLVM infrastructure provides a verification pass that may be used to verify
380 that an LLVM module is well formed. This pass is automatically run by the
381 parser after parsing input assembly and by the optimizer before it outputs
382 bitcode. The violations pointed out by the verifier pass indicate bugs in
383 transformation passes or input to the parser.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]384
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]385</div>
386
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]387<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]388
389<!-- *********************************************************************** -->
390<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
391<!-- *********************************************************************** -->
392
393<div class="doc_text">
394
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]395<p>LLVM identifiers come in two basic types: global and local. Global
396 identifiers (functions, global variables) begin with the <tt>'@'</tt>
397 character. Local identifiers (register names, types) begin with
398 the <tt>'%'</tt> character. Additionally, there are three different formats
399 for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]400
401<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]402 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]403 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
404 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
405 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
406 other characters in their names can be surrounded with quotes. Special
407 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
408 ASCII code for the character in hexadecimal. In this way, any character
409 can be used in a name value, even quotes themselves.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]410
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]411 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]412 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]413
414 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]415 constants</a>, below.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]416</ol>
417
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]418<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]419 don't need to worry about name clashes with reserved words, and the set of
420 reserved words may be expanded in the future without penalty. Additionally,
421 unnamed identifiers allow a compiler to quickly come up with a temporary
422 variable without having to avoid symbol table conflicts.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]423
424<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]425 languages. There are keywords for different opcodes
426 ('<tt><a href="#i_add">add</a></tt>',
427 '<tt><a href="#i_bitcast">bitcast</a></tt>',
428 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
429 ('<tt><a href="#t_void">void</a></tt>',
430 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
431 reserved words cannot conflict with variable names, because none of them
432 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]433
434<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]435 '<tt>%X</tt>' by 8:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]436
437<p>The easy way:</p>
438
439<div class="doc_code">
440<pre>
441%result = <a href="#i_mul">mul</a> i32 %X, 8
442</pre>
443</div>
444
445<p>After strength reduction:</p>
446
447<div class="doc_code">
448<pre>
449%result = <a href="#i_shl">shl</a> i32 %X, i8 3
450</pre>
451</div>
452
453<p>And the hard way:</p>
454
455<div class="doc_code">
456<pre>
Gabor Greifc0ea7672009-10-28 13:05:07 +0000[diff] [blame]457%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
458%1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]459%result = <a href="#i_add">add</a> i32 %1, %1
460</pre>
461</div>
462
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]463<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
464 lexical features of LLVM:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]465
466<ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]467 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]468 line.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]469
470 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]471 assigned to a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]472
473 <li>Unnamed temporaries are numbered sequentially</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]474</ol>
475
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]476<p>It also shows a convention that we follow in this document. When
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]477 demonstrating instructions, we will follow an instruction with a comment that
478 defines the type and name of value produced. Comments are shown in italic
479 text.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]480
481</div>
482
483<!-- *********************************************************************** -->
484<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
485<!-- *********************************************************************** -->
486
487<!-- ======================================================================= -->
488<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
489</div>
490
491<div class="doc_text">
492
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]493<p>LLVM programs are composed of "Module"s, each of which is a translation unit
494 of the input programs. Each module consists of functions, global variables,
495 and symbol table entries. Modules may be combined together with the LLVM
496 linker, which merges function (and global variable) definitions, resolves
497 forward declarations, and merges symbol table entries. Here is an example of
498 the "hello world" module:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]499
500<div class="doc_code">
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]501<pre>
502<i>; Declare the string constant as a global constant.</i>
503<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]504
505<i>; External declaration of the puts function</i>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]506<a href="#functionstructure">declare</a> i32 @puts(i8*) <i>; i32 (i8*)* </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]507
508<i>; Definition of main function</i>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]509define i32 @main() { <i>; i32()* </i>
510 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]511 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8*</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]512
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]513 <i>; Call puts function to write out the string to stdout.</i>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]514 <a href="#i_call">call</a> i32 @puts(i8* %cast210) <i>; i32</i>
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]515 <a href="#i_ret">ret</a> i32 0<br>}
516
517<i>; Named metadata</i>
518!1 = metadata !{i32 41}
519!foo = !{!1, null}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]520</pre>
521</div>
522
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]523<p>This example is made up of a <a href="#globalvars">global variable</a> named
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]524 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]525 a <a href="#functionstructure">function definition</a> for
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]526 "<tt>main</tt>" and <a href="#namedmetadatastructure">named metadata</a>
527 "<tt>foo"</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]528
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]529<p>In general, a module is made up of a list of global values, where both
530 functions and global variables are global values. Global values are
531 represented by a pointer to a memory location (in this case, a pointer to an
532 array of char, and a pointer to a function), and have one of the
533 following <a href="#linkage">linkage types</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]534
535</div>
536
537<!-- ======================================================================= -->
538<div class="doc_subsection">
539 <a name="linkage">Linkage Types</a>
540</div>
541
542<div class="doc_text">
543
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]544<p>All Global Variables and Functions have one of the following types of
545 linkage:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]546
547<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]548 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]549 <dd>Global values with private linkage are only directly accessible by objects
550 in the current module. In particular, linking code into a module with an
551 private global value may cause the private to be renamed as necessary to
552 avoid collisions. Because the symbol is private to the module, all
553 references can be updated. This doesn't show up in any symbol table in the
554 object file.</dd>
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]555
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]556 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]557 <dd>Similar to private, but the symbol is passed through the assembler and
Chris Lattner06c1ecc2009-08-24 04:32:16 +0000[diff] [blame]558 removed by the linker after evaluation. Note that (unlike private
559 symbols) linker_private symbols are subject to coalescing by the linker:
560 weak symbols get merged and redefinitions are rejected. However, unlike
561 normal strong symbols, they are removed by the linker from the final
562 linked image (executable or dynamic library).</dd>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]563
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]564 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]565 <dd>Similar to private, but the value shows as a local symbol
566 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
567 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]568
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]569 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]570 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]571 into the object file corresponding to the LLVM module. They exist to
572 allow inlining and other optimizations to take place given knowledge of
573 the definition of the global, which is known to be somewhere outside the
574 module. Globals with <tt>available_externally</tt> linkage are allowed to
575 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
576 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]577
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]578 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]579 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattnerd16c5512010-01-09 19:15:14 +0000[diff] [blame]580 the same name when linkage occurs. This can be used to implement
581 some forms of inline functions, templates, or other code which must be
582 generated in each translation unit that uses it, but where the body may
583 be overridden with a more definitive definition later. Unreferenced
584 <tt>linkonce</tt> globals are allowed to be discarded. Note that
585 <tt>linkonce</tt> linkage does not actually allow the optimizer to
586 inline the body of this function into callers because it doesn't know if
587 this definition of the function is the definitive definition within the
588 program or whether it will be overridden by a stronger definition.
589 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
590 linkage.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]591
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]592 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]593 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
594 <tt>linkonce</tt> linkage, except that unreferenced globals with
595 <tt>weak</tt> linkage may not be discarded. This is used for globals that
596 are declared "weak" in C source code.</dd>
597
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]598 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]599 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
600 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
601 global scope.
602 Symbols with "<tt>common</tt>" linkage are merged in the same way as
603 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattner83c4a712009-08-05 05:41:44 +0000[diff] [blame]604 <tt>common</tt> symbols may not have an explicit section,
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]605 must have a zero initializer, and may not be marked '<a
Chris Lattner83c4a712009-08-05 05:41:44 +0000[diff] [blame]606 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
607 have common linkage.</dd>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]608
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]609
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]610 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]611 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]612 pointer to array type. When two global variables with appending linkage
613 are linked together, the two global arrays are appended together. This is
614 the LLVM, typesafe, equivalent of having the system linker append together
615 "sections" with identical names when .o files are linked.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]616
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]617 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]618 <dd>The semantics of this linkage follow the ELF object file model: the symbol
619 is weak until linked, if not linked, the symbol becomes null instead of
620 being an undefined reference.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]621
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]622 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
623 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]624 <dd>Some languages allow differing globals to be merged, such as two functions
625 with different semantics. Other languages, such as <tt>C++</tt>, ensure
626 that only equivalent globals are ever merged (the "one definition rule" -
627 "ODR"). Such languages can use the <tt>linkonce_odr</tt>
628 and <tt>weak_odr</tt> linkage types to indicate that the global will only
629 be merged with equivalent globals. These linkage types are otherwise the
630 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]631
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]632 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]633 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]634 visible, meaning that it participates in linkage and can be used to
635 resolve external symbol references.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]636</dl>
637
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]638<p>The next two types of linkage are targeted for Microsoft Windows platform
639 only. They are designed to support importing (exporting) symbols from (to)
640 DLLs (Dynamic Link Libraries).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]641
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]642<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]643 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]644 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]645 or variable via a global pointer to a pointer that is set up by the DLL
646 exporting the symbol. On Microsoft Windows targets, the pointer name is
647 formed by combining <code>__imp_</code> and the function or variable
648 name.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]649
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]650 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]651 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]652 pointer to a pointer in a DLL, so that it can be referenced with the
653 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
654 name is formed by combining <code>__imp_</code> and the function or
655 variable name.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]656</dl>
657
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]658<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
659 another module defined a "<tt>.LC0</tt>" variable and was linked with this
660 one, one of the two would be renamed, preventing a collision. Since
661 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
662 declarations), they are accessible outside of the current module.</p>
663
664<p>It is illegal for a function <i>declaration</i> to have any linkage type
665 other than "externally visible", <tt>dllimport</tt>
666 or <tt>extern_weak</tt>.</p>
667
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]668<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]669 or <tt>weak_odr</tt> linkages.</p>
670
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]671</div>
672
673<!-- ======================================================================= -->
674<div class="doc_subsection">
675 <a name="callingconv">Calling Conventions</a>
676</div>
677
678<div class="doc_text">
679
680<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]681 and <a href="#i_invoke">invokes</a> can all have an optional calling
682 convention specified for the call. The calling convention of any pair of
683 dynamic caller/callee must match, or the behavior of the program is
684 undefined. The following calling conventions are supported by LLVM, and more
685 may be added in the future:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]686
687<dl>
688 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]689 <dd>This calling convention (the default if no other calling convention is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]690 specified) matches the target C calling conventions. This calling
691 convention supports varargs function calls and tolerates some mismatch in
692 the declared prototype and implemented declaration of the function (as
693 does normal C).</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]694
695 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]696 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]697 (e.g. by passing things in registers). This calling convention allows the
698 target to use whatever tricks it wants to produce fast code for the
699 target, without having to conform to an externally specified ABI
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]700 (Application Binary Interface).
701 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
Chris Lattnerac9a9392010-03-11 00:22:57 +0000[diff] [blame]702 when this or the GHC convention is used.</a> This calling convention
703 does not support varargs and requires the prototype of all callees to
704 exactly match the prototype of the function definition.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]705
706 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]707 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]708 as possible under the assumption that the call is not commonly executed.
709 As such, these calls often preserve all registers so that the call does
710 not break any live ranges in the caller side. This calling convention
711 does not support varargs and requires the prototype of all callees to
712 exactly match the prototype of the function definition.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]713
Chris Lattnerac9a9392010-03-11 00:22:57 +0000[diff] [blame]714 <dt><b>"<tt>cc <em>10</em></tt>" - GHC convention</b>:</dt>
715 <dd>This calling convention has been implemented specifically for use by the
716 <a href="http://www.haskell.org/ghc">Glasgow Haskell Compiler (GHC)</a>.
717 It passes everything in registers, going to extremes to achieve this by
718 disabling callee save registers. This calling convention should not be
719 used lightly but only for specific situations such as an alternative to
720 the <em>register pinning</em> performance technique often used when
721 implementing functional programming languages.At the moment only X86
722 supports this convention and it has the following limitations:
723 <ul>
724 <li>On <em>X86-32</em> only supports up to 4 bit type parameters. No
725 floating point types are supported.</li>
726 <li>On <em>X86-64</em> only supports up to 10 bit type parameters and
727 6 floating point parameters.</li>
728 </ul>
729 This calling convention supports
730 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> but
731 requires both the caller and callee are using it.
732 </dd>
733
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]734 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]735 <dd>Any calling convention may be specified by number, allowing
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]736 target-specific calling conventions to be used. Target specific calling
737 conventions start at 64.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]738</dl>
739
740<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]741 support Pascal conventions or any other well-known target-independent
742 convention.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]743
744</div>
745
746<!-- ======================================================================= -->
747<div class="doc_subsection">
748 <a name="visibility">Visibility Styles</a>
749</div>
750
751<div class="doc_text">
752
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]753<p>All Global Variables and Functions have one of the following visibility
754 styles:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]755
756<dl>
757 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]758 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]759 that the declaration is visible to other modules and, in shared libraries,
760 means that the declared entity may be overridden. On Darwin, default
761 visibility means that the declaration is visible to other modules. Default
762 visibility corresponds to "external linkage" in the language.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]763
764 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]765 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]766 object if they are in the same shared object. Usually, hidden visibility
767 indicates that the symbol will not be placed into the dynamic symbol
768 table, so no other module (executable or shared library) can reference it
769 directly.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]770
771 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]772 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]773 the dynamic symbol table, but that references within the defining module
774 will bind to the local symbol. That is, the symbol cannot be overridden by
775 another module.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]776</dl>
777
778</div>
779
780<!-- ======================================================================= -->
781<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]782 <a name="namedtypes">Named Types</a>
783</div>
784
785<div class="doc_text">
786
787<p>LLVM IR allows you to specify name aliases for certain types. This can make
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]788 it easier to read the IR and make the IR more condensed (particularly when
789 recursive types are involved). An example of a name specification is:</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]790
791<div class="doc_code">
792<pre>
793%mytype = type { %mytype*, i32 }
794</pre>
795</div>
796
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]797<p>You may give a name to any <a href="#typesystem">type</a> except
798 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
799 is expected with the syntax "%mytype".</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]800
801<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]802 and that you can therefore specify multiple names for the same type. This
803 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
804 uses structural typing, the name is not part of the type. When printing out
805 LLVM IR, the printer will pick <em>one name</em> to render all types of a
806 particular shape. This means that if you have code where two different
807 source types end up having the same LLVM type, that the dumper will sometimes
808 print the "wrong" or unexpected type. This is an important design point and
809 isn't going to change.</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]810
811</div>
812
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]813<!-- ======================================================================= -->
814<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]815 <a name="globalvars">Global Variables</a>
816</div>
817
818<div class="doc_text">
819
820<p>Global variables define regions of memory allocated at compilation time
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]821 instead of run-time. Global variables may optionally be initialized, may
822 have an explicit section to be placed in, and may have an optional explicit
823 alignment specified. A variable may be defined as "thread_local", which
824 means that it will not be shared by threads (each thread will have a
825 separated copy of the variable). A variable may be defined as a global
826 "constant," which indicates that the contents of the variable
827 will <b>never</b> be modified (enabling better optimization, allowing the
828 global data to be placed in the read-only section of an executable, etc).
829 Note that variables that need runtime initialization cannot be marked
830 "constant" as there is a store to the variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]831
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]832<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
833 constant, even if the final definition of the global is not. This capability
834 can be used to enable slightly better optimization of the program, but
835 requires the language definition to guarantee that optimizations based on the
836 'constantness' are valid for the translation units that do not include the
837 definition.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]838
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]839<p>As SSA values, global variables define pointer values that are in scope
840 (i.e. they dominate) all basic blocks in the program. Global variables
841 always define a pointer to their "content" type because they describe a
842 region of memory, and all memory objects in LLVM are accessed through
843 pointers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]844
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]845<p>A global variable may be declared to reside in a target-specific numbered
846 address space. For targets that support them, address spaces may affect how
847 optimizations are performed and/or what target instructions are used to
848 access the variable. The default address space is zero. The address space
849 qualifier must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]850
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]851<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]852 supports it, it will emit globals to the section specified.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]853
Chris Lattner72413b22010-04-28 00:13:42 +0000[diff] [blame]854<p>An explicit alignment may be specified for a global, which must be a power
855 of 2. If not present, or if the alignment is set to zero, the alignment of
856 the global is set by the target to whatever it feels convenient. If an
857 explicit alignment is specified, the global is forced to have exactly that
Chris Lattner50d07d02010-04-28 00:31:12 +0000[diff] [blame]858 alignment. Targets and optimizers are not allowed to over-align the global
859 if the global has an assigned section. In this case, the extra alignment
860 could be observable: for example, code could assume that the globals are
861 densely packed in their section and try to iterate over them as an array,
862 alignment padding would break this iteration.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]863
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]864<p>For example, the following defines a global in a numbered address space with
865 an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]866
867<div class="doc_code">
868<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]869@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]870</pre>
871</div>
872
873</div>
874
875
876<!-- ======================================================================= -->
877<div class="doc_subsection">
878 <a name="functionstructure">Functions</a>
879</div>
880
881<div class="doc_text">
882
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]883<p>LLVM function definitions consist of the "<tt>define</tt>" keyword, an
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]884 optional <a href="#linkage">linkage type</a>, an optional
885 <a href="#visibility">visibility style</a>, an optional
886 <a href="#callingconv">calling convention</a>, a return type, an optional
887 <a href="#paramattrs">parameter attribute</a> for the return type, a function
888 name, a (possibly empty) argument list (each with optional
889 <a href="#paramattrs">parameter attributes</a>), optional
890 <a href="#fnattrs">function attributes</a>, an optional section, an optional
891 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
892 curly brace, a list of basic blocks, and a closing curly brace.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]893
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]894<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
895 optional <a href="#linkage">linkage type</a>, an optional
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]896 <a href="#visibility">visibility style</a>, an optional
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]897 <a href="#callingconv">calling convention</a>, a return type, an optional
898 <a href="#paramattrs">parameter attribute</a> for the return type, a function
899 name, a possibly empty list of arguments, an optional alignment, and an
900 optional <a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]901
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]902<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]903 (Control Flow Graph) for the function. Each basic block may optionally start
904 with a label (giving the basic block a symbol table entry), contains a list
905 of instructions, and ends with a <a href="#terminators">terminator</a>
906 instruction (such as a branch or function return).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]907
908<p>The first basic block in a function is special in two ways: it is immediately
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]909 executed on entrance to the function, and it is not allowed to have
910 predecessor basic blocks (i.e. there can not be any branches to the entry
911 block of a function). Because the block can have no predecessors, it also
912 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]913
914<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]915 supports it, it will emit functions to the section specified.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]916
917<p>An explicit alignment may be specified for a function. If not present, or if
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]918 the alignment is set to zero, the alignment of the function is set by the
919 target to whatever it feels convenient. If an explicit alignment is
920 specified, the function is forced to have at least that much alignment. All
921 alignments must be a power of 2.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]922
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]923<h5>Syntax:</h5>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]924<div class="doc_code">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]925<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]926define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]927 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
928 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
929 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
930 [<a href="#gc">gc</a>] { ... }
931</pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]932</div>
933
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]934</div>
935
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]936<!-- ======================================================================= -->
937<div class="doc_subsection">
938 <a name="aliasstructure">Aliases</a>
939</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]940
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]941<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]942
943<p>Aliases act as "second name" for the aliasee value (which can be either
944 function, global variable, another alias or bitcast of global value). Aliases
945 may have an optional <a href="#linkage">linkage type</a>, and an
946 optional <a href="#visibility">visibility style</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]947
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]948<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]949<div class="doc_code">
950<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]951@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]952</pre>
953</div>
954
955</div>
956
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]957<!-- ======================================================================= -->
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]958<div class="doc_subsection">
959 <a name="namedmetadatastructure">Named Metadata</a>
960</div>
961
962<div class="doc_text">
963
Chris Lattnerd0d96292010-01-15 21:50:19 +0000[diff] [blame]964<p>Named metadata is a collection of metadata. <a href="#metadata">Metadata
965 nodes</a> (but not metadata strings) and null are the only valid operands for
966 a named metadata.</p>
Devang Patela4bb6792010-01-11 19:35:55 +0000[diff] [blame]967
968<h5>Syntax:</h5>
969<div class="doc_code">
970<pre>
971!1 = metadata !{metadata !"one"}
972!name = !{null, !1}
973</pre>
974</div>
975
976</div>
977
978<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]979<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]980
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]981<div class="doc_text">
982
983<p>The return type and each parameter of a function type may have a set of
984 <i>parameter attributes</i> associated with them. Parameter attributes are
985 used to communicate additional information about the result or parameters of
986 a function. Parameter attributes are considered to be part of the function,
987 not of the function type, so functions with different parameter attributes
988 can have the same function type.</p>
989
990<p>Parameter attributes are simple keywords that follow the type specified. If
991 multiple parameter attributes are needed, they are space separated. For
992 example:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]993
994<div class="doc_code">
995<pre>
Nick Lewycky3022a742009-02-15 23:06:14 +0000[diff] [blame]996declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]997declare i32 @atoi(i8 zeroext)
998declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]999</pre>
1000</div>
1001
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1002<p>Note that any attributes for the function result (<tt>nounwind</tt>,
1003 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1004
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1005<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]1006
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1007<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1008 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1009 <dd>This indicates to the code generator that the parameter or return value
1010 should be zero-extended to a 32-bit value by the caller (for a parameter)
1011 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]1012
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1013 <dt><tt><b>signext</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1014 <dd>This indicates to the code generator that the parameter or return value
1015 should be sign-extended to a 32-bit value by the caller (for a parameter)
1016 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]1017
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1018 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1019 <dd>This indicates that this parameter or return value should be treated in a
1020 special target-dependent fashion during while emitting code for a function
1021 call or return (usually, by putting it in a register as opposed to memory,
1022 though some targets use it to distinguish between two different kinds of
1023 registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]1024
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1025 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1026 <dd>This indicates that the pointer parameter should really be passed by value
1027 to the function. The attribute implies that a hidden copy of the pointee
1028 is made between the caller and the callee, so the callee is unable to
1029 modify the value in the callee. This attribute is only valid on LLVM
1030 pointer arguments. It is generally used to pass structs and arrays by
1031 value, but is also valid on pointers to scalars. The copy is considered
1032 to belong to the caller not the callee (for example,
1033 <tt><a href="#readonly">readonly</a></tt> functions should not write to
1034 <tt>byval</tt> parameters). This is not a valid attribute for return
1035 values. The byval attribute also supports specifying an alignment with
1036 the align attribute. This has a target-specific effect on the code
1037 generator that usually indicates a desired alignment for the synthesized
1038 stack slot.</dd>
1039
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1040 <dt><tt><b>sret</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1041 <dd>This indicates that the pointer parameter specifies the address of a
1042 structure that is the return value of the function in the source program.
1043 This pointer must be guaranteed by the caller to be valid: loads and
1044 stores to the structure may be assumed by the callee to not to trap. This
1045 may only be applied to the first parameter. This is not a valid attribute
1046 for return values. </dd>
1047
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1048 <dt><tt><b>noalias</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1049 <dd>This indicates that the pointer does not alias any global or any other
1050 parameter. The caller is responsible for ensuring that this is the
1051 case. On a function return value, <tt>noalias</tt> additionally indicates
1052 that the pointer does not alias any other pointers visible to the
1053 caller. For further details, please see the discussion of the NoAlias
1054 response in
1055 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
1056 analysis</a>.</dd>
1057
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1058 <dt><tt><b>nocapture</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1059 <dd>This indicates that the callee does not make any copies of the pointer
1060 that outlive the callee itself. This is not a valid attribute for return
1061 values.</dd>
1062
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1063 <dt><tt><b>nest</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1064 <dd>This indicates that the pointer parameter can be excised using the
1065 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1066 attribute for return values.</dd>
1067</dl>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1068
1069</div>
1070
1071<!-- ======================================================================= -->
1072<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1073 <a name="gc">Garbage Collector Names</a>
1074</div>
1075
1076<div class="doc_text">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1077
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1078<p>Each function may specify a garbage collector name, which is simply a
1079 string:</p>
1080
1081<div class="doc_code">
1082<pre>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1083define void @f() gc "name" { ... }
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1084</pre>
1085</div>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1086
1087<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1088 collector which will cause the compiler to alter its output in order to
1089 support the named garbage collection algorithm.</p>
1090
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1091</div>
1092
1093<!-- ======================================================================= -->
1094<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1095 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1096</div>
1097
1098<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1099
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1100<p>Function attributes are set to communicate additional information about a
1101 function. Function attributes are considered to be part of the function, not
1102 of the function type, so functions with different parameter attributes can
1103 have the same function type.</p>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1104
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1105<p>Function attributes are simple keywords that follow the type specified. If
1106 multiple attributes are needed, they are space separated. For example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1107
1108<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1109<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1110define void @f() noinline { ... }
1111define void @f() alwaysinline { ... }
1112define void @f() alwaysinline optsize { ... }
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1113define void @f() optsize { ... }
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1114</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1115</div>
1116
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1117<dl>
Charles Davisfaa8f752010-02-12 00:31:15 +0000[diff] [blame]1118 <dt><tt><b>alignstack(&lt;<em>n</em>&gt;)</b></tt></dt>
1119 <dd>This attribute indicates that, when emitting the prologue and epilogue,
1120 the backend should forcibly align the stack pointer. Specify the
1121 desired alignment, which must be a power of two, in parentheses.
1122
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1123 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1124 <dd>This attribute indicates that the inliner should attempt to inline this
1125 function into callers whenever possible, ignoring any active inlining size
1126 threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1127
Jakob Stoklund Olesen77180732010-02-06 01:16:28 +0000[diff] [blame]1128 <dt><tt><b>inlinehint</b></tt></dt>
1129 <dd>This attribute indicates that the source code contained a hint that inlining
1130 this function is desirable (such as the "inline" keyword in C/C++). It
1131 is just a hint; it imposes no requirements on the inliner.</dd>
1132
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1133 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1134 <dd>This attribute indicates that the inliner should never inline this
1135 function in any situation. This attribute may not be used together with
1136 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1137
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1138 <dt><tt><b>optsize</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1139 <dd>This attribute suggests that optimization passes and code generator passes
1140 make choices that keep the code size of this function low, and otherwise
1141 do optimizations specifically to reduce code size.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1142
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1143 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1144 <dd>This function attribute indicates that the function never returns
1145 normally. This produces undefined behavior at runtime if the function
1146 ever does dynamically return.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1147
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1148 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1149 <dd>This function attribute indicates that the function never returns with an
1150 unwind or exceptional control flow. If the function does unwind, its
1151 runtime behavior is undefined.</dd>
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1152
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1153 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1154 <dd>This attribute indicates that the function computes its result (or decides
1155 to unwind an exception) based strictly on its arguments, without
1156 dereferencing any pointer arguments or otherwise accessing any mutable
1157 state (e.g. memory, control registers, etc) visible to caller functions.
1158 It does not write through any pointer arguments
1159 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1160 changes any state visible to callers. This means that it cannot unwind
1161 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1162 could use the <tt>unwind</tt> instruction.</dd>
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1163
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1164 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1165 <dd>This attribute indicates that the function does not write through any
1166 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1167 arguments) or otherwise modify any state (e.g. memory, control registers,
1168 etc) visible to caller functions. It may dereference pointer arguments
1169 and read state that may be set in the caller. A readonly function always
1170 returns the same value (or unwinds an exception identically) when called
1171 with the same set of arguments and global state. It cannot unwind an
1172 exception by calling the <tt>C++</tt> exception throwing methods, but may
1173 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovedd7d112009-07-17 18:07:26 +0000[diff] [blame]1174
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1175 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1176 <dd>This attribute indicates that the function should emit a stack smashing
1177 protector. It is in the form of a "canary"&mdash;a random value placed on
1178 the stack before the local variables that's checked upon return from the
1179 function to see if it has been overwritten. A heuristic is used to
1180 determine if a function needs stack protectors or not.<br>
1181<br>
1182 If a function that has an <tt>ssp</tt> attribute is inlined into a
1183 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1184 function will have an <tt>ssp</tt> attribute.</dd>
1185
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1186 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1187 <dd>This attribute indicates that the function should <em>always</em> emit a
1188 stack smashing protector. This overrides
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1189 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1190<br>
1191 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1192 function that doesn't have an <tt>sspreq</tt> attribute or which has
1193 an <tt>ssp</tt> attribute, then the resulting function will have
1194 an <tt>sspreq</tt> attribute.</dd>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1195
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1196 <dt><tt><b>noredzone</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1197 <dd>This attribute indicates that the code generator should not use a red
1198 zone, even if the target-specific ABI normally permits it.</dd>
1199
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1200 <dt><tt><b>noimplicitfloat</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1201 <dd>This attributes disables implicit floating point instructions.</dd>
1202
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1203 <dt><tt><b>naked</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1204 <dd>This attribute disables prologue / epilogue emission for the function.
1205 This can have very system-specific consequences.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1206</dl>
1207
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1208</div>
1209
1210<!-- ======================================================================= -->
1211<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1212 <a name="moduleasm">Module-Level Inline Assembly</a>
1213</div>
1214
1215<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1216
1217<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1218 the GCC "file scope inline asm" blocks. These blocks are internally
1219 concatenated by LLVM and treated as a single unit, but may be separated in
1220 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1221
1222<div class="doc_code">
1223<pre>
1224module asm "inline asm code goes here"
1225module asm "more can go here"
1226</pre>
1227</div>
1228
1229<p>The strings can contain any character by escaping non-printable characters.
1230 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1231 for the number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1232
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1233<p>The inline asm code is simply printed to the machine code .s file when
1234 assembly code is generated.</p>
1235
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1236</div>
1237
1238<!-- ======================================================================= -->
1239<div class="doc_subsection">
1240 <a name="datalayout">Data Layout</a>
1241</div>
1242
1243<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1244
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1245<p>A module may specify a target specific data layout string that specifies how
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1246 data is to be laid out in memory. The syntax for the data layout is
1247 simply:</p>
1248
1249<div class="doc_code">
1250<pre>
1251target datalayout = "<i>layout specification</i>"
1252</pre>
1253</div>
1254
1255<p>The <i>layout specification</i> consists of a list of specifications
1256 separated by the minus sign character ('-'). Each specification starts with
1257 a letter and may include other information after the letter to define some
1258 aspect of the data layout. The specifications accepted are as follows:</p>
1259
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1260<dl>
1261 <dt><tt>E</tt></dt>
1262 <dd>Specifies that the target lays out data in big-endian form. That is, the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1263 bits with the most significance have the lowest address location.</dd>
1264
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1265 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1266 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1267 the bits with the least significance have the lowest address
1268 location.</dd>
1269
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1270 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1271 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1272 <i>preferred</i> alignments. All sizes are in bits. Specifying
1273 the <i>pref</i> alignment is optional. If omitted, the
1274 preceding <tt>:</tt> should be omitted too.</dd>
1275
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1276 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1277 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1278 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1279
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1280 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1281 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1282 <i>size</i>.</dd>
1283
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1284 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1285 <dd>This specifies the alignment for a floating point type of a given bit
Dale Johannesen7dabc4c2010-05-28 18:54:47 +0000[diff] [blame]1286 <i>size</i>. Only values of <i>size</i> that are supported by the target
1287 will work. 32 (float) and 64 (double) are supported on all targets;
1288 80 or 128 (different flavors of long double) are also supported on some
1289 targets.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1290
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1291 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1292 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1293 <i>size</i>.</dd>
1294
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1295 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1296 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1297 <i>size</i>.</dd>
Chris Lattnerece41802009-11-07 09:35:34 +0000[diff] [blame]1298
1299 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1300 <dd>This specifies a set of native integer widths for the target CPU
1301 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1302 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1303 this set are considered to support most general arithmetic
Chris Lattnerece41802009-11-07 09:35:34 +0000[diff] [blame]1304 operations efficiently.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1305</dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1306
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1307<p>When constructing the data layout for a given target, LLVM starts with a
Dan Gohmanfde3cd72010-04-28 00:36:01 +0000[diff] [blame]1308 default set of specifications which are then (possibly) overridden by the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1309 specifications in the <tt>datalayout</tt> keyword. The default specifications
1310 are given in this list:</p>
1311
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1312<ul>
1313 <li><tt>E</tt> - big endian</li>
Dan Gohmane78194f2010-02-23 02:44:03 +0000[diff] [blame]1314 <li><tt>p:64:64:64</tt> - 64-bit pointers with 64-bit alignment</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1315 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1316 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1317 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1318 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1319 <li><tt>i64:32:64</tt> - i64 has ABI alignment of 32-bits but preferred
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1320 alignment of 64-bits</li>
1321 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1322 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1323 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1324 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1325 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1326 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1327</ul>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1328
1329<p>When LLVM is determining the alignment for a given type, it uses the
1330 following rules:</p>
1331
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1332<ol>
1333 <li>If the type sought is an exact match for one of the specifications, that
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1334 specification is used.</li>
1335
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1336 <li>If no match is found, and the type sought is an integer type, then the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1337 smallest integer type that is larger than the bitwidth of the sought type
1338 is used. If none of the specifications are larger than the bitwidth then
1339 the the largest integer type is used. For example, given the default
1340 specifications above, the i7 type will use the alignment of i8 (next
1341 largest) while both i65 and i256 will use the alignment of i64 (largest
1342 specified).</li>
1343
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1344 <li>If no match is found, and the type sought is a vector type, then the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1345 largest vector type that is smaller than the sought vector type will be
1346 used as a fall back. This happens because &lt;128 x double&gt; can be
1347 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1348</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1349
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1350</div>
1351
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1352<!-- ======================================================================= -->
1353<div class="doc_subsection">
1354 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1355</div>
1356
1357<div class="doc_text">
1358
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1359<p>Any memory access must be done through a pointer value associated
Andreas Bolka23bece42009-07-27 20:37:10 +0000[diff] [blame]1360with an address range of the memory access, otherwise the behavior
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1361is undefined. Pointer values are associated with address ranges
1362according to the following rules:</p>
1363
1364<ul>
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1365 <li>A pointer value formed from a
1366 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1367 is associated with the addresses associated with the first operand
1368 of the <tt>getelementptr</tt>.</li>
1369 <li>An address of a global variable is associated with the address
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1370 range of the variable's storage.</li>
1371 <li>The result value of an allocation instruction is associated with
1372 the address range of the allocated storage.</li>
1373 <li>A null pointer in the default address-space is associated with
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1374 no address.</li>
1375 <li>A pointer value formed by an
1376 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1377 address ranges of all pointer values that contribute (directly or
1378 indirectly) to the computation of the pointer's value.</li>
1379 <li>The result value of a
1380 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1381 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1382 <li>An integer constant other than zero or a pointer value returned
1383 from a function not defined within LLVM may be associated with address
1384 ranges allocated through mechanisms other than those provided by
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1385 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1386 allocated by mechanisms provided by LLVM.</li>
1387 </ul>
1388
1389<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1390<tt><a href="#i_load">load</a></tt> merely indicates the size and
1391alignment of the memory from which to load, as well as the
Dan Gohmand72730e2010-06-17 19:23:50 +0000[diff] [blame]1392interpretation of the value. The first operand type of a
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1393<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1394and alignment of the store.</p>
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1395
1396<p>Consequently, type-based alias analysis, aka TBAA, aka
1397<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1398LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1399additional information which specialized optimization passes may use
1400to implement type-based alias analysis.</p>
1401
1402</div>
1403
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]1404<!-- ======================================================================= -->
1405<div class="doc_subsection">
1406 <a name="volatile">Volatile Memory Accesses</a>
1407</div>
1408
1409<div class="doc_text">
1410
1411<p>Certain memory accesses, such as <a href="#i_load"><tt>load</tt></a>s, <a
1412href="#i_store"><tt>store</tt></a>s, and <a
1413href="#int_memcpy"><tt>llvm.memcpy</tt></a>s may be marked <tt>volatile</tt>.
1414The optimizers must not change the number of volatile operations or change their
1415order of execution relative to other volatile operations. The optimizers
1416<i>may</i> change the order of volatile operations relative to non-volatile
1417operations. This is not Java's "volatile" and has no cross-thread
1418synchronization behavior.</p>
1419
1420</div>
1421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1422<!-- *********************************************************************** -->
1423<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1424<!-- *********************************************************************** -->
1425
1426<div class="doc_text">
1427
1428<p>The LLVM type system is one of the most important features of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1429 intermediate representation. Being typed enables a number of optimizations
1430 to be performed on the intermediate representation directly, without having
1431 to do extra analyses on the side before the transformation. A strong type
1432 system makes it easier to read the generated code and enables novel analyses
1433 and transformations that are not feasible to perform on normal three address
1434 code representations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1435
1436</div>
1437
1438<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1439<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1440Classifications</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1441
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1442<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1443
1444<p>The types fall into a few useful classifications:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1445
1446<table border="1" cellspacing="0" cellpadding="4">
1447 <tbody>
1448 <tr><th>Classification</th><th>Types</th></tr>
1449 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1450 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1451 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1452 </tr>
1453 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1454 <td><a href="#t_floating">floating point</a></td>
1455 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1456 </tr>
1457 <tr>
1458 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1459 <td><a href="#t_integer">integer</a>,
1460 <a href="#t_floating">floating point</a>,
1461 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1462 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1463 <a href="#t_struct">structure</a>,
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1464 <a href="#t_union">union</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1465 <a href="#t_array">array</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1466 <a href="#t_label">label</a>,
1467 <a href="#t_metadata">metadata</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1468 </td>
1469 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1470 <tr>
1471 <td><a href="#t_primitive">primitive</a></td>
1472 <td><a href="#t_label">label</a>,
1473 <a href="#t_void">void</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1474 <a href="#t_floating">floating point</a>,
1475 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1476 </tr>
1477 <tr>
1478 <td><a href="#t_derived">derived</a></td>
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1479 <td><a href="#t_array">array</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1480 <a href="#t_function">function</a>,
1481 <a href="#t_pointer">pointer</a>,
1482 <a href="#t_struct">structure</a>,
1483 <a href="#t_pstruct">packed structure</a>,
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1484 <a href="#t_union">union</a>,
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1485 <a href="#t_vector">vector</a>,
1486 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1487 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1488 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1489 </tbody>
1490</table>
1491
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1492<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1493 important. Values of these types are the only ones which can be produced by
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1494 instructions.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1495
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1496</div>
1497
1498<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1499<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1500
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1501<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1502
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1503<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1504 system.</p>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1505
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1506</div>
1507
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1508<!-- _______________________________________________________________________ -->
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1509<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1510
1511<div class="doc_text">
1512
1513<h5>Overview:</h5>
1514<p>The integer type is a very simple type that simply specifies an arbitrary
1515 bit width for the integer type desired. Any bit width from 1 bit to
1516 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1517
1518<h5>Syntax:</h5>
1519<pre>
1520 iN
1521</pre>
1522
1523<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1524 value.</p>
1525
1526<h5>Examples:</h5>
1527<table class="layout">
1528 <tr class="layout">
1529 <td class="left"><tt>i1</tt></td>
1530 <td class="left">a single-bit integer.</td>
1531 </tr>
1532 <tr class="layout">
1533 <td class="left"><tt>i32</tt></td>
1534 <td class="left">a 32-bit integer.</td>
1535 </tr>
1536 <tr class="layout">
1537 <td class="left"><tt>i1942652</tt></td>
1538 <td class="left">a really big integer of over 1 million bits.</td>
1539 </tr>
1540</table>
1541
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1542</div>
1543
1544<!-- _______________________________________________________________________ -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1545<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1546
1547<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1548
1549<table>
1550 <tbody>
1551 <tr><th>Type</th><th>Description</th></tr>
1552 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1553 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1554 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1555 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1556 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1557 </tbody>
1558</table>
1559
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1560</div>
1561
1562<!-- _______________________________________________________________________ -->
1563<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1564
1565<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1566
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1567<h5>Overview:</h5>
1568<p>The void type does not represent any value and has no size.</p>
1569
1570<h5>Syntax:</h5>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1571<pre>
1572 void
1573</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1574
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1575</div>
1576
1577<!-- _______________________________________________________________________ -->
1578<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1579
1580<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1581
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1582<h5>Overview:</h5>
1583<p>The label type represents code labels.</p>
1584
1585<h5>Syntax:</h5>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1586<pre>
1587 label
1588</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1589
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1590</div>
1591
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1592<!-- _______________________________________________________________________ -->
1593<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1594
1595<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1596
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1597<h5>Overview:</h5>
Nick Lewyckyeb021882009-09-27 23:27:42 +0000[diff] [blame]1598<p>The metadata type represents embedded metadata. No derived types may be
1599 created from metadata except for <a href="#t_function">function</a>
1600 arguments.
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1601
1602<h5>Syntax:</h5>
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1603<pre>
1604 metadata
1605</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1606
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1607</div>
1608
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1609
1610<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1611<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1612
1613<div class="doc_text">
1614
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1615<p>The real power in LLVM comes from the derived types in the system. This is
1616 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1617 useful types. Each of these types contain one or more element types which
1618 may be a primitive type, or another derived type. For example, it is
1619 possible to have a two dimensional array, using an array as the element type
1620 of another array.</p>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1621
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1622
1623</div>
1624
1625<!-- _______________________________________________________________________ -->
1626<div class="doc_subsubsection"> <a name="t_aggregate">Aggregate Types</a> </div>
1627
1628<div class="doc_text">
1629
1630<p>Aggregate Types are a subset of derived types that can contain multiple
1631 member types. <a href="#t_array">Arrays</a>,
1632 <a href="#t_struct">structs</a>, <a href="#t_vector">vectors</a> and
1633 <a href="#t_union">unions</a> are aggregate types.</p>
1634
1635</div>
1636
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1637</div>
1638
1639<!-- _______________________________________________________________________ -->
1640<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1641
1642<div class="doc_text">
1643
1644<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1645<p>The array type is a very simple derived type that arranges elements
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1646 sequentially in memory. The array type requires a size (number of elements)
1647 and an underlying data type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1648
1649<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1650<pre>
1651 [&lt;# elements&gt; x &lt;elementtype&gt;]
1652</pre>
1653
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1654<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1655 be any type with a size.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1656
1657<h5>Examples:</h5>
1658<table class="layout">
1659 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1660 <td class="left"><tt>[40 x i32]</tt></td>
1661 <td class="left">Array of 40 32-bit integer values.</td>
1662 </tr>
1663 <tr class="layout">
1664 <td class="left"><tt>[41 x i32]</tt></td>
1665 <td class="left">Array of 41 32-bit integer values.</td>
1666 </tr>
1667 <tr class="layout">
1668 <td class="left"><tt>[4 x i8]</tt></td>
1669 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1670 </tr>
1671</table>
1672<p>Here are some examples of multidimensional arrays:</p>
1673<table class="layout">
1674 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1675 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1676 <td class="left">3x4 array of 32-bit integer values.</td>
1677 </tr>
1678 <tr class="layout">
1679 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1680 <td class="left">12x10 array of single precision floating point values.</td>
1681 </tr>
1682 <tr class="layout">
1683 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1684 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1685 </tr>
1686</table>
1687
Dan Gohman87ec5ad2009-11-09 19:01:53 +0000[diff] [blame]1688<p>There is no restriction on indexing beyond the end of the array implied by
1689 a static type (though there are restrictions on indexing beyond the bounds
1690 of an allocated object in some cases). This means that single-dimension
1691 'variable sized array' addressing can be implemented in LLVM with a zero
1692 length array type. An implementation of 'pascal style arrays' in LLVM could
1693 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1694
1695</div>
1696
1697<!-- _______________________________________________________________________ -->
1698<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1699
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1700<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1701
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1702<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1703<p>The function type can be thought of as a function signature. It consists of
1704 a return type and a list of formal parameter types. The return type of a
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1705 function type is a scalar type, a void type, a struct type, or a union
1706 type. If the return type is a struct type then all struct elements must be
1707 of first class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1708
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1709<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1710<pre>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1711 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1712</pre>
1713
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1714<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1715 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1716 which indicates that the function takes a variable number of arguments.
1717 Variable argument functions can access their arguments with
1718 the <a href="#int_varargs">variable argument handling intrinsic</a>
Chris Lattner553fb1e2010-03-02 06:36:51 +0000[diff] [blame]1719 functions. '<tt>&lt;returntype&gt;</tt>' is any type except
Nick Lewyckyeb021882009-09-27 23:27:42 +0000[diff] [blame]1720 <a href="#t_label">label</a>.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1722<h5>Examples:</h5>
1723<table class="layout">
1724 <tr class="layout">
1725 <td class="left"><tt>i32 (i32)</tt></td>
1726 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1727 </td>
1728 </tr><tr class="layout">
Chris Lattner553fb1e2010-03-02 06:36:51 +0000[diff] [blame]1729 <td class="left"><tt>float&nbsp;(i16,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1730 </tt></td>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1731 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
Chris Lattner553fb1e2010-03-02 06:36:51 +0000[diff] [blame]1732 an <tt>i16</tt> and a <a href="#t_pointer">pointer</a> to <tt>i32</tt>,
1733 returning <tt>float</tt>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1734 </td>
1735 </tr><tr class="layout">
1736 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1737 <td class="left">A vararg function that takes at least one
1738 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1739 which returns an integer. This is the signature for <tt>printf</tt> in
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1740 LLVM.
1741 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1742 </tr><tr class="layout">
1743 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1744 <td class="left">A function taking an <tt>i32</tt>, returning a
1745 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1746 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1747 </tr>
1748</table>
1749
1750</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1751
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1752<!-- _______________________________________________________________________ -->
1753<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1754
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1755<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1756
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1757<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1758<p>The structure type is used to represent a collection of data members together
1759 in memory. The packing of the field types is defined to match the ABI of the
1760 underlying processor. The elements of a structure may be any type that has a
1761 size.</p>
1762
Jeffrey Yasskine05d7732010-01-11 19:19:26 +0000[diff] [blame]1763<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1764 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1765 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1766 Structures in registers are accessed using the
1767 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1768 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1769<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1770<pre>
1771 { &lt;type list&gt; }
1772</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1773
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1774<h5>Examples:</h5>
1775<table class="layout">
1776 <tr class="layout">
1777 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1778 <td class="left">A triple of three <tt>i32</tt> values</td>
1779 </tr><tr class="layout">
1780 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1781 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1782 second element is a <a href="#t_pointer">pointer</a> to a
1783 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1784 an <tt>i32</tt>.</td>
1785 </tr>
1786</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1787
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1788</div>
1789
1790<!-- _______________________________________________________________________ -->
1791<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1792</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1793
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1794<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1795
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1796<h5>Overview:</h5>
1797<p>The packed structure type is used to represent a collection of data members
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1798 together in memory. There is no padding between fields. Further, the
1799 alignment of a packed structure is 1 byte. The elements of a packed
1800 structure may be any type that has a size.</p>
1801
1802<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1803 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1804 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1805
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1806<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1807<pre>
1808 &lt; { &lt;type list&gt; } &gt;
1809</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1810
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1811<h5>Examples:</h5>
1812<table class="layout">
1813 <tr class="layout">
1814 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1815 <td class="left">A triple of three <tt>i32</tt> values</td>
1816 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1817 <td class="left">
1818<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1819 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1820 second element is a <a href="#t_pointer">pointer</a> to a
1821 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1822 an <tt>i32</tt>.</td>
1823 </tr>
1824</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1825
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1826</div>
1827
1828<!-- _______________________________________________________________________ -->
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1829<div class="doc_subsubsection"> <a name="t_union">Union Type</a> </div>
1830
1831<div class="doc_text">
1832
1833<h5>Overview:</h5>
1834<p>A union type describes an object with size and alignment suitable for
1835 an object of any one of a given set of types (also known as an "untagged"
1836 union). It is similar in concept and usage to a
1837 <a href="#t_struct">struct</a>, except that all members of the union
1838 have an offset of zero. The elements of a union may be any type that has a
1839 size. Unions must have at least one member - empty unions are not allowed.
1840 </p>
1841
1842<p>The size of the union as a whole will be the size of its largest member,
1843 and the alignment requirements of the union as a whole will be the largest
1844 alignment requirement of any member.</p>
1845
Dan Gohmanef8400c2010-02-25 16:51:31 +0000[diff] [blame]1846<p>Union members are accessed using '<tt><a href="#i_load">load</a></tt> and
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]1847 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1848 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1849 Since all members are at offset zero, the getelementptr instruction does
1850 not affect the address, only the type of the resulting pointer.</p>
1851
1852<h5>Syntax:</h5>
1853<pre>
1854 union { &lt;type list&gt; }
1855</pre>
1856
1857<h5>Examples:</h5>
1858<table class="layout">
1859 <tr class="layout">
1860 <td class="left"><tt>union { i32, i32*, float }</tt></td>
1861 <td class="left">A union of three types: an <tt>i32</tt>, a pointer to
1862 an <tt>i32</tt>, and a <tt>float</tt>.</td>
1863 </tr><tr class="layout">
1864 <td class="left">
1865 <tt>union {&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1866 <td class="left">A union, where the first element is a <tt>float</tt> and the
1867 second element is a <a href="#t_pointer">pointer</a> to a
1868 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1869 an <tt>i32</tt>.</td>
1870 </tr>
1871</table>
1872
1873</div>
1874
1875<!-- _______________________________________________________________________ -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1876<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1877
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1878<div class="doc_text">
1879
1880<h5>Overview:</h5>
Dan Gohmanb2f72c82010-02-25 16:50:07 +0000[diff] [blame]1881<p>The pointer type is used to specify memory locations.
1882 Pointers are commonly used to reference objects in memory.</p>
1883
1884<p>Pointer types may have an optional address space attribute defining the
1885 numbered address space where the pointed-to object resides. The default
1886 address space is number zero. The semantics of non-zero address
1887 spaces are target-specific.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1888
1889<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1890 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1891
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1892<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1893<pre>
1894 &lt;type&gt; *
1895</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1896
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1897<h5>Examples:</h5>
1898<table class="layout">
1899 <tr class="layout">
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]1900 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1901 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1902 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1903 </tr>
1904 <tr class="layout">
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]1905 <td class="left"><tt>i32 (i32*) *</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1906 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1907 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1908 <tt>i32</tt>.</td>
1909 </tr>
1910 <tr class="layout">
1911 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1912 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1913 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1914 </tr>
1915</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1916
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1917</div>
1918
1919<!-- _______________________________________________________________________ -->
1920<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1921
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1922<div class="doc_text">
1923
1924<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1925<p>A vector type is a simple derived type that represents a vector of elements.
1926 Vector types are used when multiple primitive data are operated in parallel
1927 using a single instruction (SIMD). A vector type requires a size (number of
Duncan Sands87598b02009-11-27 13:38:03 +0000[diff] [blame]1928 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1929 <a href="#t_firstclass">first class</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1930
1931<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1932<pre>
1933 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1934</pre>
1935
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1936<p>The number of elements is a constant integer value; elementtype may be any
1937 integer or floating point type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1938
1939<h5>Examples:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1940<table class="layout">
1941 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1942 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1943 <td class="left">Vector of 4 32-bit integer values.</td>
1944 </tr>
1945 <tr class="layout">
1946 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1947 <td class="left">Vector of 8 32-bit floating-point values.</td>
1948 </tr>
1949 <tr class="layout">
1950 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1951 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1952 </tr>
1953</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1954
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1955</div>
1956
1957<!-- _______________________________________________________________________ -->
1958<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1959<div class="doc_text">
1960
1961<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1962<p>Opaque types are used to represent unknown types in the system. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1963 corresponds (for example) to the C notion of a forward declared structure
1964 type. In LLVM, opaque types can eventually be resolved to any type (not just
1965 a structure type).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1966
1967<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1968<pre>
1969 opaque
1970</pre>
1971
1972<h5>Examples:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1973<table class="layout">
1974 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1975 <td class="left"><tt>opaque</tt></td>
1976 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1977 </tr>
1978</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1979
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1980</div>
1981
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1982<!-- ======================================================================= -->
1983<div class="doc_subsection">
1984 <a name="t_uprefs">Type Up-references</a>
1985</div>
1986
1987<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1988
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1989<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1990<p>An "up reference" allows you to refer to a lexically enclosing type without
1991 requiring it to have a name. For instance, a structure declaration may
1992 contain a pointer to any of the types it is lexically a member of. Example
1993 of up references (with their equivalent as named type declarations)
1994 include:</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1995
1996<pre>
Chris Lattner5ad632d2009-02-09 10:00:56 +0000[diff] [blame]1997 { \2 * } %x = type { %x* }
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1998 { \2 }* %y = type { %y }*
1999 \1* %z = type %z*
2000</pre>
2001
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2002<p>An up reference is needed by the asmprinter for printing out cyclic types
2003 when there is no declared name for a type in the cycle. Because the
2004 asmprinter does not want to print out an infinite type string, it needs a
2005 syntax to handle recursive types that have no names (all names are optional
2006 in llvm IR).</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]2007
2008<h5>Syntax:</h5>
2009<pre>
2010 \&lt;level&gt;
2011</pre>
2012
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2013<p>The level is the count of the lexical type that is being referred to.</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]2014
2015<h5>Examples:</h5>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]2016<table class="layout">
2017 <tr class="layout">
2018 <td class="left"><tt>\1*</tt></td>
2019 <td class="left">Self-referential pointer.</td>
2020 </tr>
2021 <tr class="layout">
2022 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
2023 <td class="left">Recursive structure where the upref refers to the out-most
2024 structure.</td>
2025 </tr>
2026</table>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]2027
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2028</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2029
2030<!-- *********************************************************************** -->
2031<div class="doc_section"> <a name="constants">Constants</a> </div>
2032<!-- *********************************************************************** -->
2033
2034<div class="doc_text">
2035
2036<p>LLVM has several different basic types of constants. This section describes
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2037 them all and their syntax.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2038
2039</div>
2040
2041<!-- ======================================================================= -->
2042<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
2043
2044<div class="doc_text">
2045
2046<dl>
2047 <dt><b>Boolean constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2048 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]2049 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2050
2051 <dt><b>Integer constants</b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2052 <dd>Standard integers (such as '4') are constants of
2053 the <a href="#t_integer">integer</a> type. Negative numbers may be used
2054 with integer types.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2055
2056 <dt><b>Floating point constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2057 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2058 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
2059 notation (see below). The assembler requires the exact decimal value of a
2060 floating-point constant. For example, the assembler accepts 1.25 but
2061 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
2062 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2063
2064 <dt><b>Null pointer constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2065 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2066 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2067</dl>
2068
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2069<p>The one non-intuitive notation for constants is the hexadecimal form of
2070 floating point constants. For example, the form '<tt>double
2071 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
2072 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
2073 constants are required (and the only time that they are generated by the
2074 disassembler) is when a floating point constant must be emitted but it cannot
2075 be represented as a decimal floating point number in a reasonable number of
2076 digits. For example, NaN's, infinities, and other special values are
2077 represented in their IEEE hexadecimal format so that assembly and disassembly
2078 do not cause any bits to change in the constants.</p>
2079
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]2080<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2081 represented using the 16-digit form shown above (which matches the IEEE754
2082 representation for double); float values must, however, be exactly
2083 representable as IEE754 single precision. Hexadecimal format is always used
2084 for long double, and there are three forms of long double. The 80-bit format
2085 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
2086 The 128-bit format used by PowerPC (two adjacent doubles) is represented
2087 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
2088 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
2089 currently supported target uses this format. Long doubles will only work if
2090 they match the long double format on your target. All hexadecimal formats
2091 are big-endian (sign bit at the left).</p>
2092
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2093</div>
2094
2095<!-- ======================================================================= -->
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]2096<div class="doc_subsection">
Bill Wendling1a2630a2009-07-20 02:32:41 +0000[diff] [blame]2097<a name="aggregateconstants"></a> <!-- old anchor -->
2098<a name="complexconstants">Complex Constants</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2099</div>
2100
2101<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2102
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]2103<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2104 constants and smaller complex constants.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2105
2106<dl>
2107 <dt><b>Structure constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2108 <dd>Structure constants are represented with notation similar to structure
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2109 type definitions (a comma separated list of elements, surrounded by braces
2110 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
2111 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
2112 Structure constants must have <a href="#t_struct">structure type</a>, and
2113 the number and types of elements must match those specified by the
2114 type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2115
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]2116 <dt><b>Union constants</b></dt>
2117 <dd>Union constants are represented with notation similar to a structure with
2118 a single element - that is, a single typed element surrounded
2119 by braces (<tt>{}</tt>)). For example: "<tt>{ i32 4 }</tt>". The
2120 <a href="#t_union">union type</a> can be initialized with a single-element
2121 struct as long as the type of the struct element matches the type of
2122 one of the union members.</dd>
2123
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2124 <dt><b>Array constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2125 <dd>Array constants are represented with notation similar to array type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2126 definitions (a comma separated list of elements, surrounded by square
2127 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
2128 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
2129 the number and types of elements must match those specified by the
2130 type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2131
2132 <dt><b>Vector constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2133 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2134 definitions (a comma separated list of elements, surrounded by
2135 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
2136 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
2137 have <a href="#t_vector">vector type</a>, and the number and types of
2138 elements must match those specified by the type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2139
2140 <dt><b>Zero initialization</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2141 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]2142 value to zero of <em>any</em> type, including scalar and
2143 <a href="#t_aggregate">aggregate</a> types.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2144 This is often used to avoid having to print large zero initializers
2145 (e.g. for large arrays) and is always exactly equivalent to using explicit
2146 zero initializers.</dd>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2147
2148 <dt><b>Metadata node</b></dt>
Nick Lewyckyf122c7e2009-05-30 16:08:30 +0000[diff] [blame]2149 <dd>A metadata node is a structure-like constant with
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2150 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
2151 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
2152 be interpreted as part of the instruction stream, metadata is a place to
2153 attach additional information such as debug info.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2154</dl>
2155
2156</div>
2157
2158<!-- ======================================================================= -->
2159<div class="doc_subsection">
2160 <a name="globalconstants">Global Variable and Function Addresses</a>
2161</div>
2162
2163<div class="doc_text">
2164
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2165<p>The addresses of <a href="#globalvars">global variables</a>
2166 and <a href="#functionstructure">functions</a> are always implicitly valid
2167 (link-time) constants. These constants are explicitly referenced when
2168 the <a href="#identifiers">identifier for the global</a> is used and always
2169 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2170 legal LLVM file:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2171
2172<div class="doc_code">
2173<pre>
2174@X = global i32 17
2175@Y = global i32 42
2176@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2177</pre>
2178</div>
2179
2180</div>
2181
2182<!-- ======================================================================= -->
2183<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2184<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2185
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2186<p>The string '<tt>undef</tt>' can be used anywhere a constant is expected, and
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2187 indicates that the user of the value may receive an unspecified bit-pattern.
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2188 Undefined values may be of any type (other than label or void) and be used
2189 anywhere a constant is permitted.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2190
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2191<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2192 program is well defined no matter what value is used. This gives the
2193 compiler more freedom to optimize. Here are some examples of (potentially
2194 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2195
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2196
2197<div class="doc_code">
2198<pre>
2199 %A = add %X, undef
2200 %B = sub %X, undef
2201 %C = xor %X, undef
2202Safe:
2203 %A = undef
2204 %B = undef
2205 %C = undef
2206</pre>
2207</div>
2208
2209<p>This is safe because all of the output bits are affected by the undef bits.
2210Any output bit can have a zero or one depending on the input bits.</p>
2211
2212<div class="doc_code">
2213<pre>
2214 %A = or %X, undef
2215 %B = and %X, undef
2216Safe:
2217 %A = -1
2218 %B = 0
2219Unsafe:
2220 %A = undef
2221 %B = undef
2222</pre>
2223</div>
2224
2225<p>These logical operations have bits that are not always affected by the input.
2226For example, if "%X" has a zero bit, then the output of the 'and' operation will
2227always be a zero, no matter what the corresponding bit from the undef is. As
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2228such, it is unsafe to optimize or assume that the result of the and is undef.
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2229However, it is safe to assume that all bits of the undef could be 0, and
2230optimize the and to 0. Likewise, it is safe to assume that all the bits of
2231the undef operand to the or could be set, allowing the or to be folded to
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2232-1.</p>
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2233
2234<div class="doc_code">
2235<pre>
2236 %A = select undef, %X, %Y
2237 %B = select undef, 42, %Y
2238 %C = select %X, %Y, undef
2239Safe:
2240 %A = %X (or %Y)
2241 %B = 42 (or %Y)
2242 %C = %Y
2243Unsafe:
2244 %A = undef
2245 %B = undef
2246 %C = undef
2247</pre>
2248</div>
2249
2250<p>This set of examples show that undefined select (and conditional branch)
2251conditions can go "either way" but they have to come from one of the two
2252operands. In the %A example, if %X and %Y were both known to have a clear low
2253bit, then %A would have to have a cleared low bit. However, in the %C example,
2254the optimizer is allowed to assume that the undef operand could be the same as
2255%Y, allowing the whole select to be eliminated.</p>
2256
2257
2258<div class="doc_code">
2259<pre>
2260 %A = xor undef, undef
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2261
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2262 %B = undef
2263 %C = xor %B, %B
2264
2265 %D = undef
2266 %E = icmp lt %D, 4
2267 %F = icmp gte %D, 4
2268
2269Safe:
2270 %A = undef
2271 %B = undef
2272 %C = undef
2273 %D = undef
2274 %E = undef
2275 %F = undef
2276</pre>
2277</div>
2278
2279<p>This example points out that two undef operands are not necessarily the same.
2280This can be surprising to people (and also matches C semantics) where they
2281assume that "X^X" is always zero, even if X is undef. This isn't true for a
2282number of reasons, but the short answer is that an undef "variable" can
2283arbitrarily change its value over its "live range". This is true because the
2284"variable" doesn't actually <em>have a live range</em>. Instead, the value is
2285logically read from arbitrary registers that happen to be around when needed,
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2286so the value is not necessarily consistent over time. In fact, %A and %C need
Chris Lattnerc802e482009-09-08 15:13:16 +0000[diff] [blame]2287to have the same semantics or the core LLVM "replace all uses with" concept
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2288would not hold.</p>
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2289
2290<div class="doc_code">
2291<pre>
2292 %A = fdiv undef, %X
2293 %B = fdiv %X, undef
2294Safe:
2295 %A = undef
2296b: unreachable
2297</pre>
2298</div>
2299
2300<p>These examples show the crucial difference between an <em>undefined
2301value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2302allowed to have an arbitrary bit-pattern. This means that the %A operation
2303can be constant folded to undef because the undef could be an SNaN, and fdiv is
2304not (currently) defined on SNaN's. However, in the second example, we can make
2305a more aggressive assumption: because the undef is allowed to be an arbitrary
2306value, we are allowed to assume that it could be zero. Since a divide by zero
Chris Lattner83d45d72009-09-08 19:45:34 +0000[diff] [blame]2307has <em>undefined behavior</em>, we are allowed to assume that the operation
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2308does not execute at all. This allows us to delete the divide and all code after
2309it: since the undefined operation "can't happen", the optimizer can assume that
2310it occurs in dead code.
2311</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2312
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2313<div class="doc_code">
2314<pre>
2315a: store undef -> %X
2316b: store %X -> undef
2317Safe:
2318a: &lt;deleted&gt;
2319b: unreachable
2320</pre>
2321</div>
2322
2323<p>These examples reiterate the fdiv example: a store "of" an undefined value
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2324can be assumed to not have any effect: we can assume that the value is
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2325overwritten with bits that happen to match what was already there. However, a
2326store "to" an undefined location could clobber arbitrary memory, therefore, it
2327has undefined behavior.</p>
2328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2329</div>
2330
2331<!-- ======================================================================= -->
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]2332<div class="doc_subsection"><a name="trapvalues">Trap Values</a></div>
2333<div class="doc_text">
2334
Dan Gohman67bf37f2010-04-26 20:21:21 +0000[diff] [blame]2335<p>Trap values are similar to <a href="#undefvalues">undef values</a>, however
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]2336 instead of representing an unspecified bit pattern, they represent the
2337 fact that an instruction or constant expression which cannot evoke side
2338 effects has nevertheless detected a condition which results in undefined
Dan Gohman67bf37f2010-04-26 20:21:21 +0000[diff] [blame]2339 behavior.</p>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]2340
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2341<p>There is currently no way of representing a trap value in the IR; they
Dan Gohmanc4bfe502010-05-03 14:51:43 +0000[diff] [blame]2342 only exist when produced by operations such as
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2343 <a href="#i_add"><tt>add</tt></a> with the <tt>nsw</tt> flag.</p>
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2344
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2345<p>Trap value behavior is defined in terms of value <i>dependence</i>:</p>
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2346
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2347<p>
2348<ul>
2349<li>Values other than <a href="#i_phi"><tt>phi</tt></a> nodes depend on
2350 their operands.</li>
2351
2352<li><a href="#i_phi"><tt>Phi</tt></a> nodes depend on the operand corresponding
2353 to their dynamic predecessor basic block.</li>
2354
2355<li>Function arguments depend on the corresponding actual argument values in
2356 the dynamic callers of their functions.</li>
2357
2358<li><a href="#i_call"><tt>Call</tt></a> instructions depend on the
2359 <a href="#i_ret"><tt>ret</tt></a> instructions that dynamically transfer
2360 control back to them.</li>
2361
Dan Gohman5e7b8fb2010-05-03 14:55:22 +0000[diff] [blame]2362<li><a href="#i_invoke"><tt>Invoke</tt></a> instructions depend on the
2363 <a href="#i_ret"><tt>ret</tt></a>, <a href="#i_unwind"><tt>unwind</tt></a>,
2364 or exception-throwing call instructions that dynamically transfer control
2365 back to them.</li>
2366
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2367<li>Non-volatile loads and stores depend on the most recent stores to all of the
2368 referenced memory addresses, following the order in the IR
2369 (including loads and stores implied by intrinsics such as
2370 <a href="#int_memcpy"><tt>@llvm.memcpy</tt></a>.)</li>
2371
Dan Gohmane2c06b62010-05-03 14:59:34 +0000[diff] [blame]2372<!-- TODO: In the case of multiple threads, this only applies if the store
2373 "happens-before" the load or store. -->
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2374
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2375<!-- TODO: floating-point exception state -->
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2376
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2377<li>An instruction with externally visible side effects depends on the most
2378 recent preceding instruction with externally visible side effects, following
2379 the order in the IR. (This includes volatile loads and stores.)</li>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]2380
Dan Gohman5e7b8fb2010-05-03 14:55:22 +0000[diff] [blame]2381<li>An instruction <i>control-depends</i> on a
2382 <a href="#terminators">terminator instruction</a>
2383 if the terminator instruction has multiple successors and the instruction
2384 is always executed when control transfers to one of the successors, and
2385 may not be executed when control is transfered to another.</li>
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2386
2387<li>Dependence is transitive.</li>
2388
2389</ul>
2390</p>
2391
2392<p>Whenever a trap value is generated, all values which depend on it evaluate
2393 to trap. If they have side effects, the evoke their side effects as if each
2394 operand with a trap value were undef. If they have externally-visible side
2395 effects, the behavior is undefined.</p>
2396
2397<p>Here are some examples:</p>
Dan Gohman54884272010-04-26 20:54:53 +0000[diff] [blame]2398
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2399<div class="doc_code">
2400<pre>
2401entry:
2402 %trap = sub nuw i32 0, 1 ; Results in a trap value.
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2403 %still_trap = and i32 %trap, 0 ; Whereas (and i32 undef, 0) would return 0.
2404 %trap_yet_again = getelementptr i32* @h, i32 %still_trap
2405 store i32 0, i32* %trap_yet_again ; undefined behavior
2406
2407 store i32 %trap, i32* @g ; Trap value conceptually stored to memory.
2408 %trap2 = load i32* @g ; Returns a trap value, not just undef.
2409
2410 volatile store i32 %trap, i32* @g ; External observation; undefined behavior.
2411
2412 %narrowaddr = bitcast i32* @g to i16*
2413 %wideaddr = bitcast i32* @g to i64*
2414 %trap3 = load 16* %narrowaddr ; Returns a trap value.
2415 %trap4 = load i64* %widaddr ; Returns a trap value.
2416
2417 %cmp = icmp i32 slt %trap, 0 ; Returns a trap value.
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2418 %br i1 %cmp, %true, %end ; Branch to either destination.
2419
2420true:
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2421 volatile store i32 0, i32* @g ; This is control-dependent on %cmp, so
2422 ; it has undefined behavior.
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2423 br label %end
2424
2425end:
2426 %p = phi i32 [ 0, %entry ], [ 1, %true ]
2427 ; Both edges into this PHI are
2428 ; control-dependent on %cmp, so this
Dan Gohman762c0362010-04-28 00:49:41 +0000[diff] [blame]2429 ; always results in a trap value.
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2430
2431 volatile store i32 0, i32* @g ; %end is control-equivalent to %entry
2432 ; so this is defined (ignoring earlier
2433 ; undefined behavior in this example).
Dan Gohman568ca042010-04-26 23:36:52 +0000[diff] [blame]2434</pre>
2435</div>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]2436
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]2437</div>
2438
2439<!-- ======================================================================= -->
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]2440<div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2441 Blocks</a></div>
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2442<div class="doc_text">
2443
Chris Lattner620cead2009-11-01 01:27:45 +0000[diff] [blame]2444<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2445
2446<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner24f2e852009-10-27 21:49:40 +0000[diff] [blame]2447 basic block in the specified function, and always has an i8* type. Taking
Chris Lattner620cead2009-11-01 01:27:45 +0000[diff] [blame]2448 the address of the entry block is illegal.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2449
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2450<p>This value only has defined behavior when used as an operand to the
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2451 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2452 against null. Pointer equality tests between labels addresses is undefined
2453 behavior - though, again, comparison against null is ok, and no label is
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]2454 equal to the null pointer. This may also be passed around as an opaque
2455 pointer sized value as long as the bits are not inspected. This allows
Chris Lattner0bae7b32009-10-27 21:44:20 +0000[diff] [blame]2456 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2457 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2458
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]2459<p>Finally, some targets may provide defined semantics when
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2460 using the value as the operand to an inline assembly, but that is target
2461 specific.
2462 </p>
2463
2464</div>
2465
2466
2467<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2468<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2469</div>
2470
2471<div class="doc_text">
2472
2473<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2474 to be used as constants. Constant expressions may be of
2475 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2476 operation that does not have side effects (e.g. load and call are not
2477 supported). The following is the syntax for constant expressions:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2478
2479<dl>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2480 <dt><b><tt>trunc (CST to TYPE)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2481 <dd>Truncate a constant to another type. The bit size of CST must be larger
2482 than the bit size of TYPE. Both types must be integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2483
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2484 <dt><b><tt>zext (CST to TYPE)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2485 <dd>Zero extend a constant to another type. The bit size of CST must be
2486 smaller or equal to the bit size of TYPE. Both types must be
2487 integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2488
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2489 <dt><b><tt>sext (CST to TYPE)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2490 <dd>Sign extend a constant to another type. The bit size of CST must be
2491 smaller or equal to the bit size of TYPE. Both types must be
2492 integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2493
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2494 <dt><b><tt>fptrunc (CST to TYPE)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2495 <dd>Truncate a floating point constant to another floating point type. The
2496 size of CST must be larger than the size of TYPE. Both types must be
2497 floating point.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2498
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2499 <dt><b><tt>fpext (CST to TYPE)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2500 <dd>Floating point extend a constant to another type. The size of CST must be
2501 smaller or equal to the size of TYPE. Both types must be floating
2502 point.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2503
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2504 <dt><b><tt>fptoui (CST to TYPE)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2505 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2506 constant. TYPE must be a scalar or vector integer type. CST must be of
2507 scalar or vector floating point type. Both CST and TYPE must be scalars,
2508 or vectors of the same number of elements. If the value won't fit in the
2509 integer type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2510
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2511 <dt><b><tt>fptosi (CST to TYPE)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2512 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2513 constant. TYPE must be a scalar or vector integer type. CST must be of
2514 scalar or vector floating point type. Both CST and TYPE must be scalars,
2515 or vectors of the same number of elements. If the value won't fit in the
2516 integer type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2517
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2518 <dt><b><tt>uitofp (CST to TYPE)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2519 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2520 constant. TYPE must be a scalar or vector floating point type. CST must be
2521 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2522 vectors of the same number of elements. If the value won't fit in the
2523 floating point type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2524
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2525 <dt><b><tt>sitofp (CST to TYPE)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2526 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2527 constant. TYPE must be a scalar or vector floating point type. CST must be
2528 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2529 vectors of the same number of elements. If the value won't fit in the
2530 floating point type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2531
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2532 <dt><b><tt>ptrtoint (CST to TYPE)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2533 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2534 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2535 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2536 make it fit in <tt>TYPE</tt>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2537
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2538 <dt><b><tt>inttoptr (CST to TYPE)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2539 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2540 type. CST must be of integer type. The CST value is zero extended,
2541 truncated, or unchanged to make it fit in a pointer size. This one is
2542 <i>really</i> dangerous!</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2543
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2544 <dt><b><tt>bitcast (CST to TYPE)</tt></b></dt>
Chris Lattner557bc5d2009-02-28 18:27:03 +0000[diff] [blame]2545 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2546 are the same as those for the <a href="#i_bitcast">bitcast
2547 instruction</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2548
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2549 <dt><b><tt>getelementptr (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
2550 <dt><b><tt>getelementptr inbounds (CSTPTR, IDX0, IDX1, ...)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2551 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2552 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2553 instruction, the index list may have zero or more indexes, which are
2554 required to make sense for the type of "CSTPTR".</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2555
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2556 <dt><b><tt>select (COND, VAL1, VAL2)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2557 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2558
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2559 <dt><b><tt>icmp COND (VAL1, VAL2)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2560 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2561
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2562 <dt><b><tt>fcmp COND (VAL1, VAL2)</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2563 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2564
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2565 <dt><b><tt>extractelement (VAL, IDX)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2566 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2567 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2568
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2569 <dt><b><tt>insertelement (VAL, ELT, IDX)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2570 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2571 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2572
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2573 <dt><b><tt>shufflevector (VEC1, VEC2, IDXMASK)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2574 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2575 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2576
Nick Lewyckyeb94e312010-05-29 06:44:15 +0000[diff] [blame]2577 <dt><b><tt>extractvalue (VAL, IDX0, IDX1, ...)</tt></b></dt>
2578 <dd>Perform the <a href="#i_extractvalue">extractvalue operation</a> on
2579 constants. The index list is interpreted in a similar manner as indices in
2580 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2581 index value must be specified.</dd>
2582
2583 <dt><b><tt>insertvalue (VAL, ELT, IDX0, IDX1, ...)</tt></b></dt>
2584 <dd>Perform the <a href="#i_insertvalue">insertvalue operation</a> on
2585 constants. The index list is interpreted in a similar manner as indices in
2586 a '<a href="#i_getelementptr">getelementptr</a>' operation. At least one
2587 index value must be specified.</dd>
2588
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]2589 <dt><b><tt>OPCODE (LHS, RHS)</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2590 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2591 be any of the <a href="#binaryops">binary</a>
2592 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2593 on operands are the same as those for the corresponding instruction
2594 (e.g. no bitwise operations on floating point values are allowed).</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2595</dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2596
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2597</div>
2598
2599<!-- *********************************************************************** -->
2600<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2601<!-- *********************************************************************** -->
2602
2603<!-- ======================================================================= -->
2604<div class="doc_subsection">
2605<a name="inlineasm">Inline Assembler Expressions</a>
2606</div>
2607
2608<div class="doc_text">
2609
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2610<p>LLVM supports inline assembler expressions (as opposed
2611 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2612 a special value. This value represents the inline assembler as a string
2613 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2614 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2615 expression has side effects, and a flag indicating whether the function
2616 containing the asm needs to align its stack conservatively. An example
2617 inline assembler expression is:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2618
2619<div class="doc_code">
2620<pre>
2621i32 (i32) asm "bswap $0", "=r,r"
2622</pre>
2623</div>
2624
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2625<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2626 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2627 have:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2628
2629<div class="doc_code">
2630<pre>
2631%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2632</pre>
2633</div>
2634
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2635<p>Inline asms with side effects not visible in the constraint list must be
2636 marked as having side effects. This is done through the use of the
2637 '<tt>sideeffect</tt>' keyword, like so:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2638
2639<div class="doc_code">
2640<pre>
2641call void asm sideeffect "eieio", ""()
2642</pre>
2643</div>
2644
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2645<p>In some cases inline asms will contain code that will not work unless the
2646 stack is aligned in some way, such as calls or SSE instructions on x86,
2647 yet will not contain code that does that alignment within the asm.
2648 The compiler should make conservative assumptions about what the asm might
2649 contain and should generate its usual stack alignment code in the prologue
2650 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2651
2652<div class="doc_code">
2653<pre>
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2654call void asm alignstack "eieio", ""()
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2655</pre>
2656</div>
2657
2658<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2659 first.</p>
2660
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2661<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2662 documented here. Constraints on what can be done (e.g. duplication, moving,
2663 etc need to be documented). This is probably best done by reference to
2664 another document that covers inline asm from a holistic perspective.</p>
Chris Lattnerbafc8372010-04-07 05:38:05 +0000[diff] [blame]2665</div>
2666
2667<div class="doc_subsubsection">
2668<a name="inlineasm_md">Inline Asm Metadata</a>
2669</div>
2670
2671<div class="doc_text">
2672
2673<p>The call instructions that wrap inline asm nodes may have a "!srcloc" MDNode
2674 attached to it that contains a constant integer. If present, the code
2675 generator will use the integer as the location cookie value when report
2676 errors through the LLVMContext error reporting mechanisms. This allows a
Dan Gohmanfde3cd72010-04-28 00:36:01 +0000[diff] [blame]2677 front-end to correlate backend errors that occur with inline asm back to the
Chris Lattnerbafc8372010-04-07 05:38:05 +0000[diff] [blame]2678 source code that produced it. For example:</p>
2679
2680<div class="doc_code">
2681<pre>
2682call void asm sideeffect "something bad", ""()<b>, !srcloc !42</b>
2683...
2684!42 = !{ i32 1234567 }
2685</pre>
2686</div>
2687
2688<p>It is up to the front-end to make sense of the magic numbers it places in the
2689 IR.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2690
2691</div>
2692
Chris Lattnerd0d96292010-01-15 21:50:19 +0000[diff] [blame]2693<!-- ======================================================================= -->
2694<div class="doc_subsection"><a name="metadata">Metadata Nodes and Metadata
2695 Strings</a>
2696</div>
2697
2698<div class="doc_text">
2699
2700<p>LLVM IR allows metadata to be attached to instructions in the program that
2701 can convey extra information about the code to the optimizers and code
2702 generator. One example application of metadata is source-level debug
2703 information. There are two metadata primitives: strings and nodes. All
2704 metadata has the <tt>metadata</tt> type and is identified in syntax by a
2705 preceding exclamation point ('<tt>!</tt>').</p>
2706
2707<p>A metadata string is a string surrounded by double quotes. It can contain
2708 any character by escaping non-printable characters with "\xx" where "xx" is
2709 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
2710
2711<p>Metadata nodes are represented with notation similar to structure constants
2712 (a comma separated list of elements, surrounded by braces and preceded by an
2713 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2714 10}</tt>". Metadata nodes can have any values as their operand.</p>
2715
2716<p>A <a href="#namedmetadatastructure">named metadata</a> is a collection of
2717 metadata nodes, which can be looked up in the module symbol table. For
2718 example: "<tt>!foo = metadata !{!4, !3}</tt>".
2719
Devang Patelb1586922010-03-04 23:44:48 +0000[diff] [blame]2720<p>Metadata can be used as function arguments. Here <tt>llvm.dbg.value</tt>
2721 function is using two metadata arguments.
2722
2723 <div class="doc_code">
2724 <pre>
2725 call void @llvm.dbg.value(metadata !24, i64 0, metadata !25)
2726 </pre>
2727 </div></p>
2728
2729<p>Metadata can be attached with an instruction. Here metadata <tt>!21</tt> is
2730 attached with <tt>add</tt> instruction using <tt>!dbg</tt> identifier.
2731
2732 <div class="doc_code">
2733 <pre>
2734 %indvar.next = add i64 %indvar, 1, !dbg !21
2735 </pre>
2736 </div></p>
Chris Lattnerd0d96292010-01-15 21:50:19 +0000[diff] [blame]2737</div>
2738
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2739
2740<!-- *********************************************************************** -->
2741<div class="doc_section">
2742 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2743</div>
2744<!-- *********************************************************************** -->
2745
2746<p>LLVM has a number of "magic" global variables that contain data that affect
2747code generation or other IR semantics. These are documented here. All globals
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]2748of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2749section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2750by LLVM.</p>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2751
2752<!-- ======================================================================= -->
2753<div class="doc_subsection">
2754<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2755</div>
2756
2757<div class="doc_text">
2758
2759<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2760href="#linkage_appending">appending linkage</a>. This array contains a list of
2761pointers to global variables and functions which may optionally have a pointer
2762cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2763
2764<pre>
2765 @X = global i8 4
2766 @Y = global i32 123
2767
2768 @llvm.used = appending global [2 x i8*] [
2769 i8* @X,
2770 i8* bitcast (i32* @Y to i8*)
2771 ], section "llvm.metadata"
2772</pre>
2773
2774<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2775compiler, assembler, and linker are required to treat the symbol as if there is
2776a reference to the global that it cannot see. For example, if a variable has
2777internal linkage and no references other than that from the <tt>@llvm.used</tt>
2778list, it cannot be deleted. This is commonly used to represent references from
2779inline asms and other things the compiler cannot "see", and corresponds to
2780"attribute((used))" in GNU C.</p>
2781
2782<p>On some targets, the code generator must emit a directive to the assembler or
2783object file to prevent the assembler and linker from molesting the symbol.</p>
2784
2785</div>
2786
2787<!-- ======================================================================= -->
2788<div class="doc_subsection">
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]2789<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2790</div>
2791
2792<div class="doc_text">
2793
2794<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2795<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2796touching the symbol. On targets that support it, this allows an intelligent
2797linker to optimize references to the symbol without being impeded as it would be
2798by <tt>@llvm.used</tt>.</p>
2799
2800<p>This is a rare construct that should only be used in rare circumstances, and
2801should not be exposed to source languages.</p>
2802
2803</div>
2804
2805<!-- ======================================================================= -->
2806<div class="doc_subsection">
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2807<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2808</div>
2809
2810<div class="doc_text">
David Chisnall47e8b772010-04-30 19:23:49 +0000[diff] [blame]2811<pre>
2812%0 = type { i32, void ()* }
David Chisnalla07a3c22010-04-30 19:27:35 +0000[diff] [blame]2813@llvm.global_ctors = appending global [1 x %0] [%0 { i32 65535, void ()* @ctor }]
David Chisnall47e8b772010-04-30 19:23:49 +0000[diff] [blame]2814</pre>
2815<p>The <tt>@llvm.global_ctors</tt> array contains a list of constructor functions and associated priorities. The functions referenced by this array will be called in ascending order of priority (i.e. lowest first) when the module is loaded. The order of functions with the same priority is not defined.
2816</p>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2817
2818</div>
2819
2820<!-- ======================================================================= -->
2821<div class="doc_subsection">
2822<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2823</div>
2824
2825<div class="doc_text">
David Chisnall47e8b772010-04-30 19:23:49 +0000[diff] [blame]2826<pre>
2827%0 = type { i32, void ()* }
David Chisnalla07a3c22010-04-30 19:27:35 +0000[diff] [blame]2828@llvm.global_dtors = appending global [1 x %0] [%0 { i32 65535, void ()* @dtor }]
David Chisnall47e8b772010-04-30 19:23:49 +0000[diff] [blame]2829</pre>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2830
David Chisnall47e8b772010-04-30 19:23:49 +0000[diff] [blame]2831<p>The <tt>@llvm.global_dtors</tt> array contains a list of destructor functions and associated priorities. The functions referenced by this array will be called in descending order of priority (i.e. highest first) when the module is loaded. The order of functions with the same priority is not defined.
2832</p>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2833
2834</div>
2835
2836
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2837<!-- *********************************************************************** -->
2838<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2839<!-- *********************************************************************** -->
2840
2841<div class="doc_text">
2842
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2843<p>The LLVM instruction set consists of several different classifications of
2844 instructions: <a href="#terminators">terminator
2845 instructions</a>, <a href="#binaryops">binary instructions</a>,
2846 <a href="#bitwiseops">bitwise binary instructions</a>,
2847 <a href="#memoryops">memory instructions</a>, and
2848 <a href="#otherops">other instructions</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2849
2850</div>
2851
2852<!-- ======================================================================= -->
2853<div class="doc_subsection"> <a name="terminators">Terminator
2854Instructions</a> </div>
2855
2856<div class="doc_text">
2857
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2858<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2859 in a program ends with a "Terminator" instruction, which indicates which
2860 block should be executed after the current block is finished. These
2861 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2862 control flow, not values (the one exception being the
2863 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2864
Duncan Sands048d8062010-04-15 20:35:54 +0000[diff] [blame]2865<p>There are seven different terminator instructions: the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2866 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2867 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2868 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
Bill Wendlingb4ae2292009-11-02 00:25:26 +0000[diff] [blame]2869 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2870 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2871 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2872 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2873
2874</div>
2875
2876<!-- _______________________________________________________________________ -->
2877<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2878Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2879
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2880<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2881
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2882<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2883<pre>
2884 ret &lt;type&gt; &lt;value&gt; <i>; Return a value from a non-void function</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2885 ret void <i>; Return from void function</i>
2886</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2887
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2888<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2889<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2890 a value) from a function back to the caller.</p>
2891
2892<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2893 value and then causes control flow, and one that just causes control flow to
2894 occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2895
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2896<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2897<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2898 return value. The type of the return value must be a
2899 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2900
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2901<p>A function is not <a href="#wellformed">well formed</a> if it it has a
2902 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2903 value or a return value with a type that does not match its type, or if it
2904 has a void return type and contains a '<tt>ret</tt>' instruction with a
2905 return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2906
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2907<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2908<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2909 the calling function's context. If the caller is a
2910 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2911 instruction after the call. If the caller was an
2912 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2913 the beginning of the "normal" destination block. If the instruction returns
2914 a value, that value shall set the call or invoke instruction's return
2915 value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2916
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2917<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2918<pre>
2919 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2920 ret void <i>; Return from a void function</i>
Bill Wendlingd163e2d2009-02-28 22:12:54 +0000[diff] [blame]2921 ret { i32, i8 } { i32 4, i8 2 } <i>; Return a struct of values 4 and 2</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2922</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2923
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2924</div>
2925<!-- _______________________________________________________________________ -->
2926<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2927
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2928<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2929
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2930<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2931<pre>
2932 br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2933</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2934
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2935<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2936<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2937 different basic block in the current function. There are two forms of this
2938 instruction, corresponding to a conditional branch and an unconditional
2939 branch.</p>
2940
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2941<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2942<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2943 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2944 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2945 target.</p>
2946
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2947<h5>Semantics:</h5>
2948<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2949 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2950 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2951 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2952
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2953<h5>Example:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]2954<pre>
2955Test:
2956 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2957 br i1 %cond, label %IfEqual, label %IfUnequal
2958IfEqual:
2959 <a href="#i_ret">ret</a> i32 1
2960IfUnequal:
2961 <a href="#i_ret">ret</a> i32 0
2962</pre>
2963
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2964</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2965
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2966<!-- _______________________________________________________________________ -->
2967<div class="doc_subsubsection">
2968 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2969</div>
2970
2971<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2972
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2973<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2974<pre>
2975 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2976</pre>
2977
2978<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2979<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2980 several different places. It is a generalization of the '<tt>br</tt>'
2981 instruction, allowing a branch to occur to one of many possible
2982 destinations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2983
2984<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2985<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2986 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2987 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2988 The table is not allowed to contain duplicate constant entries.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2989
2990<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2991<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2992 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2993 is searched for the given value. If the value is found, control flow is
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2994 transferred to the corresponding destination; otherwise, control flow is
2995 transferred to the default destination.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2996
2997<h5>Implementation:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2998<p>Depending on properties of the target machine and the particular
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2999 <tt>switch</tt> instruction, this instruction may be code generated in
3000 different ways. For example, it could be generated as a series of chained
3001 conditional branches or with a lookup table.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3002
3003<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3004<pre>
3005 <i>; Emulate a conditional br instruction</i>
3006 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]3007 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3008
3009 <i>; Emulate an unconditional br instruction</i>
3010 switch i32 0, label %dest [ ]
3011
3012 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]3013 switch i32 %val, label %otherwise [ i32 0, label %onzero
3014 i32 1, label %onone
3015 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3016</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3017
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3018</div>
3019
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3020
3021<!-- _______________________________________________________________________ -->
3022<div class="doc_subsubsection">
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]3023 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3024</div>
3025
3026<div class="doc_text">
3027
3028<h5>Syntax:</h5>
3029<pre>
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]3030 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3031</pre>
3032
3033<h5>Overview:</h5>
3034
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]3035<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3036 within the current function, whose address is specified by
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]3037 "<tt>address</tt>". Address must be derived from a <a
3038 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3039
3040<h5>Arguments:</h5>
3041
3042<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
3043 rest of the arguments indicate the full set of possible destinations that the
3044 address may point to. Blocks are allowed to occur multiple times in the
3045 destination list, though this isn't particularly useful.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3046
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3047<p>This destination list is required so that dataflow analysis has an accurate
3048 understanding of the CFG.</p>
3049
3050<h5>Semantics:</h5>
3051
3052<p>Control transfers to the block specified in the address argument. All
3053 possible destination blocks must be listed in the label list, otherwise this
3054 instruction has undefined behavior. This implies that jumps to labels
3055 defined in other functions have undefined behavior as well.</p>
3056
3057<h5>Implementation:</h5>
3058
3059<p>This is typically implemented with a jump through a register.</p>
3060
3061<h5>Example:</h5>
3062<pre>
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]3063 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]3064</pre>
3065
3066</div>
3067
3068
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3069<!-- _______________________________________________________________________ -->
3070<div class="doc_subsubsection">
3071 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
3072</div>
3073
3074<div class="doc_text">
3075
3076<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3077<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]3078 &lt;result&gt; = invoke [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ptr to function ty&gt; &lt;function ptr val&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3079 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
3080</pre>
3081
3082<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3083<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3084 function, with the possibility of control flow transfer to either the
3085 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
3086 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
3087 control flow will return to the "normal" label. If the callee (or any
3088 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
3089 instruction, control is interrupted and continued at the dynamically nearest
3090 "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3091
3092<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3093<p>This instruction requires several arguments:</p>
3094
3095<ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3096 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
3097 convention</a> the call should use. If none is specified, the call
3098 defaults to using C calling conventions.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]3099
3100 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3101 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
3102 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]3103
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3104 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3105 function value being invoked. In most cases, this is a direct function
3106 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
3107 off an arbitrary pointer to function value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3108
3109 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3110 function to be invoked. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3111
3112 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner553fb1e2010-03-02 06:36:51 +0000[diff] [blame]3113 signature argument types and parameter attributes. All arguments must be
3114 of <a href="#t_firstclass">first class</a> type. If the function
3115 signature indicates the function accepts a variable number of arguments,
3116 the extra arguments can be specified.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3117
3118 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3119 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3120
3121 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3122 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3123
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]3124 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3125 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
3126 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3127</ol>
3128
3129<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3130<p>This instruction is designed to operate as a standard
3131 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
3132 primary difference is that it establishes an association with a label, which
3133 is used by the runtime library to unwind the stack.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3134
3135<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3136 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
3137 exception. Additionally, this is important for implementation of
3138 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3139
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3140<p>For the purposes of the SSA form, the definition of the value returned by the
3141 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
3142 block to the "normal" label. If the callee unwinds then no return value is
3143 available.</p>
Dan Gohman140ba5d2009-05-22 21:47:08 +0000[diff] [blame]3144
Chris Lattner4a91ef42010-01-15 18:08:37 +0000[diff] [blame]3145<p>Note that the code generator does not yet completely support unwind, and
3146that the invoke/unwind semantics are likely to change in future versions.</p>
3147
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3148<h5>Example:</h5>
3149<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]3150 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3151 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]3152 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3153 unwind label %TestCleanup <i>; {i32}:retval set</i>
3154</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3155
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3156</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3157
3158<!-- _______________________________________________________________________ -->
3159
3160<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
3161Instruction</a> </div>
3162
3163<div class="doc_text">
3164
3165<h5>Syntax:</h5>
3166<pre>
3167 unwind
3168</pre>
3169
3170<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3171<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3172 at the first callee in the dynamic call stack which used
3173 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
3174 This is primarily used to implement exception handling.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3175
3176<h5>Semantics:</h5>
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3177<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3178 immediately halt. The dynamic call stack is then searched for the
3179 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
3180 Once found, execution continues at the "exceptional" destination block
3181 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
3182 instruction in the dynamic call chain, undefined behavior results.</p>
3183
Chris Lattner4a91ef42010-01-15 18:08:37 +0000[diff] [blame]3184<p>Note that the code generator does not yet completely support unwind, and
3185that the invoke/unwind semantics are likely to change in future versions.</p>
3186
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3187</div>
3188
3189<!-- _______________________________________________________________________ -->
3190
3191<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
3192Instruction</a> </div>
3193
3194<div class="doc_text">
3195
3196<h5>Syntax:</h5>
3197<pre>
3198 unreachable
3199</pre>
3200
3201<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3202<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3203 instruction is used to inform the optimizer that a particular portion of the
3204 code is not reachable. This can be used to indicate that the code after a
3205 no-return function cannot be reached, and other facts.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3206
3207<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3208<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3209
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3210</div>
3211
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3212<!-- ======================================================================= -->
3213<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3214
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3215<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3216
3217<p>Binary operators are used to do most of the computation in a program. They
3218 require two operands of the same type, execute an operation on them, and
3219 produce a single value. The operands might represent multiple data, as is
3220 the case with the <a href="#t_vector">vector</a> data type. The result value
3221 has the same type as its operands.</p>
3222
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3223<p>There are several different binary operators:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3224
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3225</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3226
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3227<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3228<div class="doc_subsubsection">
3229 <a name="i_add">'<tt>add</tt>' Instruction</a>
3230</div>
3231
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3232<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3233
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3234<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3235<pre>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3236 &lt;result&gt; = add &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanb3b03482009-09-02 17:31:42 +0000[diff] [blame]3237 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3238 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3239 &lt;result&gt; = add nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3240</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3241
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3242<h5>Overview:</h5>
3243<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3244
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3245<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3246<p>The two arguments to the '<tt>add</tt>' instruction must
3247 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3248 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3249
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3250<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3251<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3252
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3253<p>If the sum has unsigned overflow, the result returned is the mathematical
3254 result modulo 2<sup>n</sup>, where n is the bit width of the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3255
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3256<p>Because LLVM integers use a two's complement representation, this instruction
3257 is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3258
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3259<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3260 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3261 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]3262 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3263 respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3264
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3265<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3266<pre>
3267 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3268</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3269
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3270</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3271
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3272<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3273<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3274 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
3275</div>
3276
3277<div class="doc_text">
3278
3279<h5>Syntax:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3280<pre>
3281 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3282</pre>
3283
3284<h5>Overview:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3285<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
3286
3287<h5>Arguments:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3288<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3289 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3290 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3291
3292<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3293<p>The value produced is the floating point sum of the two operands.</p>
3294
3295<h5>Example:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3296<pre>
3297 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
3298</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3299
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3300</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3301
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3302<!-- _______________________________________________________________________ -->
3303<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3304 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
3305</div>
3306
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3307<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3308
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3309<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3310<pre>
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3311 &lt;result&gt; = sub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanb3b03482009-09-02 17:31:42 +0000[diff] [blame]3312 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3313 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3314 &lt;result&gt; = sub nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3315</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3316
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3317<h5>Overview:</h5>
3318<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3319 operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3320
3321<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3322 '<tt>neg</tt>' instruction present in most other intermediate
3323 representations.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3324
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3325<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3326<p>The two arguments to the '<tt>sub</tt>' instruction must
3327 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3328 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3329
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3330<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3331<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3332
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3333<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3334 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3335 result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3336
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3337<p>Because LLVM integers use a two's complement representation, this instruction
3338 is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3339
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3340<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3341 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3342 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]3343 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3344 respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3345
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3346<h5>Example:</h5>
3347<pre>
3348 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3349 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3350</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3351
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3352</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3354<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3355<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3356 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3357</div>
3358
3359<div class="doc_text">
3360
3361<h5>Syntax:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3362<pre>
3363 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3364</pre>
3365
3366<h5>Overview:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3367<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3368 operands.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3369
3370<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3371 '<tt>fneg</tt>' instruction present in most other intermediate
3372 representations.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3373
3374<h5>Arguments:</h5>
Bill Wendling1a2630a2009-07-20 02:32:41 +0000[diff] [blame]3375<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3376 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3377 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3378
3379<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3380<p>The value produced is the floating point difference of the two operands.</p>
3381
3382<h5>Example:</h5>
3383<pre>
3384 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3385 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3386</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3387
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3388</div>
3389
3390<!-- _______________________________________________________________________ -->
3391<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3392 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3393</div>
3394
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3395<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3396
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3397<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3398<pre>
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3399 &lt;result&gt; = mul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanb3b03482009-09-02 17:31:42 +0000[diff] [blame]3400 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3401 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3402 &lt;result&gt; = mul nuw nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3403</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3404
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3405<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3406<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3408<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3409<p>The two arguments to the '<tt>mul</tt>' instruction must
3410 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3411 integer values. Both arguments must have identical types.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3412
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3413<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3414<p>The value produced is the integer product of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3415
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3416<p>If the result of the multiplication has unsigned overflow, the result
3417 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3418 width of the result.</p>
3419
3420<p>Because LLVM integers use a two's complement representation, and the result
3421 is the same width as the operands, this instruction returns the correct
3422 result for both signed and unsigned integers. If a full product
3423 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3424 be sign-extended or zero-extended as appropriate to the width of the full
3425 product.</p>
3426
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3427<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3428 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3429 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
Dan Gohmanb8ddf022010-04-22 23:14:21 +0000[diff] [blame]3430 is a <a href="#trapvalues">trap value</a> if unsigned and/or signed overflow,
3431 respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3432
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3433<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3434<pre>
3435 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3436</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3437
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3438</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3439
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3440<!-- _______________________________________________________________________ -->
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3441<div class="doc_subsubsection">
3442 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3443</div>
3444
3445<div class="doc_text">
3446
3447<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3448<pre>
3449 &lt;result&gt; = fmul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3450</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3451
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3452<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3453<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3454
3455<h5>Arguments:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3456<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3457 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3458 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3459
3460<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3461<p>The value produced is the floating point product of the two operands.</p>
3462
3463<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3464<pre>
3465 &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3466</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3467
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3468</div>
3469
3470<!-- _______________________________________________________________________ -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3471<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3472</a></div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3473
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3474<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3476<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3477<pre>
3478 &lt;result&gt; = udiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3479</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3480
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3481<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3482<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3483
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3484<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3485<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3486 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3487 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3489<h5>Semantics:</h5>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3490<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3491
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3492<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3493 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3494
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3495<p>Division by zero leads to undefined behavior.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3496
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3497<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3498<pre>
3499 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3500</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3501
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3502</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3503
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3504<!-- _______________________________________________________________________ -->
3505<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3506</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3508<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3509
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3510<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3511<pre>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3512 &lt;result&gt; = sdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanb3b03482009-09-02 17:31:42 +0000[diff] [blame]3513 &lt;result&gt; = sdiv exact &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3514</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3515
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3516<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3517<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3518
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3519<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3520<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3521 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3522 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3523
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3524<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3525<p>The value produced is the signed integer quotient of the two operands rounded
3526 towards zero.</p>
3527
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3528<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3529 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3530
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3531<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3532 undefined behavior; this is a rare case, but can occur, for example, by doing
3533 a 32-bit division of -2147483648 by -1.</p>
3534
Dan Gohman67fa48e2009-07-22 00:04:19 +0000[diff] [blame]3535<p>If the <tt>exact</tt> keyword is present, the result value of the
Dan Gohmanf7263862010-04-23 15:23:32 +0000[diff] [blame]3536 <tt>sdiv</tt> is a <a href="#trapvalues">trap value</a> if the result would
3537 be rounded or if overflow would occur.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3538
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3539<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3540<pre>
3541 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3542</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3543
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3544</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3545
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3546<!-- _______________________________________________________________________ -->
3547<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3548Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3549
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3550<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3551
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3552<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3553<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3554 &lt;result&gt; = fdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3555</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3556
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3557<h5>Overview:</h5>
3558<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3559
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3560<h5>Arguments:</h5>
3561<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3562 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3563 floating point values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3564
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3565<h5>Semantics:</h5>
3566<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3567
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3568<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3569<pre>
3570 &lt;result&gt; = fdiv float 4.0, %var <i>; yields {float}:result = 4.0 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3571</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3572
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3573</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3574
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3575<!-- _______________________________________________________________________ -->
3576<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3577</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3578
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3579<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3580
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3581<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3582<pre>
3583 &lt;result&gt; = urem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3584</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3585
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3586<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3587<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3588 division of its two arguments.</p>
3589
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3590<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3591<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3592 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3593 values. Both arguments must have identical types.</p>
3594
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3595<h5>Semantics:</h5>
3596<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3597 This instruction always performs an unsigned division to get the
3598 remainder.</p>
3599
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3600<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3601 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3602
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3603<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3604
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3605<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3606<pre>
3607 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3608</pre>
3609
3610</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3611
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3612<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3613<div class="doc_subsubsection">
3614 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3615</div>
3616
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3617<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3618
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3619<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3620<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3621 &lt;result&gt; = srem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3622</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3623
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3624<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3625<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3626 division of its two operands. This instruction can also take
3627 <a href="#t_vector">vector</a> versions of the values in which case the
3628 elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]3629
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3630<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3631<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3632 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3633 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3634
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3635<h5>Semantics:</h5>
3636<p>This instruction returns the <i>remainder</i> of a division (where the result
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3637 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3638 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3639 a value. For more information about the difference,
3640 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3641 Math Forum</a>. For a table of how this is implemented in various languages,
3642 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3643 Wikipedia: modulo operation</a>.</p>
3644
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3645<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3646 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3647
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3648<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3649 Overflow also leads to undefined behavior; this is a rare case, but can
3650 occur, for example, by taking the remainder of a 32-bit division of
3651 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3652 lets srem be implemented using instructions that return both the result of
3653 the division and the remainder.)</p>
3654
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3655<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3656<pre>
3657 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3658</pre>
3659
3660</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3661
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3662<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3663<div class="doc_subsubsection">
3664 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3665
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3666<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3667
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3668<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3669<pre>
3670 &lt;result&gt; = frem &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3671</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3672
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3673<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3674<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3675 its two operands.</p>
3676
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3677<h5>Arguments:</h5>
3678<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3679 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3680 floating point values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3681
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3682<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3683<p>This instruction returns the <i>remainder</i> of a division. The remainder
3684 has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3685
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3686<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3687<pre>
3688 &lt;result&gt; = frem float 4.0, %var <i>; yields {float}:result = 4.0 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3689</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3690
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3691</div>
3692
3693<!-- ======================================================================= -->
3694<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3695Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3696
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3697<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3698
3699<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3700 program. They are generally very efficient instructions and can commonly be
3701 strength reduced from other instructions. They require two operands of the
3702 same type, execute an operation on them, and produce a single value. The
3703 resulting value is the same type as its operands.</p>
3704
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3705</div>
3706
3707<!-- _______________________________________________________________________ -->
3708<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3709Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3710
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3711<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3712
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3713<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3714<pre>
3715 &lt;result&gt; = shl &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3716</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3717
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3718<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3719<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3720 a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3722<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3723<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3724 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3725 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3726
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3727<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3728<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3729 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3730 is (statically or dynamically) negative or equal to or larger than the number
3731 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3732 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3733 shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3734
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3735<h5>Example:</h5>
3736<pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3737 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3738 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3739 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3740 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3741 &lt;result&gt; = shl &lt;2 x i32&gt; &lt; i32 1, i32 1&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 2, i32 4&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3742</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3743
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3744</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3745
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3746<!-- _______________________________________________________________________ -->
3747<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3748Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3749
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3750<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3751
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3752<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3753<pre>
3754 &lt;result&gt; = lshr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3755</pre>
3756
3757<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3758<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3759 operand shifted to the right a specified number of bits with zero fill.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3760
3761<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3762<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3763 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3764 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3765
3766<h5>Semantics:</h5>
3767<p>This instruction always performs a logical shift right operation. The most
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3768 significant bits of the result will be filled with zero bits after the shift.
3769 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3770 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3771 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3772 shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3773
3774<h5>Example:</h5>
3775<pre>
3776 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3777 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3778 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3779 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3780 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3781 &lt;result&gt; = lshr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 2&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 0x7FFFFFFF, i32 1&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3782</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3783
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3784</div>
3785
3786<!-- _______________________________________________________________________ -->
3787<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3788Instruction</a> </div>
3789<div class="doc_text">
3790
3791<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3792<pre>
3793 &lt;result&gt; = ashr &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3794</pre>
3795
3796<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3797<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3798 operand shifted to the right a specified number of bits with sign
3799 extension.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3800
3801<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3802<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3803 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3804 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3805
3806<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3807<p>This instruction always performs an arithmetic shift right operation, The
3808 most significant bits of the result will be filled with the sign bit
3809 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3810 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3811 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3812 the corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3813
3814<h5>Example:</h5>
3815<pre>
3816 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3817 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3818 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3819 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3820 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3821 &lt;result&gt; = ashr &lt;2 x i32&gt; &lt; i32 -2, i32 4&gt;, &lt; i32 1, i32 3&gt; <i>; yields: result=&lt;2 x i32&gt; &lt; i32 -1, i32 0&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3822</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3823
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3824</div>
3825
3826<!-- _______________________________________________________________________ -->
3827<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3828Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3829
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3830<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3831
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3832<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3833<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3834 &lt;result&gt; = and &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3835</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3836
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3837<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3838<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3839 operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3840
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3841<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3842<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3843 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3844 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3845
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3846<h5>Semantics:</h5>
3847<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3848
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3849<table border="1" cellspacing="0" cellpadding="4">
3850 <tbody>
3851 <tr>
3852 <td>In0</td>
3853 <td>In1</td>
3854 <td>Out</td>
3855 </tr>
3856 <tr>
3857 <td>0</td>
3858 <td>0</td>
3859 <td>0</td>
3860 </tr>
3861 <tr>
3862 <td>0</td>
3863 <td>1</td>
3864 <td>0</td>
3865 </tr>
3866 <tr>
3867 <td>1</td>
3868 <td>0</td>
3869 <td>0</td>
3870 </tr>
3871 <tr>
3872 <td>1</td>
3873 <td>1</td>
3874 <td>1</td>
3875 </tr>
3876 </tbody>
3877</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3878
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3879<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3880<pre>
3881 &lt;result&gt; = and i32 4, %var <i>; yields {i32}:result = 4 &amp; %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3882 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3883 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3884</pre>
3885</div>
3886<!-- _______________________________________________________________________ -->
3887<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3888
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3889<div class="doc_text">
3890
3891<h5>Syntax:</h5>
3892<pre>
3893 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3894</pre>
3895
3896<h5>Overview:</h5>
3897<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3898 two operands.</p>
3899
3900<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3901<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3902 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3903 values. Both arguments must have identical types.</p>
3904
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3905<h5>Semantics:</h5>
3906<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3907
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3908<table border="1" cellspacing="0" cellpadding="4">
3909 <tbody>
3910 <tr>
3911 <td>In0</td>
3912 <td>In1</td>
3913 <td>Out</td>
3914 </tr>
3915 <tr>
3916 <td>0</td>
3917 <td>0</td>
3918 <td>0</td>
3919 </tr>
3920 <tr>
3921 <td>0</td>
3922 <td>1</td>
3923 <td>1</td>
3924 </tr>
3925 <tr>
3926 <td>1</td>
3927 <td>0</td>
3928 <td>1</td>
3929 </tr>
3930 <tr>
3931 <td>1</td>
3932 <td>1</td>
3933 <td>1</td>
3934 </tr>
3935 </tbody>
3936</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3937
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3938<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3939<pre>
3940 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3941 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3942 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3943</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3944
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3945</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3946
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3947<!-- _______________________________________________________________________ -->
3948<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3949Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3950
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3951<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3952
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3953<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3954<pre>
3955 &lt;result&gt; = xor &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3956</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3957
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3958<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3959<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3960 its two operands. The <tt>xor</tt> is used to implement the "one's
3961 complement" operation, which is the "~" operator in C.</p>
3962
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3963<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3964<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3965 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3966 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3967
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3968<h5>Semantics:</h5>
3969<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3970
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3971<table border="1" cellspacing="0" cellpadding="4">
3972 <tbody>
3973 <tr>
3974 <td>In0</td>
3975 <td>In1</td>
3976 <td>Out</td>
3977 </tr>
3978 <tr>
3979 <td>0</td>
3980 <td>0</td>
3981 <td>0</td>
3982 </tr>
3983 <tr>
3984 <td>0</td>
3985 <td>1</td>
3986 <td>1</td>
3987 </tr>
3988 <tr>
3989 <td>1</td>
3990 <td>0</td>
3991 <td>1</td>
3992 </tr>
3993 <tr>
3994 <td>1</td>
3995 <td>1</td>
3996 <td>0</td>
3997 </tr>
3998 </tbody>
3999</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4000
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4001<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4002<pre>
4003 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4004 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
4005 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
4006 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
4007</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4008
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4009</div>
4010
4011<!-- ======================================================================= -->
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4012<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4013 <a name="vectorops">Vector Operations</a>
4014</div>
4015
4016<div class="doc_text">
4017
4018<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4019 target-independent manner. These instructions cover the element-access and
4020 vector-specific operations needed to process vectors effectively. While LLVM
4021 does directly support these vector operations, many sophisticated algorithms
4022 will want to use target-specific intrinsics to take full advantage of a
4023 specific target.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4024
4025</div>
4026
4027<!-- _______________________________________________________________________ -->
4028<div class="doc_subsubsection">
4029 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
4030</div>
4031
4032<div class="doc_text">
4033
4034<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4035<pre>
4036 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
4037</pre>
4038
4039<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4040<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
4041 from a vector at a specified index.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4042
4043
4044<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4045<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
4046 of <a href="#t_vector">vector</a> type. The second operand is an index
4047 indicating the position from which to extract the element. The index may be
4048 a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4049
4050<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4051<p>The result is a scalar of the same type as the element type of
4052 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
4053 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4054 results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4055
4056<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4057<pre>
Gabor Greifb2c73142009-10-28 13:14:50 +0000[diff] [blame]4058 &lt;result&gt; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4059</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4060
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4061</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4062
4063<!-- _______________________________________________________________________ -->
4064<div class="doc_subsubsection">
4065 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
4066</div>
4067
4068<div class="doc_text">
4069
4070<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4071<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]4072 &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>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4073</pre>
4074
4075<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4076<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
4077 vector at a specified index.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4078
4079<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4080<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
4081 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
4082 whose type must equal the element type of the first operand. The third
4083 operand is an index indicating the position at which to insert the value.
4084 The index may be a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4085
4086<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4087<p>The result is a vector of the same type as <tt>val</tt>. Its element values
4088 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
4089 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
4090 results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4091
4092<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4093<pre>
Gabor Greifb2c73142009-10-28 13:14:50 +0000[diff] [blame]4094 &lt;result&gt; = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4095</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4096
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4097</div>
4098
4099<!-- _______________________________________________________________________ -->
4100<div class="doc_subsubsection">
4101 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
4102</div>
4103
4104<div class="doc_text">
4105
4106<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4107<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]4108 &lt;result&gt; = shufflevector &lt;n x &lt;ty&gt;&gt; &lt;v1&gt;, &lt;n x &lt;ty&gt;&gt; &lt;v2&gt;, &lt;m x i32&gt; &lt;mask&gt; <i>; yields &lt;m x &lt;ty&gt;&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4109</pre>
4110
4111<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4112<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
4113 from two input vectors, returning a vector with the same element type as the
4114 input and length that is the same as the shuffle mask.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4115
4116<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4117<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
4118 with types that match each other. The third argument is a shuffle mask whose
4119 element type is always 'i32'. The result of the instruction is a vector
4120 whose length is the same as the shuffle mask and whose element type is the
4121 same as the element type of the first two operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4122
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4123<p>The shuffle mask operand is required to be a constant vector with either
4124 constant integer or undef values.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4125
4126<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4127<p>The elements of the two input vectors are numbered from left to right across
4128 both of the vectors. The shuffle mask operand specifies, for each element of
4129 the result vector, which element of the two input vectors the result element
4130 gets. The element selector may be undef (meaning "don't care") and the
4131 second operand may be undef if performing a shuffle from only one vector.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4132
4133<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4134<pre>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4135 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4136 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4137 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4138 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4139 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]4140 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4141 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]4142 &lt;8 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7 &gt; <i>; yields &lt;8 x i32&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4143</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4144
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4145</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4146
4147<!-- ======================================================================= -->
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4148<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4149 <a name="aggregateops">Aggregate Operations</a>
4150</div>
4151
4152<div class="doc_text">
4153
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4154<p>LLVM supports several instructions for working with
4155 <a href="#t_aggregate">aggregate</a> values.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4156
4157</div>
4158
4159<!-- _______________________________________________________________________ -->
4160<div class="doc_subsubsection">
4161 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
4162</div>
4163
4164<div class="doc_text">
4165
4166<h5>Syntax:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4167<pre>
4168 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
4169</pre>
4170
4171<h5>Overview:</h5>
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4172<p>The '<tt>extractvalue</tt>' instruction extracts the value of a member field
4173 from an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4174
4175<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4176<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4177 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4178 <a href="#t_array">array</a> type. The operands are constant indices to
4179 specify which value to extract in a similar manner as indices in a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4180 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4181
4182<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4183<p>The result is the value at the position in the aggregate specified by the
4184 index operands.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4185
4186<h5>Example:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4187<pre>
Gabor Greifb2c73142009-10-28 13:14:50 +0000[diff] [blame]4188 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4189</pre>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4190
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4191</div>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4192
4193<!-- _______________________________________________________________________ -->
4194<div class="doc_subsubsection">
4195 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
4196</div>
4197
4198<div class="doc_text">
4199
4200<h5>Syntax:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4201<pre>
Jeffrey Yasskine05d7732010-01-11 19:19:26 +0000[diff] [blame]4202 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, &lt;idx&gt; <i>; yields &lt;aggregate type&gt;</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4203</pre>
4204
4205<h5>Overview:</h5>
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4206<p>The '<tt>insertvalue</tt>' instruction inserts a value into a member field
4207 in an <a href="#t_aggregate">aggregate</a> value.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4208
4209<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4210<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4211 of <a href="#t_struct">struct</a>, <a href="#t_union">union</a> or
4212 <a href="#t_array">array</a> type. The second operand is a first-class
4213 value to insert. The following operands are constant indices indicating
4214 the position at which to insert the value in a similar manner as indices in a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4215 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
4216 value to insert must have the same type as the value identified by the
4217 indices.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4218
4219<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4220<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
4221 that of <tt>val</tt> except that the value at the position specified by the
4222 indices is that of <tt>elt</tt>.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4223
4224<h5>Example:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4225<pre>
Jeffrey Yasskine05d7732010-01-11 19:19:26 +0000[diff] [blame]4226 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
4227 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4228</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4229
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]4230</div>
4231
4232
4233<!-- ======================================================================= -->
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4234<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4235 <a name="memoryops">Memory Access and Addressing Operations</a>
4236</div>
4237
4238<div class="doc_text">
4239
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4240<p>A key design point of an SSA-based representation is how it represents
4241 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]4242 very simple. This section describes how to read, write, and allocate
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4243 memory in LLVM.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4244
4245</div>
4246
4247<!-- _______________________________________________________________________ -->
4248<div class="doc_subsubsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4249 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
4250</div>
4251
4252<div class="doc_text">
4253
4254<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4255<pre>
Dan Gohman3eb67d52010-05-28 01:14:11 +0000[diff] [blame]4256 &lt;result&gt; = alloca &lt;type&gt;[, &lt;ty&gt; &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4257</pre>
4258
4259<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4260<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4261 currently executing function, to be automatically released when this function
4262 returns to its caller. The object is always allocated in the generic address
4263 space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4264
4265<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4266<p>The '<tt>alloca</tt>' instruction
4267 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
4268 runtime stack, returning a pointer of the appropriate type to the program.
4269 If "NumElements" is specified, it is the number of elements allocated,
4270 otherwise "NumElements" is defaulted to be one. If a constant alignment is
4271 specified, the value result of the allocation is guaranteed to be aligned to
4272 at least that boundary. If not specified, or if zero, the target can choose
4273 to align the allocation on any convenient boundary compatible with the
4274 type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4275
4276<p>'<tt>type</tt>' may be any sized type.</p>
4277
4278<h5>Semantics:</h5>
Bill Wendling2a454572009-05-08 20:49:29 +0000[diff] [blame]4279<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4280 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
4281 memory is automatically released when the function returns. The
4282 '<tt>alloca</tt>' instruction is commonly used to represent automatic
4283 variables that must have an address available. When the function returns
4284 (either with the <tt><a href="#i_ret">ret</a></tt>
4285 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
4286 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4287
4288<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4289<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]4290 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
4291 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
4292 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
4293 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4294</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4295
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4296</div>
4297
4298<!-- _______________________________________________________________________ -->
4299<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
4300Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4301
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4302<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4303
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4304<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4305<pre>
Bill Wendling4197e452010-02-25 21:23:24 +0000[diff] [blame]4306 &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4307 &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !&lt;index&gt;]
4308 !&lt;index&gt; = !{ i32 1 }
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4309</pre>
4310
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4311<h5>Overview:</h5>
4312<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4313
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4314<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4315<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
4316 from which to load. The pointer must point to
4317 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
4318 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]4319 number or order of execution of this <tt>load</tt> with other <a
4320 href="#volatile">volatile operations</a>.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4321
Bill Wendling4197e452010-02-25 21:23:24 +0000[diff] [blame]4322<p>The optional constant <tt>align</tt> argument specifies the alignment of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4323 operation (that is, the alignment of the memory address). A value of 0 or an
Bill Wendling4197e452010-02-25 21:23:24 +0000[diff] [blame]4324 omitted <tt>align</tt> argument means that the operation has the preferential
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4325 alignment for the target. It is the responsibility of the code emitter to
4326 ensure that the alignment information is correct. Overestimating the
Bill Wendling4197e452010-02-25 21:23:24 +0000[diff] [blame]4327 alignment results in undefined behavior. Underestimating the alignment may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4328 produce less efficient code. An alignment of 1 is always safe.</p>
4329
Bill Wendling4197e452010-02-25 21:23:24 +0000[diff] [blame]4330<p>The optional <tt>!nontemporal</tt> metadata must reference a single
4331 metatadata name &lt;index&gt; corresponding to a metadata node with
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]4332 one <tt>i32</tt> entry of value 1. The existence of
Bill Wendling4197e452010-02-25 21:23:24 +0000[diff] [blame]4333 the <tt>!nontemporal</tt> metatadata on the instruction tells the optimizer
4334 and code generator that this load is not expected to be reused in the cache.
4335 The code generator may select special instructions to save cache bandwidth,
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]4336 such as the <tt>MOVNT</tt> instruction on x86.</p>
David Greene02dfe202010-02-16 20:50:18 +0000[diff] [blame]4337
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4338<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4339<p>The location of memory pointed to is loaded. If the value being loaded is of
4340 scalar type then the number of bytes read does not exceed the minimum number
4341 of bytes needed to hold all bits of the type. For example, loading an
4342 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4343 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4344 is undefined if the value was not originally written using a store of the
4345 same type.</p>
4346
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4347<h5>Examples:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4348<pre>
4349 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4350 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4351 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4352</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4354</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4355
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4356<!-- _______________________________________________________________________ -->
4357<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4358Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4359
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4360<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4361
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4362<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4363<pre>
David Greene02dfe202010-02-16 20:50:18 +0000[diff] [blame]4364 store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !<index>] <i>; yields {void}</i>
4365 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;][, !nontemporal !<index>] <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4366</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4367
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4368<h5>Overview:</h5>
4369<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4370
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4371<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4372<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4373 and an address at which to store it. The type of the
4374 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4375 the <a href="#t_firstclass">first class</a> type of the
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]4376 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked as
4377 <tt>volatile</tt>, then the optimizer is not allowed to modify the number or
4378 order of execution of this <tt>store</tt> with other <a
4379 href="#volatile">volatile operations</a>.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4380
4381<p>The optional constant "align" argument specifies the alignment of the
4382 operation (that is, the alignment of the memory address). A value of 0 or an
4383 omitted "align" argument means that the operation has the preferential
4384 alignment for the target. It is the responsibility of the code emitter to
4385 ensure that the alignment information is correct. Overestimating the
4386 alignment results in an undefined behavior. Underestimating the alignment may
4387 produce less efficient code. An alignment of 1 is always safe.</p>
4388
David Greene02dfe202010-02-16 20:50:18 +0000[diff] [blame]4389<p>The optional !nontemporal metadata must reference a single metatadata
4390 name <index> corresponding to a metadata node with one i32 entry of
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]4391 value 1. The existence of the !nontemporal metatadata on the
David Greene02dfe202010-02-16 20:50:18 +0000[diff] [blame]4392 instruction tells the optimizer and code generator that this load is
4393 not expected to be reused in the cache. The code generator may
4394 select special instructions to save cache bandwidth, such as the
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]4395 MOVNT instruction on x86.</p>
David Greene02dfe202010-02-16 20:50:18 +0000[diff] [blame]4396
4397
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4398<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4399<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4400 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4401 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4402 does not exceed the minimum number of bytes needed to hold all bits of the
4403 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4404 writing a value of a type like <tt>i20</tt> with a size that is not an
4405 integral number of bytes, it is unspecified what happens to the extra bits
4406 that do not belong to the type, but they will typically be overwritten.</p>
4407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4408<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4409<pre>
4410 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]4411 store i32 3, i32* %ptr <i>; yields {void}</i>
4412 %val = <a href="#i_load">load</a> i32* %ptr <i>; yields {i32}:val = i32 3</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4413</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4414
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4415</div>
4416
4417<!-- _______________________________________________________________________ -->
4418<div class="doc_subsubsection">
4419 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4420</div>
4421
4422<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4423
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4424<h5>Syntax:</h5>
4425<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4426 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4427 &lt;result&gt; = getelementptr inbounds &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4428</pre>
4429
4430<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4431<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4432 subelement of an <a href="#t_aggregate">aggregate</a> data structure.
4433 It performs address calculation only and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4434
4435<h5>Arguments:</h5>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4436<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4437 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4438 elements of the aggregate object are indexed. The interpretation of each
4439 index is dependent on the type being indexed into. The first index always
4440 indexes the pointer value given as the first argument, the second index
4441 indexes a value of the type pointed to (not necessarily the value directly
4442 pointed to, since the first index can be non-zero), etc. The first type
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4443 indexed into must be a pointer value, subsequent types can be arrays,
4444 vectors, structs and unions. Note that subsequent types being indexed into
4445 can never be pointers, since that would require loading the pointer before
4446 continuing calculation.</p>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4447
4448<p>The type of each index argument depends on the type it is indexing into.
Chris Lattnerd5d51722010-02-12 20:49:41 +0000[diff] [blame]4449 When indexing into a (optionally packed) structure or union, only <tt>i32</tt>
4450 integer <b>constants</b> are allowed. When indexing into an array, pointer
4451 or vector, integers of any width are allowed, and they are not required to be
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4452 constant.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4453
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4454<p>For example, let's consider a C code fragment and how it gets compiled to
4455 LLVM:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4456
4457<div class="doc_code">
4458<pre>
4459struct RT {
4460 char A;
4461 int B[10][20];
4462 char C;
4463};
4464struct ST {
4465 int X;
4466 double Y;
4467 struct RT Z;
4468};
4469
4470int *foo(struct ST *s) {
4471 return &amp;s[1].Z.B[5][13];
4472}
4473</pre>
4474</div>
4475
4476<p>The LLVM code generated by the GCC frontend is:</p>
4477
4478<div class="doc_code">
4479<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]4480%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4481%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4482
Dan Gohman47360842009-07-25 02:23:48 +0000[diff] [blame]4483define i32* @foo(%ST* %s) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4484entry:
4485 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4486 ret i32* %reg
4487}
4488</pre>
4489</div>
4490
4491<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4492<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4493 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4494 }</tt>' type, a structure. The second index indexes into the third element
4495 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4496 i8 }</tt>' type, another structure. The third index indexes into the second
4497 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4498 array. The two dimensions of the array are subscripted into, yielding an
4499 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4500 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4501
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4502<p>Note that it is perfectly legal to index partially through a structure,
4503 returning a pointer to an inner element. Because of this, the LLVM code for
4504 the given testcase is equivalent to:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4505
4506<pre>
Dan Gohman47360842009-07-25 02:23:48 +0000[diff] [blame]4507 define i32* @foo(%ST* %s) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4508 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4509 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4510 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4511 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4512 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4513 ret i32* %t5
4514 }
4515</pre>
4516
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4517<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohmanf7263862010-04-23 15:23:32 +0000[diff] [blame]4518 <tt>getelementptr</tt> is a <a href="#trapvalues">trap value</a> if the
4519 base pointer is not an <i>in bounds</i> address of an allocated object,
4520 or if any of the addresses that would be formed by successive addition of
4521 the offsets implied by the indices to the base address with infinitely
4522 precise arithmetic are not an <i>in bounds</i> address of that allocated
4523 object. The <i>in bounds</i> addresses for an allocated object are all
4524 the addresses that point into the object, plus the address one byte past
4525 the end.</p>
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4526
4527<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4528 the base address with silently-wrapping two's complement arithmetic, and
4529 the result value of the <tt>getelementptr</tt> may be outside the object
4530 pointed to by the base pointer. The result value may not necessarily be
4531 used to access memory though, even if it happens to point into allocated
4532 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4533 section for more information.</p>
4534
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4535<p>The getelementptr instruction is often confusing. For some more insight into
4536 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4537
4538<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4539<pre>
4540 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4541 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4542 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4543 %vptr = getelementptr {i32, &lt;2 x i8&gt;}* %svptr, i64 0, i32 1, i32 1
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4544 <i>; yields i8*:eptr</i>
4545 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta4f9a0dc2009-04-25 07:27:44 +0000[diff] [blame]4546 <i>; yields i32*:iptr</i>
Sanjiv Gupta1e46c582009-04-24 16:38:13 +0000[diff] [blame]4547 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4548</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4549
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4550</div>
4551
4552<!-- ======================================================================= -->
4553<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4554</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4555
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4556<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4557
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4558<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4559 which all take a single operand and a type. They perform various bit
4560 conversions on the operand.</p>
4561
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4562</div>
4563
4564<!-- _______________________________________________________________________ -->
4565<div class="doc_subsubsection">
4566 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4567</div>
4568<div class="doc_text">
4569
4570<h5>Syntax:</h5>
4571<pre>
4572 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4573</pre>
4574
4575<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4576<p>The '<tt>trunc</tt>' instruction truncates its operand to the
4577 type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4578
4579<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4580<p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4581 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4582 size and type of the result, which must be
4583 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4584 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4585 allowed.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4586
4587<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4588<p>The '<tt>trunc</tt>' instruction truncates the high order bits
4589 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4590 source size must be larger than the destination size, <tt>trunc</tt> cannot
4591 be a <i>no-op cast</i>. It will always truncate bits.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4592
4593<h5>Example:</h5>
4594<pre>
4595 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4596 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4597 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4598</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4599
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4600</div>
4601
4602<!-- _______________________________________________________________________ -->
4603<div class="doc_subsubsection">
4604 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4605</div>
4606<div class="doc_text">
4607
4608<h5>Syntax:</h5>
4609<pre>
4610 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4611</pre>
4612
4613<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4614<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4615 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4616
4617
4618<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4619<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4620 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4621 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4622 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4623 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4624
4625<h5>Semantics:</h5>
4626<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4627 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4628
4629<p>When zero extending from i1, the result will always be either 0 or 1.</p>
4630
4631<h5>Example:</h5>
4632<pre>
4633 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4634 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4635</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4636
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4637</div>
4638
4639<!-- _______________________________________________________________________ -->
4640<div class="doc_subsubsection">
4641 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4642</div>
4643<div class="doc_text">
4644
4645<h5>Syntax:</h5>
4646<pre>
4647 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4648</pre>
4649
4650<h5>Overview:</h5>
4651<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4652
4653<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4654<p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4655 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4656 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4657 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4658 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4659
4660<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4661<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4662 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4663 of the type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4664
4665<p>When sign extending from i1, the extension always results in -1 or 0.</p>
4666
4667<h5>Example:</h5>
4668<pre>
4669 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4670 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4671</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4672
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4673</div>
4674
4675<!-- _______________________________________________________________________ -->
4676<div class="doc_subsubsection">
4677 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4678</div>
4679
4680<div class="doc_text">
4681
4682<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4683<pre>
4684 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4685</pre>
4686
4687<h5>Overview:</h5>
4688<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4689 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4690
4691<h5>Arguments:</h5>
4692<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4693 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4694 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4695 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4696 <i>no-op cast</i>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4697
4698<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4699<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4700 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4701 <a href="#t_floating">floating point</a> type. If the value cannot fit
4702 within the destination type, <tt>ty2</tt>, then the results are
4703 undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4704
4705<h5>Example:</h5>
4706<pre>
4707 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4708 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4709</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4710
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4711</div>
4712
4713<!-- _______________________________________________________________________ -->
4714<div class="doc_subsubsection">
4715 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4716</div>
4717<div class="doc_text">
4718
4719<h5>Syntax:</h5>
4720<pre>
4721 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4722</pre>
4723
4724<h5>Overview:</h5>
4725<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4726 floating point value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4727
4728<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4729<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4730 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4731 a <a href="#t_floating">floating point</a> type to cast it to. The source
4732 type must be smaller than the destination type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4733
4734<h5>Semantics:</h5>
4735<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4736 <a href="#t_floating">floating point</a> type to a larger
4737 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4738 used to make a <i>no-op cast</i> because it always changes bits. Use
4739 <tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4740
4741<h5>Example:</h5>
4742<pre>
4743 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4744 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4745</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4746
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4747</div>
4748
4749<!-- _______________________________________________________________________ -->
4750<div class="doc_subsubsection">
4751 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4752</div>
4753<div class="doc_text">
4754
4755<h5>Syntax:</h5>
4756<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4757 &lt;result&gt; = fptoui &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4758</pre>
4759
4760<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4761<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4762 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4763
4764<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4765<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4766 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4767 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4768 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4769 vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4770
4771<h5>Semantics:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4772<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4773 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4774 towards zero) unsigned integer value. If the value cannot fit
4775 in <tt>ty2</tt>, the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4776
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4777<h5>Example:</h5>
4778<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4779 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4780 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4781 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4782</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4783
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4784</div>
4785
4786<!-- _______________________________________________________________________ -->
4787<div class="doc_subsubsection">
4788 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4789</div>
4790<div class="doc_text">
4791
4792<h5>Syntax:</h5>
4793<pre>
4794 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4795</pre>
4796
4797<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4798<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4799 <a href="#t_floating">floating point</a> <tt>value</tt> to
4800 type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4801
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4802<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4803<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4804 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4805 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4806 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4807 vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4808
4809<h5>Semantics:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4810<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4811 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4812 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4813 the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4814
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4815<h5>Example:</h5>
4816<pre>
4817 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4818 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4819 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4820</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4821
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4822</div>
4823
4824<!-- _______________________________________________________________________ -->
4825<div class="doc_subsubsection">
4826 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4827</div>
4828<div class="doc_text">
4829
4830<h5>Syntax:</h5>
4831<pre>
4832 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4833</pre>
4834
4835<h5>Overview:</h5>
4836<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4837 integer and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4838
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4839<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4840<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4841 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4842 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4843 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4844 floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4845
4846<h5>Semantics:</h5>
4847<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4848 integer quantity and converts it to the corresponding floating point
4849 value. If the value cannot fit in the floating point value, the results are
4850 undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4851
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4852<h5>Example:</h5>
4853<pre>
4854 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4855 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4856</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4857
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4858</div>
4859
4860<!-- _______________________________________________________________________ -->
4861<div class="doc_subsubsection">
4862 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4863</div>
4864<div class="doc_text">
4865
4866<h5>Syntax:</h5>
4867<pre>
4868 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4869</pre>
4870
4871<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4872<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4873 and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4874
4875<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4876<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4877 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4878 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4879 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4880 floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4881
4882<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4883<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4884 quantity and converts it to the corresponding floating point value. If the
4885 value cannot fit in the floating point value, the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4886
4887<h5>Example:</h5>
4888<pre>
4889 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4890 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4891</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4892
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4893</div>
4894
4895<!-- _______________________________________________________________________ -->
4896<div class="doc_subsubsection">
4897 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4898</div>
4899<div class="doc_text">
4900
4901<h5>Syntax:</h5>
4902<pre>
4903 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4904</pre>
4905
4906<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4907<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4908 the integer type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4909
4910<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4911<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4912 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4913 <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4914
4915<h5>Semantics:</h5>
4916<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4917 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4918 truncating or zero extending that value to the size of the integer type. If
4919 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4920 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4921 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4922 change.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4923
4924<h5>Example:</h5>
4925<pre>
4926 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4927 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4928</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4929
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4930</div>
4931
4932<!-- _______________________________________________________________________ -->
4933<div class="doc_subsubsection">
4934 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4935</div>
4936<div class="doc_text">
4937
4938<h5>Syntax:</h5>
4939<pre>
4940 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4941</pre>
4942
4943<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4944<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4945 pointer type, <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4946
4947<h5>Arguments:</h5>
4948<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4949 value to cast, and a type to cast it to, which must be a
4950 <a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4951
4952<h5>Semantics:</h5>
4953<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4954 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4955 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4956 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4957 than the size of a pointer then a zero extension is done. If they are the
4958 same size, nothing is done (<i>no-op cast</i>).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4959
4960<h5>Example:</h5>
4961<pre>
4962 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4963 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4964 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4965</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4966
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4967</div>
4968
4969<!-- _______________________________________________________________________ -->
4970<div class="doc_subsubsection">
4971 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4972</div>
4973<div class="doc_text">
4974
4975<h5>Syntax:</h5>
4976<pre>
4977 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4978</pre>
4979
4980<h5>Overview:</h5>
4981<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4982 <tt>ty2</tt> without changing any bits.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4983
4984<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4985<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4986 non-aggregate first class value, and a type to cast it to, which must also be
4987 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4988 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4989 identical. If the source type is a pointer, the destination type must also be
4990 a pointer. This instruction supports bitwise conversion of vectors to
4991 integers and to vectors of other types (as long as they have the same
4992 size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4993
4994<h5>Semantics:</h5>
4995<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4996 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4997 this conversion. The conversion is done as if the <tt>value</tt> had been
4998 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4999 be converted to other pointer types with this instruction. To convert
5000 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
5001 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5002
5003<h5>Example:</h5>
5004<pre>
5005 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
5006 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5007 %Z = bitcast &lt;2 x int&gt; %V to i64; <i>; yields i64: %V</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5008</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5009
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5010</div>
5011
5012<!-- ======================================================================= -->
5013<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5014
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5015<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5016
5017<p>The instructions in this category are the "miscellaneous" instructions, which
5018 defy better classification.</p>
5019
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5020</div>
5021
5022<!-- _______________________________________________________________________ -->
5023<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
5024</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5025
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5026<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5027
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5028<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5029<pre>
5030 &lt;result&gt; = icmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5031</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5032
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5033<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5034<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
5035 boolean values based on comparison of its two integer, integer vector, or
5036 pointer operands.</p>
5037
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5038<h5>Arguments:</h5>
5039<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5040 the condition code indicating the kind of comparison to perform. It is not a
5041 value, just a keyword. The possible condition code are:</p>
5042
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5043<ol>
5044 <li><tt>eq</tt>: equal</li>
5045 <li><tt>ne</tt>: not equal </li>
5046 <li><tt>ugt</tt>: unsigned greater than</li>
5047 <li><tt>uge</tt>: unsigned greater or equal</li>
5048 <li><tt>ult</tt>: unsigned less than</li>
5049 <li><tt>ule</tt>: unsigned less or equal</li>
5050 <li><tt>sgt</tt>: signed greater than</li>
5051 <li><tt>sge</tt>: signed greater or equal</li>
5052 <li><tt>slt</tt>: signed less than</li>
5053 <li><tt>sle</tt>: signed less or equal</li>
5054</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5055
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5056<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5057 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
5058 typed. They must also be identical types.</p>
5059
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5060<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5061<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
5062 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]5063 either an <a href="#t_integer"><tt>i1</tt></a> or vector of <tt>i1</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5064 result, as follows:</p>
5065
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5066<ol>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5067 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5068 <tt>false</tt> otherwise. No sign interpretation is necessary or
5069 performed.</li>
5070
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5071 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5072 <tt>false</tt> otherwise. No sign interpretation is necessary or
5073 performed.</li>
5074
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5075 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5076 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5077
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5078 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5079 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5080 to <tt>op2</tt>.</li>
5081
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5082 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5083 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5084
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5085 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5086 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5087
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5088 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5089 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5090
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5091 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5092 <tt>true</tt> if <tt>op1</tt> is greater than or equal
5093 to <tt>op2</tt>.</li>
5094
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5095 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5096 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
5097
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5098 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5099 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5100</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5101
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5102<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5103 values are compared as if they were integers.</p>
5104
5105<p>If the operands are integer vectors, then they are compared element by
5106 element. The result is an <tt>i1</tt> vector with the same number of elements
5107 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5108
5109<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5110<pre>
5111 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5112 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
5113 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
5114 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
5115 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
5116 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
5117</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]5118
5119<p>Note that the code generator does not yet support vector types with
5120 the <tt>icmp</tt> instruction.</p>
5121
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5122</div>
5123
5124<!-- _______________________________________________________________________ -->
5125<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
5126</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5127
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5128<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5129
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5130<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5131<pre>
5132 &lt;result&gt; = fcmp &lt;cond&gt; &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {i1} or {&lt;N x i1&gt;}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5133</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5134
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5135<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5136<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
5137 values based on comparison of its operands.</p>
5138
5139<p>If the operands are floating point scalars, then the result type is a boolean
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]5140(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5141
5142<p>If the operands are floating point vectors, then the result type is a vector
5143 of boolean with the same number of elements as the operands being
5144 compared.</p>
5145
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5146<h5>Arguments:</h5>
5147<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5148 the condition code indicating the kind of comparison to perform. It is not a
5149 value, just a keyword. The possible condition code are:</p>
5150
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5151<ol>
5152 <li><tt>false</tt>: no comparison, always returns false</li>
5153 <li><tt>oeq</tt>: ordered and equal</li>
5154 <li><tt>ogt</tt>: ordered and greater than </li>
5155 <li><tt>oge</tt>: ordered and greater than or equal</li>
5156 <li><tt>olt</tt>: ordered and less than </li>
5157 <li><tt>ole</tt>: ordered and less than or equal</li>
5158 <li><tt>one</tt>: ordered and not equal</li>
5159 <li><tt>ord</tt>: ordered (no nans)</li>
5160 <li><tt>ueq</tt>: unordered or equal</li>
5161 <li><tt>ugt</tt>: unordered or greater than </li>
5162 <li><tt>uge</tt>: unordered or greater than or equal</li>
5163 <li><tt>ult</tt>: unordered or less than </li>
5164 <li><tt>ule</tt>: unordered or less than or equal</li>
5165 <li><tt>une</tt>: unordered or not equal</li>
5166 <li><tt>uno</tt>: unordered (either nans)</li>
5167 <li><tt>true</tt>: no comparison, always returns true</li>
5168</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5169
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5170<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5171 <i>unordered</i> means that either operand may be a QNAN.</p>
5172
5173<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
5174 a <a href="#t_floating">floating point</a> type or
5175 a <a href="#t_vector">vector</a> of floating point type. They must have
5176 identical types.</p>
5177
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5178<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]5179<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5180 according to the condition code given as <tt>cond</tt>. If the operands are
5181 vectors, then the vectors are compared element by element. Each comparison
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]5182 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5183 follows:</p>
5184
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5185<ol>
5186 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5187
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5188 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5189 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5190
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5191 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]5192 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5193
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5194 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5195 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5196
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5197 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5198 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5199
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5200 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5201 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5202
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5203 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5204 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5205
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5206 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5207
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5208 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5209 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
5210
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5211 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5212 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
5213
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5214 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5215 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
5216
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5217 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5218 <tt>op1</tt> is less than <tt>op2</tt>.</li>
5219
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5220 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5221 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
5222
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5223 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5224 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
5225
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5226 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5227
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5228 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
5229</ol>
5230
5231<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5232<pre>
5233 &lt;result&gt; = fcmp oeq float 4.0, 5.0 <i>; yields: result=false</i>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]5234 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
5235 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
5236 &lt;result&gt; = fcmp ueq double 1.0, 2.0 <i>; yields: result=false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5237</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]5238
5239<p>Note that the code generator does not yet support vector types with
5240 the <tt>fcmp</tt> instruction.</p>
5241
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5242</div>
5243
5244<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]5245<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5246 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
5247</div>
5248
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5249<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5250
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5251<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5252<pre>
5253 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
5254</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5255
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5256<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5257<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
5258 SSA graph representing the function.</p>
5259
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5260<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5261<p>The type of the incoming values is specified with the first type field. After
5262 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
5263 one pair for each predecessor basic block of the current block. Only values
5264 of <a href="#t_firstclass">first class</a> type may be used as the value
5265 arguments to the PHI node. Only labels may be used as the label
5266 arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5267
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5268<p>There must be no non-phi instructions between the start of a basic block and
5269 the PHI instructions: i.e. PHI instructions must be first in a basic
5270 block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5271
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5272<p>For the purposes of the SSA form, the use of each incoming value is deemed to
5273 occur on the edge from the corresponding predecessor block to the current
5274 block (but after any definition of an '<tt>invoke</tt>' instruction's return
5275 value on the same edge).</p>
Jay Foad8e2fd2c2009-06-03 10:20:10 +0000[diff] [blame]5276
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5277<h5>Semantics:</h5>
5278<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5279 specified by the pair corresponding to the predecessor basic block that
5280 executed just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5281
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5282<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]5283<pre>
5284Loop: ; Infinite loop that counts from 0 on up...
5285 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
5286 %nextindvar = add i32 %indvar, 1
5287 br label %Loop
5288</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5289
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5290</div>
5291
5292<!-- _______________________________________________________________________ -->
5293<div class="doc_subsubsection">
5294 <a name="i_select">'<tt>select</tt>' Instruction</a>
5295</div>
5296
5297<div class="doc_text">
5298
5299<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5300<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]5301 &lt;result&gt; = select <i>selty</i> &lt;cond&gt;, &lt;ty&gt; &lt;val1&gt;, &lt;ty&gt; &lt;val2&gt; <i>; yields ty</i>
5302
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5303 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5304</pre>
5305
5306<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5307<p>The '<tt>select</tt>' instruction is used to choose one value based on a
5308 condition, without branching.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5309
5310
5311<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5312<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
5313 values indicating the condition, and two values of the
5314 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
5315 vectors and the condition is a scalar, then entire vectors are selected, not
5316 individual elements.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5317
5318<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5319<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
5320 first value argument; otherwise, it returns the second value argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5321
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5322<p>If the condition is a vector of i1, then the value arguments must be vectors
5323 of the same size, and the selection is done element by element.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5324
5325<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5326<pre>
5327 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
5328</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]5329
5330<p>Note that the code generator does not yet support conditions
5331 with vector type.</p>
5332
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5333</div>
5334
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5335<!-- _______________________________________________________________________ -->
5336<div class="doc_subsubsection">
5337 <a name="i_call">'<tt>call</tt>' Instruction</a>
5338</div>
5339
5340<div class="doc_text">
5341
5342<h5>Syntax:</h5>
5343<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]5344 &lt;result&gt; = [tail] call [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>] &lt;ty&gt; [&lt;fnty&gt;*] &lt;fnptrval&gt;(&lt;function args&gt;) [<a href="#fnattrs">fn attrs</a>]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5345</pre>
5346
5347<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5348<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
5349
5350<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5351<p>This instruction requires several arguments:</p>
5352
5353<ol>
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]5354 <li>The optional "tail" marker indicates that the callee function does not
5355 access any allocas or varargs in the caller. Note that calls may be
5356 marked "tail" even if they do not occur before
5357 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5358 present, the function call is eligible for tail call optimization,
5359 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
Evan Chengcc7495c2010-03-08 21:05:02 +0000[diff] [blame]5360 optimized into a jump</a>. The code generator may optimize calls marked
5361 "tail" with either 1) automatic <a href="CodeGenerator.html#sibcallopt">
5362 sibling call optimization</a> when the caller and callee have
5363 matching signatures, or 2) forced tail call optimization when the
5364 following extra requirements are met:
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]5365 <ul>
5366 <li>Caller and callee both have the calling
5367 convention <tt>fastcc</tt>.</li>
5368 <li>The call is in tail position (ret immediately follows call and ret
5369 uses value of call or is void).</li>
5370 <li>Option <tt>-tailcallopt</tt> is enabled,
Dan Gohman1be84f02010-03-02 01:08:11 +0000[diff] [blame]5371 or <code>llvm::GuaranteedTailCallOpt</code> is <code>true</code>.</li>
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]5372 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5373 constraints are met.</a></li>
5374 </ul>
5375 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5376
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5377 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5378 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]5379 defaults to using C calling conventions. The calling convention of the
5380 call must match the calling convention of the target function, or else the
5381 behavior is undefined.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5382
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5383 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5384 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5385 '<tt>inreg</tt>' attributes are valid here.</li>
5386
5387 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5388 type of the return value. Functions that return no value are marked
5389 <tt><a href="#t_void">void</a></tt>.</li>
5390
5391 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5392 being invoked. The argument types must match the types implied by this
5393 signature. This type can be omitted if the function is not varargs and if
5394 the function type does not return a pointer to a function.</li>
5395
5396 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5397 be invoked. In most cases, this is a direct function invocation, but
5398 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5399 to function value.</li>
5400
5401 <li>'<tt>function args</tt>': argument list whose types match the function
Chris Lattner553fb1e2010-03-02 06:36:51 +0000[diff] [blame]5402 signature argument types and parameter attributes. All arguments must be
5403 of <a href="#t_firstclass">first class</a> type. If the function
5404 signature indicates the function accepts a variable number of arguments,
5405 the extra arguments can be specified.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5406
5407 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5408 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5409 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5410</ol>
5411
5412<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5413<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5414 a specified function, with its incoming arguments bound to the specified
5415 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5416 function, control flow continues with the instruction after the function
5417 call, and the return value of the function is bound to the result
5418 argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5419
5420<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5421<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]5422 %retval = call i32 @test(i32 %argc)
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]5423 call i32 (i8*, ...)* @printf(i8* %msg, i32 12, i8 42) <i>; yields i32</i>
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]5424 %X = tail call i32 @foo() <i>; yields i32</i>
5425 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5426 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]5427
5428 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5429 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]5430 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5431 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]5432 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]5433 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5434</pre>
5435
Dale Johannesen9fcf4a92009-09-24 18:38:21 +0000[diff] [blame]5436<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen4730cd32009-09-25 17:04:42 +0000[diff] [blame]5437standard C99 library as being the C99 library functions, and may perform
5438optimizations or generate code for them under that assumption. This is
5439something we'd like to change in the future to provide better support for
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]5440freestanding environments and non-C-based languages.</p>
Dale Johannesen9fcf4a92009-09-24 18:38:21 +0000[diff] [blame]5441
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5442</div>
5443
5444<!-- _______________________________________________________________________ -->
5445<div class="doc_subsubsection">
5446 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5447</div>
5448
5449<div class="doc_text">
5450
5451<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5452<pre>
5453 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
5454</pre>
5455
5456<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5457<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5458 the "variable argument" area of a function call. It is used to implement the
5459 <tt>va_arg</tt> macro in C.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5460
5461<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5462<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5463 argument. It returns a value of the specified argument type and increments
5464 the <tt>va_list</tt> to point to the next argument. The actual type
5465 of <tt>va_list</tt> is target specific.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5466
5467<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5468<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5469 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5470 to the next argument. For more information, see the variable argument
5471 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5472
5473<p>It is legal for this instruction to be called in a function which does not
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5474 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5475 function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5476
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5477<p><tt>va_arg</tt> is an LLVM instruction instead of
5478 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5479 argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5480
5481<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5482<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5483
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5484<p>Note that the code generator does not yet fully support va_arg on many
5485 targets. Also, it does not currently support va_arg with aggregate types on
5486 any target.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]5487
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5488</div>
5489
5490<!-- *********************************************************************** -->
5491<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5492<!-- *********************************************************************** -->
5493
5494<div class="doc_text">
5495
5496<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5497 well known names and semantics and are required to follow certain
5498 restrictions. Overall, these intrinsics represent an extension mechanism for
5499 the LLVM language that does not require changing all of the transformations
5500 in LLVM when adding to the language (or the bitcode reader/writer, the
5501 parser, etc...).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5502
5503<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5504 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5505 begin with this prefix. Intrinsic functions must always be external
5506 functions: you cannot define the body of intrinsic functions. Intrinsic
5507 functions may only be used in call or invoke instructions: it is illegal to
5508 take the address of an intrinsic function. Additionally, because intrinsic
5509 functions are part of the LLVM language, it is required if any are added that
5510 they be documented here.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5511
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5512<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5513 family of functions that perform the same operation but on different data
5514 types. Because LLVM can represent over 8 million different integer types,
5515 overloading is used commonly to allow an intrinsic function to operate on any
5516 integer type. One or more of the argument types or the result type can be
5517 overloaded to accept any integer type. Argument types may also be defined as
5518 exactly matching a previous argument's type or the result type. This allows
5519 an intrinsic function which accepts multiple arguments, but needs all of them
5520 to be of the same type, to only be overloaded with respect to a single
5521 argument or the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5522
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5523<p>Overloaded intrinsics will have the names of its overloaded argument types
5524 encoded into its function name, each preceded by a period. Only those types
5525 which are overloaded result in a name suffix. Arguments whose type is matched
5526 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5527 can take an integer of any width and returns an integer of exactly the same
5528 integer width. This leads to a family of functions such as
5529 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5530 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5531 suffix is required. Because the argument's type is matched against the return
5532 type, it does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5533
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5534<p>To learn how to add an intrinsic function, please see the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5535 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5536
5537</div>
5538
5539<!-- ======================================================================= -->
5540<div class="doc_subsection">
5541 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5542</div>
5543
5544<div class="doc_text">
5545
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5546<p>Variable argument support is defined in LLVM with
5547 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5548 intrinsic functions. These functions are related to the similarly named
5549 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5550
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5551<p>All of these functions operate on arguments that use a target-specific value
5552 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5553 not define what this type is, so all transformations should be prepared to
5554 handle these functions regardless of the type used.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5555
5556<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5557 instruction and the variable argument handling intrinsic functions are
5558 used.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5559
5560<div class="doc_code">
5561<pre>
5562define i32 @test(i32 %X, ...) {
5563 ; Initialize variable argument processing
5564 %ap = alloca i8*
5565 %ap2 = bitcast i8** %ap to i8*
5566 call void @llvm.va_start(i8* %ap2)
5567
5568 ; Read a single integer argument
5569 %tmp = va_arg i8** %ap, i32
5570
5571 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5572 %aq = alloca i8*
5573 %aq2 = bitcast i8** %aq to i8*
5574 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5575 call void @llvm.va_end(i8* %aq2)
5576
5577 ; Stop processing of arguments.
5578 call void @llvm.va_end(i8* %ap2)
5579 ret i32 %tmp
5580}
5581
5582declare void @llvm.va_start(i8*)
5583declare void @llvm.va_copy(i8*, i8*)
5584declare void @llvm.va_end(i8*)
5585</pre>
5586</div>
5587
5588</div>
5589
5590<!-- _______________________________________________________________________ -->
5591<div class="doc_subsubsection">
5592 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5593</div>
5594
5595
5596<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5597
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5598<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5599<pre>
5600 declare void %llvm.va_start(i8* &lt;arglist&gt;)
5601</pre>
5602
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5603<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5604<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
5605 for subsequent use by <tt><a href="#i_va_arg">va_arg</a></tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5606
5607<h5>Arguments:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5608<p>The argument is a pointer to a <tt>va_list</tt> element to initialize.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5609
5610<h5>Semantics:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5611<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5612 macro available in C. In a target-dependent way, it initializes
5613 the <tt>va_list</tt> element to which the argument points, so that the next
5614 call to <tt>va_arg</tt> will produce the first variable argument passed to
5615 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5616 need to know the last argument of the function as the compiler can figure
5617 that out.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5618
5619</div>
5620
5621<!-- _______________________________________________________________________ -->
5622<div class="doc_subsubsection">
5623 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5624</div>
5625
5626<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5627
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5628<h5>Syntax:</h5>
5629<pre>
5630 declare void @llvm.va_end(i8* &lt;arglist&gt;)
5631</pre>
5632
5633<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5634<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5635 which has been initialized previously
5636 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5637 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5638
5639<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5640<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5641
5642<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5643<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5644 macro available in C. In a target-dependent way, it destroys
5645 the <tt>va_list</tt> element to which the argument points. Calls
5646 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5647 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5648 with calls to <tt>llvm.va_end</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5649
5650</div>
5651
5652<!-- _______________________________________________________________________ -->
5653<div class="doc_subsubsection">
5654 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5655</div>
5656
5657<div class="doc_text">
5658
5659<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5660<pre>
5661 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
5662</pre>
5663
5664<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5665<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5666 from the source argument list to the destination argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5667
5668<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5669<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5670 The second argument is a pointer to a <tt>va_list</tt> element to copy
5671 from.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5672
5673<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5674<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5675 macro available in C. In a target-dependent way, it copies the
5676 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5677 element. This intrinsic is necessary because
5678 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5679 arbitrarily complex and require, for example, memory allocation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5680
5681</div>
5682
5683<!-- ======================================================================= -->
5684<div class="doc_subsection">
5685 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5686</div>
5687
5688<div class="doc_text">
5689
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5690<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5691Collection</a> (GC) requires the implementation and generation of these
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5692intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5693roots on the stack</a>, as well as garbage collector implementations that
5694require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5695barriers. Front-ends for type-safe garbage collected languages should generate
5696these intrinsics to make use of the LLVM garbage collectors. For more details,
5697see <a href="GarbageCollection.html">Accurate Garbage Collection with
5698LLVM</a>.</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]5699
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5700<p>The garbage collection intrinsics only operate on objects in the generic
5701 address space (address space zero).</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]5702
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5703</div>
5704
5705<!-- _______________________________________________________________________ -->
5706<div class="doc_subsubsection">
5707 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5708</div>
5709
5710<div class="doc_text">
5711
5712<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5713<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5714 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5715</pre>
5716
5717<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5718<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5719 the code generator, and allows some metadata to be associated with it.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5720
5721<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5722<p>The first argument specifies the address of a stack object that contains the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5723 root pointer. The second pointer (which must be either a constant or a
5724 global value address) contains the meta-data to be associated with the
5725 root.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5726
5727<h5>Semantics:</h5>
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]5728<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5729 location. At compile-time, the code generator generates information to allow
5730 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5731 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5732 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5733
5734</div>
5735
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5736<!-- _______________________________________________________________________ -->
5737<div class="doc_subsubsection">
5738 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5739</div>
5740
5741<div class="doc_text">
5742
5743<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5744<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5745 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5746</pre>
5747
5748<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5749<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5750 locations, allowing garbage collector implementations that require read
5751 barriers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5752
5753<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5754<p>The second argument is the address to read from, which should be an address
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5755 allocated from the garbage collector. The first object is a pointer to the
5756 start of the referenced object, if needed by the language runtime (otherwise
5757 null).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5758
5759<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5760<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5761 instruction, but may be replaced with substantially more complex code by the
5762 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5763 may only be used in a function which <a href="#gc">specifies a GC
5764 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5765
5766</div>
5767
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5768<!-- _______________________________________________________________________ -->
5769<div class="doc_subsubsection">
5770 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5771</div>
5772
5773<div class="doc_text">
5774
5775<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5776<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5777 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5778</pre>
5779
5780<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5781<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5782 locations, allowing garbage collector implementations that require write
5783 barriers (such as generational or reference counting collectors).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5784
5785<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5786<p>The first argument is the reference to store, the second is the start of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5787 object to store it to, and the third is the address of the field of Obj to
5788 store to. If the runtime does not require a pointer to the object, Obj may
5789 be null.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5790
5791<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5792<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5793 instruction, but may be replaced with substantially more complex code by the
5794 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5795 may only be used in a function which <a href="#gc">specifies a GC
5796 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5797
5798</div>
5799
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5800<!-- ======================================================================= -->
5801<div class="doc_subsection">
5802 <a name="int_codegen">Code Generator Intrinsics</a>
5803</div>
5804
5805<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5806
5807<p>These intrinsics are provided by LLVM to expose special features that may
5808 only be implemented with code generator support.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5809
5810</div>
5811
5812<!-- _______________________________________________________________________ -->
5813<div class="doc_subsubsection">
5814 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5815</div>
5816
5817<div class="doc_text">
5818
5819<h5>Syntax:</h5>
5820<pre>
5821 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
5822</pre>
5823
5824<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5825<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5826 target-specific value indicating the return address of the current function
5827 or one of its callers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5828
5829<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5830<p>The argument to this intrinsic indicates which function to return the address
5831 for. Zero indicates the calling function, one indicates its caller, etc.
5832 The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5833
5834<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5835<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5836 indicating the return address of the specified call frame, or zero if it
5837 cannot be identified. The value returned by this intrinsic is likely to be
5838 incorrect or 0 for arguments other than zero, so it should only be used for
5839 debugging purposes.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5840
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5841<p>Note that calling this intrinsic does not prevent function inlining or other
5842 aggressive transformations, so the value returned may not be that of the
5843 obvious source-language caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5844
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5845</div>
5846
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5847<!-- _______________________________________________________________________ -->
5848<div class="doc_subsubsection">
5849 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5850</div>
5851
5852<div class="doc_text">
5853
5854<h5>Syntax:</h5>
5855<pre>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]5856 declare i8* @llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5857</pre>
5858
5859<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5860<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5861 target-specific frame pointer value for the specified stack frame.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5862
5863<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5864<p>The argument to this intrinsic indicates which function to return the frame
5865 pointer for. Zero indicates the calling function, one indicates its caller,
5866 etc. The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5867
5868<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5869<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5870 indicating the frame address of the specified call frame, or zero if it
5871 cannot be identified. The value returned by this intrinsic is likely to be
5872 incorrect or 0 for arguments other than zero, so it should only be used for
5873 debugging purposes.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5874
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5875<p>Note that calling this intrinsic does not prevent function inlining or other
5876 aggressive transformations, so the value returned may not be that of the
5877 obvious source-language caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5878
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5879</div>
5880
5881<!-- _______________________________________________________________________ -->
5882<div class="doc_subsubsection">
5883 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5884</div>
5885
5886<div class="doc_text">
5887
5888<h5>Syntax:</h5>
5889<pre>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]5890 declare i8* @llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5891</pre>
5892
5893<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5894<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5895 of the function stack, for use
5896 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5897 useful for implementing language features like scoped automatic variable
5898 sized arrays in C99.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5899
5900<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5901<p>This intrinsic returns a opaque pointer value that can be passed
5902 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5903 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5904 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5905 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5906 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5907 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5908
5909</div>
5910
5911<!-- _______________________________________________________________________ -->
5912<div class="doc_subsubsection">
5913 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5914</div>
5915
5916<div class="doc_text">
5917
5918<h5>Syntax:</h5>
5919<pre>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]5920 declare void @llvm.stackrestore(i8* %ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5921</pre>
5922
5923<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5924<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5925 the function stack to the state it was in when the
5926 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5927 executed. This is useful for implementing language features like scoped
5928 automatic variable sized arrays in C99.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5929
5930<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5931<p>See the description
5932 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5933
5934</div>
5935
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5936<!-- _______________________________________________________________________ -->
5937<div class="doc_subsubsection">
5938 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5939</div>
5940
5941<div class="doc_text">
5942
5943<h5>Syntax:</h5>
5944<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5945 declare void @llvm.prefetch(i8* &lt;address&gt;, i32 &lt;rw&gt;, i32 &lt;locality&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5946</pre>
5947
5948<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5949<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5950 insert a prefetch instruction if supported; otherwise, it is a noop.
5951 Prefetches have no effect on the behavior of the program but can change its
5952 performance characteristics.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5953
5954<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5955<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5956 specifier determining if the fetch should be for a read (0) or write (1),
5957 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5958 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5959 and <tt>locality</tt> arguments must be constant integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5960
5961<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5962<p>This intrinsic does not modify the behavior of the program. In particular,
5963 prefetches cannot trap and do not produce a value. On targets that support
5964 this intrinsic, the prefetch can provide hints to the processor cache for
5965 better performance.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5966
5967</div>
5968
5969<!-- _______________________________________________________________________ -->
5970<div class="doc_subsubsection">
5971 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5972</div>
5973
5974<div class="doc_text">
5975
5976<h5>Syntax:</h5>
5977<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5978 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5979</pre>
5980
5981<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5982<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5983 Counter (PC) in a region of code to simulators and other tools. The method
5984 is target specific, but it is expected that the marker will use exported
5985 symbols to transmit the PC of the marker. The marker makes no guarantees
5986 that it will remain with any specific instruction after optimizations. It is
5987 possible that the presence of a marker will inhibit optimizations. The
5988 intended use is to be inserted after optimizations to allow correlations of
5989 simulation runs.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5990
5991<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5992<p><tt>id</tt> is a numerical id identifying the marker.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5993
5994<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5995<p>This intrinsic does not modify the behavior of the program. Backends that do
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]5996 not support this intrinsic may ignore it.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5997
5998</div>
5999
6000<!-- _______________________________________________________________________ -->
6001<div class="doc_subsubsection">
6002 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
6003</div>
6004
6005<div class="doc_text">
6006
6007<h5>Syntax:</h5>
6008<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]6009 declare i64 @llvm.readcyclecounter()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6010</pre>
6011
6012<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6013<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
6014 counter register (or similar low latency, high accuracy clocks) on those
6015 targets that support it. On X86, it should map to RDTSC. On Alpha, it
6016 should map to RPCC. As the backing counters overflow quickly (on the order
6017 of 9 seconds on alpha), this should only be used for small timings.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6018
6019<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6020<p>When directly supported, reading the cycle counter should not modify any
6021 memory. Implementations are allowed to either return a application specific
6022 value or a system wide value. On backends without support, this is lowered
6023 to a constant 0.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6024
6025</div>
6026
6027<!-- ======================================================================= -->
6028<div class="doc_subsection">
6029 <a name="int_libc">Standard C Library Intrinsics</a>
6030</div>
6031
6032<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6033
6034<p>LLVM provides intrinsics for a few important standard C library functions.
6035 These intrinsics allow source-language front-ends to pass information about
6036 the alignment of the pointer arguments to the code generator, providing
6037 opportunity for more efficient code generation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6038
6039</div>
6040
6041<!-- _______________________________________________________________________ -->
6042<div class="doc_subsubsection">
6043 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
6044</div>
6045
6046<div class="doc_text">
6047
6048<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6049<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
Mon P Wang238462c2010-04-07 06:35:53 +0000[diff] [blame]6050 integer bit width and for different address spaces. Not all targets support
6051 all bit widths however.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6052
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6053<pre>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]6054 declare void @llvm.memcpy.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6055 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]6056 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6057 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6058</pre>
6059
6060<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6061<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6062 source location to the destination location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6063
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6064<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6065 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6066 and the pointers can be in specified address spaces.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6067
6068<h5>Arguments:</h5>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6069
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6070<p>The first argument is a pointer to the destination, the second is a pointer
6071 to the source. The third argument is an integer argument specifying the
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6072 number of bytes to copy, the fourth argument is the alignment of the
6073 source and destination locations, and the fifth is a boolean indicating a
6074 volatile access.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6075
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]6076<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6077 then the caller guarantees that both the source and destination pointers are
6078 aligned to that boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6079
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]6080<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6081 <tt>llvm.memcpy</tt> call is a <a href="#volatile">volatile operation</a>.
6082 The detailed access behavior is not very cleanly specified and it is unwise
6083 to depend on it.</p>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6084
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6085<h5>Semantics:</h5>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6086
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6087<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
6088 source location to the destination location, which are not allowed to
6089 overlap. It copies "len" bytes of memory over. If the argument is known to
6090 be aligned to some boundary, this can be specified as the fourth argument,
6091 otherwise it should be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6092
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6093</div>
6094
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6095<!-- _______________________________________________________________________ -->
6096<div class="doc_subsubsection">
6097 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
6098</div>
6099
6100<div class="doc_text">
6101
6102<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]6103<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Mon P Wang238462c2010-04-07 06:35:53 +0000[diff] [blame]6104 width and for different address space. Not all targets support all bit
6105 widths however.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6106
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6107<pre>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]6108 declare void @llvm.memmove.p0i8.p0i8.i32(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6109 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]6110 declare void @llvm.memmove.p0i8.p0i8.i64(i8* &lt;dest&gt;, i8* &lt;src&gt;,
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6111 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6112</pre>
6113
6114<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6115<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
6116 source location to the destination location. It is similar to the
6117 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
6118 overlap.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6119
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6120<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6121 intrinsics do not return a value, takes extra alignment/isvolatile arguments
6122 and the pointers can be in specified address spaces.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6123
6124<h5>Arguments:</h5>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6125
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6126<p>The first argument is a pointer to the destination, the second is a pointer
6127 to the source. The third argument is an integer argument specifying the
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6128 number of bytes to copy, the fourth argument is the alignment of the
6129 source and destination locations, and the fifth is a boolean indicating a
6130 volatile access.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6131
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]6132<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6133 then the caller guarantees that the source and destination pointers are
6134 aligned to that boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6135
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]6136<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6137 <tt>llvm.memmove</tt> call is a <a href="#volatile">volatile operation</a>.
6138 The detailed access behavior is not very cleanly specified and it is unwise
6139 to depend on it.</p>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6140
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6141<h5>Semantics:</h5>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6142
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6143<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
6144 source location to the destination location, which may overlap. It copies
6145 "len" bytes of memory over. If the argument is known to be aligned to some
6146 boundary, this can be specified as the fourth argument, otherwise it should
6147 be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6148
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6149</div>
6150
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6151<!-- _______________________________________________________________________ -->
6152<div class="doc_subsubsection">
6153 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
6154</div>
6155
6156<div class="doc_text">
6157
6158<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]6159<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
Mon P Wang238462c2010-04-07 06:35:53 +0000[diff] [blame]6160 width and for different address spaces. Not all targets support all bit
6161 widths however.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6162
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6163<pre>
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]6164 declare void @llvm.memset.p0i8.i32(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerc0dfafe2010-04-08 00:54:34 +0000[diff] [blame]6165 i32 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanecfb95c2010-05-28 17:13:49 +0000[diff] [blame]6166 declare void @llvm.memset.p0i8.i64(i8* &lt;dest&gt;, i8 &lt;val&gt;,
Chris Lattnerc0dfafe2010-04-08 00:54:34 +0000[diff] [blame]6167 i64 &lt;len&gt;, i32 &lt;align&gt;, i1 &lt;isvolatile&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6168</pre>
6169
6170<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6171<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
6172 particular byte value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6173
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6174<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6175 intrinsic does not return a value, takes extra alignment/volatile arguments,
6176 and the destination can be in an arbitrary address space.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6177
6178<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6179<p>The first argument is a pointer to the destination to fill, the second is the
6180 byte value to fill it with, the third argument is an integer argument
6181 specifying the number of bytes to fill, and the fourth argument is the known
6182 alignment of destination location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6183
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]6184<p>If the call to this intrinsic has an alignment value that is not 0 or 1,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6185 then the caller guarantees that the destination pointer is aligned to that
6186 boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6187
Jeffrey Yasskin6dc467f2010-04-26 21:21:24 +0000[diff] [blame]6188<p>If the <tt>isvolatile</tt> parameter is <tt>true</tt>, the
6189 <tt>llvm.memset</tt> call is a <a href="#volatile">volatile operation</a>.
6190 The detailed access behavior is not very cleanly specified and it is unwise
6191 to depend on it.</p>
Chris Lattner5aa5aab2010-04-08 00:53:57 +0000[diff] [blame]6192
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6193<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6194<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
6195 at the destination location. If the argument is known to be aligned to some
6196 boundary, this can be specified as the fourth argument, otherwise it should
6197 be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6198
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6199</div>
6200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6201<!-- _______________________________________________________________________ -->
6202<div class="doc_subsubsection">
6203 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
6204</div>
6205
6206<div class="doc_text">
6207
6208<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6209<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
6210 floating point or vector of floating point type. Not all targets support all
6211 types however.</p>
6212
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6213<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]6214 declare float @llvm.sqrt.f32(float %Val)
6215 declare double @llvm.sqrt.f64(double %Val)
6216 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
6217 declare fp128 @llvm.sqrt.f128(fp128 %Val)
6218 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6219</pre>
6220
6221<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6222<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
6223 returning the same value as the libm '<tt>sqrt</tt>' functions would.
6224 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
6225 behavior for negative numbers other than -0.0 (which allows for better
6226 optimization, because there is no need to worry about errno being
6227 set). <tt>llvm.sqrt(-0.0)</tt> is defined to return -0.0 like IEEE sqrt.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6228
6229<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6230<p>The argument and return value are floating point numbers of the same
6231 type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6232
6233<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6234<p>This function returns the sqrt of the specified operand if it is a
6235 nonnegative floating point number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6236
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6237</div>
6238
6239<!-- _______________________________________________________________________ -->
6240<div class="doc_subsubsection">
6241 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
6242</div>
6243
6244<div class="doc_text">
6245
6246<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6247<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
6248 floating point or vector of floating point type. Not all targets support all
6249 types however.</p>
6250
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6251<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]6252 declare float @llvm.powi.f32(float %Val, i32 %power)
6253 declare double @llvm.powi.f64(double %Val, i32 %power)
6254 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
6255 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
6256 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6257</pre>
6258
6259<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6260<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
6261 specified (positive or negative) power. The order of evaluation of
6262 multiplications is not defined. When a vector of floating point type is
6263 used, the second argument remains a scalar integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6264
6265<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6266<p>The second argument is an integer power, and the first is a value to raise to
6267 that power.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6268
6269<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6270<p>This function returns the first value raised to the second power with an
6271 unspecified sequence of rounding operations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6272
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6273</div>
6274
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6275<!-- _______________________________________________________________________ -->
6276<div class="doc_subsubsection">
6277 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
6278</div>
6279
6280<div class="doc_text">
6281
6282<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6283<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
6284 floating point or vector of floating point type. Not all targets support all
6285 types however.</p>
6286
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6287<pre>
6288 declare float @llvm.sin.f32(float %Val)
6289 declare double @llvm.sin.f64(double %Val)
6290 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
6291 declare fp128 @llvm.sin.f128(fp128 %Val)
6292 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
6293</pre>
6294
6295<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6296<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6297
6298<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6299<p>The argument and return value are floating point numbers of the same
6300 type.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6301
6302<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6303<p>This function returns the sine of the specified operand, returning the same
6304 values as the libm <tt>sin</tt> functions would, and handles error conditions
6305 in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6306
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6307</div>
6308
6309<!-- _______________________________________________________________________ -->
6310<div class="doc_subsubsection">
6311 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
6312</div>
6313
6314<div class="doc_text">
6315
6316<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6317<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
6318 floating point or vector of floating point type. Not all targets support all
6319 types however.</p>
6320
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6321<pre>
6322 declare float @llvm.cos.f32(float %Val)
6323 declare double @llvm.cos.f64(double %Val)
6324 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
6325 declare fp128 @llvm.cos.f128(fp128 %Val)
6326 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
6327</pre>
6328
6329<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6330<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6331
6332<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6333<p>The argument and return value are floating point numbers of the same
6334 type.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6335
6336<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6337<p>This function returns the cosine of the specified operand, returning the same
6338 values as the libm <tt>cos</tt> functions would, and handles error conditions
6339 in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6340
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6341</div>
6342
6343<!-- _______________________________________________________________________ -->
6344<div class="doc_subsubsection">
6345 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
6346</div>
6347
6348<div class="doc_text">
6349
6350<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6351<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
6352 floating point or vector of floating point type. Not all targets support all
6353 types however.</p>
6354
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6355<pre>
6356 declare float @llvm.pow.f32(float %Val, float %Power)
6357 declare double @llvm.pow.f64(double %Val, double %Power)
6358 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
6359 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
6360 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
6361</pre>
6362
6363<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6364<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
6365 specified (positive or negative) power.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6366
6367<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6368<p>The second argument is a floating point power, and the first is a value to
6369 raise to that power.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6370
6371<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6372<p>This function returns the first value raised to the second power, returning
6373 the same values as the libm <tt>pow</tt> functions would, and handles error
6374 conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6375
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6376</div>
6377
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6378<!-- ======================================================================= -->
6379<div class="doc_subsection">
6380 <a name="int_manip">Bit Manipulation Intrinsics</a>
6381</div>
6382
6383<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6384
6385<p>LLVM provides intrinsics for a few important bit manipulation operations.
6386 These allow efficient code generation for some algorithms.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6387
6388</div>
6389
6390<!-- _______________________________________________________________________ -->
6391<div class="doc_subsubsection">
6392 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6393</div>
6394
6395<div class="doc_text">
6396
6397<h5>Syntax:</h5>
6398<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6399 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6400
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6401<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6402 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
6403 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
6404 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6405</pre>
6406
6407<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6408<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6409 values with an even number of bytes (positive multiple of 16 bits). These
6410 are useful for performing operations on data that is not in the target's
6411 native byte order.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6412
6413<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6414<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6415 and low byte of the input i16 swapped. Similarly,
6416 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6417 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6418 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6419 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6420 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6421 more, respectively).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6422
6423</div>
6424
6425<!-- _______________________________________________________________________ -->
6426<div class="doc_subsubsection">
6427 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6428</div>
6429
6430<div class="doc_text">
6431
6432<h5>Syntax:</h5>
6433<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6434 width. Not all targets support all bit widths however.</p>
6435
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6436<pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6437 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6438 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6439 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6440 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
6441 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6442</pre>
6443
6444<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6445<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6446 in a value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6447
6448<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6449<p>The only argument is the value to be counted. The argument may be of any
6450 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6451
6452<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6453<p>The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6454
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6455</div>
6456
6457<!-- _______________________________________________________________________ -->
6458<div class="doc_subsubsection">
6459 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6460</div>
6461
6462<div class="doc_text">
6463
6464<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6465<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6466 integer bit width. Not all targets support all bit widths however.</p>
6467
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6468<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6469 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
6470 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6471 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6472 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
6473 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6474</pre>
6475
6476<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6477<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6478 leading zeros in a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6479
6480<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6481<p>The only argument is the value to be counted. The argument may be of any
6482 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6483
6484<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6485<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6486 zeros in a variable. If the src == 0 then the result is the size in bits of
6487 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6489</div>
6490
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6491<!-- _______________________________________________________________________ -->
6492<div class="doc_subsubsection">
6493 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6494</div>
6495
6496<div class="doc_text">
6497
6498<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6499<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6500 integer bit width. Not all targets support all bit widths however.</p>
6501
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6502<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6503 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
6504 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6505 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6506 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
6507 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6508</pre>
6509
6510<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6511<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6512 trailing zeros.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6513
6514<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6515<p>The only argument is the value to be counted. The argument may be of any
6516 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6517
6518<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6519<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6520 zeros in a variable. If the src == 0 then the result is the size in bits of
6521 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6522
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6523</div>
6524
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6525<!-- ======================================================================= -->
6526<div class="doc_subsection">
6527 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6528</div>
6529
6530<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6531
6532<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6533
6534</div>
6535
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6536<!-- _______________________________________________________________________ -->
6537<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6538 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6539</div>
6540
6541<div class="doc_text">
6542
6543<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6544<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6545 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6546
6547<pre>
6548 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6549 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6550 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6551</pre>
6552
6553<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6554<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6555 a signed addition of the two arguments, and indicate whether an overflow
6556 occurred during the signed summation.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6557
6558<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6559<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6560 be of integer types of any bit width, but they must have the same bit
6561 width. The second element of the result structure must be of
6562 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6563 undergo signed addition.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6564
6565<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6566<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6567 a signed addition of the two variables. They return a structure &mdash; the
6568 first element of which is the signed summation, and the second element of
6569 which is a bit specifying if the signed summation resulted in an
6570 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6571
6572<h5>Examples:</h5>
6573<pre>
6574 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6575 %sum = extractvalue {i32, i1} %res, 0
6576 %obit = extractvalue {i32, i1} %res, 1
6577 br i1 %obit, label %overflow, label %normal
6578</pre>
6579
6580</div>
6581
6582<!-- _______________________________________________________________________ -->
6583<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6584 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6585</div>
6586
6587<div class="doc_text">
6588
6589<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6590<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6591 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6592
6593<pre>
6594 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6595 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6596 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6597</pre>
6598
6599<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6600<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6601 an unsigned addition of the two arguments, and indicate whether a carry
6602 occurred during the unsigned summation.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6603
6604<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6605<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6606 be of integer types of any bit width, but they must have the same bit
6607 width. The second element of the result structure must be of
6608 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6609 undergo unsigned addition.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6610
6611<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6612<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6613 an unsigned addition of the two arguments. They return a structure &mdash;
6614 the first element of which is the sum, and the second element of which is a
6615 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6616
6617<h5>Examples:</h5>
6618<pre>
6619 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6620 %sum = extractvalue {i32, i1} %res, 0
6621 %obit = extractvalue {i32, i1} %res, 1
6622 br i1 %obit, label %carry, label %normal
6623</pre>
6624
6625</div>
6626
6627<!-- _______________________________________________________________________ -->
6628<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6629 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6630</div>
6631
6632<div class="doc_text">
6633
6634<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6635<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6636 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6637
6638<pre>
6639 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6640 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6641 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6642</pre>
6643
6644<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6645<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6646 a signed subtraction of the two arguments, and indicate whether an overflow
6647 occurred during the signed subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6648
6649<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6650<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6651 be of integer types of any bit width, but they must have the same bit
6652 width. The second element of the result structure must be of
6653 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6654 undergo signed subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6655
6656<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6657<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6658 a signed subtraction of the two arguments. They return a structure &mdash;
6659 the first element of which is the subtraction, and the second element of
6660 which is a bit specifying if the signed subtraction resulted in an
6661 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6662
6663<h5>Examples:</h5>
6664<pre>
6665 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6666 %sum = extractvalue {i32, i1} %res, 0
6667 %obit = extractvalue {i32, i1} %res, 1
6668 br i1 %obit, label %overflow, label %normal
6669</pre>
6670
6671</div>
6672
6673<!-- _______________________________________________________________________ -->
6674<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6675 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6676</div>
6677
6678<div class="doc_text">
6679
6680<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6681<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6682 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6683
6684<pre>
6685 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6686 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6687 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6688</pre>
6689
6690<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6691<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6692 an unsigned subtraction of the two arguments, and indicate whether an
6693 overflow occurred during the unsigned subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6694
6695<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6696<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6697 be of integer types of any bit width, but they must have the same bit
6698 width. The second element of the result structure must be of
6699 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6700 undergo unsigned subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6701
6702<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6703<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6704 an unsigned subtraction of the two arguments. They return a structure &mdash;
6705 the first element of which is the subtraction, and the second element of
6706 which is a bit specifying if the unsigned subtraction resulted in an
6707 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6708
6709<h5>Examples:</h5>
6710<pre>
6711 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6712 %sum = extractvalue {i32, i1} %res, 0
6713 %obit = extractvalue {i32, i1} %res, 1
6714 br i1 %obit, label %overflow, label %normal
6715</pre>
6716
6717</div>
6718
6719<!-- _______________________________________________________________________ -->
6720<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6721 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6722</div>
6723
6724<div class="doc_text">
6725
6726<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6727<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6728 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6729
6730<pre>
6731 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6732 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6733 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6734</pre>
6735
6736<h5>Overview:</h5>
6737
6738<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6739 a signed multiplication of the two arguments, and indicate whether an
6740 overflow occurred during the signed multiplication.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6741
6742<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6743<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6744 be of integer types of any bit width, but they must have the same bit
6745 width. The second element of the result structure must be of
6746 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6747 undergo signed multiplication.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6748
6749<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6750<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6751 a signed multiplication of the two arguments. They return a structure &mdash;
6752 the first element of which is the multiplication, and the second element of
6753 which is a bit specifying if the signed multiplication resulted in an
6754 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6755
6756<h5>Examples:</h5>
6757<pre>
6758 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6759 %sum = extractvalue {i32, i1} %res, 0
6760 %obit = extractvalue {i32, i1} %res, 1
6761 br i1 %obit, label %overflow, label %normal
6762</pre>
6763
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6764</div>
6765
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6766<!-- _______________________________________________________________________ -->
6767<div class="doc_subsubsection">
6768 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6769</div>
6770
6771<div class="doc_text">
6772
6773<h5>Syntax:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6774<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6775 on any integer bit width.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6776
6777<pre>
6778 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6779 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6780 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6781</pre>
6782
6783<h5>Overview:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6784<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6785 a unsigned multiplication of the two arguments, and indicate whether an
6786 overflow occurred during the unsigned multiplication.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6787
6788<h5>Arguments:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6789<p>The arguments (%a and %b) and the first element of the result structure may
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6790 be of integer types of any bit width, but they must have the same bit
6791 width. The second element of the result structure must be of
6792 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6793 undergo unsigned multiplication.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6794
6795<h5>Semantics:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6796<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6797 an unsigned multiplication of the two arguments. They return a structure
6798 &mdash; the first element of which is the multiplication, and the second
6799 element of which is a bit specifying if the unsigned multiplication resulted
6800 in an overflow.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6801
6802<h5>Examples:</h5>
6803<pre>
6804 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6805 %sum = extractvalue {i32, i1} %res, 0
6806 %obit = extractvalue {i32, i1} %res, 1
6807 br i1 %obit, label %overflow, label %normal
6808</pre>
6809
6810</div>
6811
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6812<!-- ======================================================================= -->
6813<div class="doc_subsection">
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]6814 <a name="int_fp16">Half Precision Floating Point Intrinsics</a>
6815</div>
6816
6817<div class="doc_text">
6818
Chris Lattnere5969c62010-03-15 04:12:21 +0000[diff] [blame]6819<p>Half precision floating point is a storage-only format. This means that it is
6820 a dense encoding (in memory) but does not support computation in the
6821 format.</p>
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6822
Chris Lattnere5969c62010-03-15 04:12:21 +0000[diff] [blame]6823<p>This means that code must first load the half-precision floating point
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6824 value as an i16, then convert it to float with <a
6825 href="#int_convert_from_fp16"><tt>llvm.convert.from.fp16</tt></a>.
6826 Computation can then be performed on the float value (including extending to
Chris Lattnere5969c62010-03-15 04:12:21 +0000[diff] [blame]6827 double etc). To store the value back to memory, it is first converted to
6828 float if needed, then converted to i16 with
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6829 <a href="#int_convert_to_fp16"><tt>llvm.convert.to.fp16</tt></a>, then
6830 storing as an i16 value.</p>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]6831</div>
6832
6833<!-- _______________________________________________________________________ -->
6834<div class="doc_subsubsection">
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6835 <a name="int_convert_to_fp16">'<tt>llvm.convert.to.fp16</tt>' Intrinsic</a>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]6836</div>
6837
6838<div class="doc_text">
6839
6840<h5>Syntax:</h5>
6841<pre>
6842 declare i16 @llvm.convert.to.fp16(f32 %a)
6843</pre>
6844
6845<h5>Overview:</h5>
6846<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6847 a conversion from single precision floating point format to half precision
6848 floating point format.</p>
6849
6850<h5>Arguments:</h5>
6851<p>The intrinsic function contains single argument - the value to be
6852 converted.</p>
6853
6854<h5>Semantics:</h5>
6855<p>The '<tt>llvm.convert.to.fp16</tt>' intrinsic function performs
6856 a conversion from single precision floating point format to half precision
Chris Lattnere5969c62010-03-15 04:12:21 +0000[diff] [blame]6857 floating point format. The return value is an <tt>i16</tt> which
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6858 contains the converted number.</p>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]6859
6860<h5>Examples:</h5>
6861<pre>
6862 %res = call i16 @llvm.convert.to.fp16(f32 %a)
6863 store i16 %res, i16* @x, align 2
6864</pre>
6865
6866</div>
6867
6868<!-- _______________________________________________________________________ -->
6869<div class="doc_subsubsection">
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6870 <a name="int_convert_from_fp16">'<tt>llvm.convert.from.fp16</tt>' Intrinsic</a>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]6871</div>
6872
6873<div class="doc_text">
6874
6875<h5>Syntax:</h5>
6876<pre>
6877 declare f32 @llvm.convert.from.fp16(i16 %a)
6878</pre>
6879
6880<h5>Overview:</h5>
6881<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs
6882 a conversion from half precision floating point format to single precision
6883 floating point format.</p>
6884
6885<h5>Arguments:</h5>
6886<p>The intrinsic function contains single argument - the value to be
6887 converted.</p>
6888
6889<h5>Semantics:</h5>
6890<p>The '<tt>llvm.convert.from.fp16</tt>' intrinsic function performs a
Chris Lattnere5969c62010-03-15 04:12:21 +0000[diff] [blame]6891 conversion from half single precision floating point format to single
Chris Lattnerebc48e52010-03-14 23:03:31 +0000[diff] [blame]6892 precision floating point format. The input half-float value is represented by
6893 an <tt>i16</tt> value.</p>
Anton Korobeynikov8c98a892010-03-14 18:42:47 +0000[diff] [blame]6894
6895<h5>Examples:</h5>
6896<pre>
6897 %a = load i16* @x, align 2
6898 %res = call f32 @llvm.convert.from.fp16(i16 %a)
6899</pre>
6900
6901</div>
6902
6903<!-- ======================================================================= -->
6904<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6905 <a name="int_debugger">Debugger Intrinsics</a>
6906</div>
6907
6908<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6909
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6910<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6911 prefix), are described in
6912 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6913 Level Debugging</a> document.</p>
6914
6915</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6916
6917<!-- ======================================================================= -->
6918<div class="doc_subsection">
6919 <a name="int_eh">Exception Handling Intrinsics</a>
6920</div>
6921
6922<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6923
6924<p>The LLVM exception handling intrinsics (which all start with
6925 <tt>llvm.eh.</tt> prefix), are described in
6926 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6927 Handling</a> document.</p>
6928
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6929</div>
6930
6931<!-- ======================================================================= -->
6932<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6933 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6934</div>
6935
6936<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6937
6938<p>This intrinsic makes it possible to excise one parameter, marked with
6939 the <tt>nest</tt> attribute, from a function. The result is a callable
6940 function pointer lacking the nest parameter - the caller does not need to
6941 provide a value for it. Instead, the value to use is stored in advance in a
6942 "trampoline", a block of memory usually allocated on the stack, which also
6943 contains code to splice the nest value into the argument list. This is used
6944 to implement the GCC nested function address extension.</p>
6945
6946<p>For example, if the function is
6947 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6948 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6949 follows:</p>
6950
6951<div class="doc_code">
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6952<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6953 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6954 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]6955 %p = call i8* @llvm.init.trampoline(i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval)
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6956 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6957</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6958</div>
6959
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]6960<p>The call <tt>%val = call i32 %fp(i32 %x, i32 %y)</tt> is then equivalent
6961 to <tt>%val = call i32 %f(i8* %nval, i32 %x, i32 %y)</tt>.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6962
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6963</div>
6964
6965<!-- _______________________________________________________________________ -->
6966<div class="doc_subsubsection">
6967 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6968</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6969
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6970<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6971
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6972<h5>Syntax:</h5>
6973<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6974 declare i8* @llvm.init.trampoline(i8* &lt;tramp&gt;, i8* &lt;func&gt;, i8* &lt;nval&gt;)
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6975</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6976
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6977<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6978<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6979 function pointer suitable for executing it.</p>
6980
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6981<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6982<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6983 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6984 sufficiently aligned block of memory; this memory is written to by the
6985 intrinsic. Note that the size and the alignment are target-specific - LLVM
6986 currently provides no portable way of determining them, so a front-end that
6987 generates this intrinsic needs to have some target-specific knowledge.
6988 The <tt>func</tt> argument must hold a function bitcast to
6989 an <tt>i8*</tt>.</p>
6990
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6991<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6992<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6993 dependent code, turning it into a function. A pointer to this function is
6994 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6995 function pointer type</a> before being called. The new function's signature
6996 is the same as that of <tt>func</tt> with any arguments marked with
6997 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6998 is allowed, and it must be of pointer type. Calling the new function is
6999 equivalent to calling <tt>func</tt> with the same argument list, but
7000 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
7001 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
7002 by <tt>tramp</tt> is modified, then the effect of any later call to the
7003 returned function pointer is undefined.</p>
7004
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]7005</div>
7006
7007<!-- ======================================================================= -->
7008<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7009 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
7010</div>
7011
7012<div class="doc_text">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7013
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7014<p>These intrinsic functions expand the "universal IR" of LLVM to represent
7015 hardware constructs for atomic operations and memory synchronization. This
7016 provides an interface to the hardware, not an interface to the programmer. It
7017 is aimed at a low enough level to allow any programming models or APIs
7018 (Application Programming Interfaces) which need atomic behaviors to map
7019 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
7020 hardware provides a "universal IR" for source languages, it also provides a
7021 starting point for developing a "universal" atomic operation and
7022 synchronization IR.</p>
7023
7024<p>These do <em>not</em> form an API such as high-level threading libraries,
7025 software transaction memory systems, atomic primitives, and intrinsic
7026 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
7027 application libraries. The hardware interface provided by LLVM should allow
7028 a clean implementation of all of these APIs and parallel programming models.
7029 No one model or paradigm should be selected above others unless the hardware
7030 itself ubiquitously does so.</p>
7031
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7032</div>
7033
7034<!-- _______________________________________________________________________ -->
7035<div class="doc_subsubsection">
7036 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
7037</div>
7038<div class="doc_text">
7039<h5>Syntax:</h5>
7040<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7041 declare void @llvm.memory.barrier(i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;, i1 &lt;device&gt;)
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7042</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7043
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7044<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7045<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
7046 specific pairs of memory access types.</p>
7047
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7048<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7049<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
7050 The first four arguments enables a specific barrier as listed below. The
Dan Gohman22dc6682010-03-01 17:41:39 +0000[diff] [blame]7051 fifth argument specifies that the barrier applies to io or device or uncached
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7052 memory.</p>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7053
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7054<ul>
7055 <li><tt>ll</tt>: load-load barrier</li>
7056 <li><tt>ls</tt>: load-store barrier</li>
7057 <li><tt>sl</tt>: store-load barrier</li>
7058 <li><tt>ss</tt>: store-store barrier</li>
7059 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
7060</ul>
7061
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7062<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7063<p>This intrinsic causes the system to enforce some ordering constraints upon
7064 the loads and stores of the program. This barrier does not
7065 indicate <em>when</em> any events will occur, it only enforces
7066 an <em>order</em> in which they occur. For any of the specified pairs of load
7067 and store operations (f.ex. load-load, or store-load), all of the first
7068 operations preceding the barrier will complete before any of the second
7069 operations succeeding the barrier begin. Specifically the semantics for each
7070 pairing is as follows:</p>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7071
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7072<ul>
7073 <li><tt>ll</tt>: All loads before the barrier must complete before any load
7074 after the barrier begins.</li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7075 <li><tt>ls</tt>: All loads before the barrier must complete before any
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7076 store after the barrier begins.</li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7077 <li><tt>ss</tt>: All stores before the barrier must complete before any
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7078 store after the barrier begins.</li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7079 <li><tt>sl</tt>: All stores before the barrier must complete before any
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7080 load after the barrier begins.</li>
7081</ul>
7082
7083<p>These semantics are applied with a logical "and" behavior when more than one
7084 is enabled in a single memory barrier intrinsic.</p>
7085
7086<p>Backends may implement stronger barriers than those requested when they do
7087 not support as fine grained a barrier as requested. Some architectures do
7088 not need all types of barriers and on such architectures, these become
7089 noops.</p>
7090
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7091<h5>Example:</h5>
7092<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7093%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7094%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7095 store i32 4, %ptr
7096
7097%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7098 call void @llvm.memory.barrier(i1 false, i1 true, i1 false, i1 false)
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7099 <i>; guarantee the above finishes</i>
7100 store i32 8, %ptr <i>; before this begins</i>
7101</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7102
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7103</div>
7104
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7105<!-- _______________________________________________________________________ -->
7106<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7107 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7108</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7109
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7110<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7111
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7112<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7113<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
7114 any integer bit width and for different address spaces. Not all targets
7115 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7116
7117<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7118 declare i8 @llvm.atomic.cmp.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt;)
7119 declare i16 @llvm.atomic.cmp.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt;)
7120 declare i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt;)
7121 declare i64 @llvm.atomic.cmp.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;cmp&gt;, i64 &lt;val&gt;)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7122</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7123
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7124<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7125<p>This loads a value in memory and compares it to a given value. If they are
7126 equal, it stores a new value into the memory.</p>
7127
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7128<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7129<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
7130 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
7131 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
7132 this integer type. While any bit width integer may be used, targets may only
7133 lower representations they support in hardware.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7134
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7135<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7136<p>This entire intrinsic must be executed atomically. It first loads the value
7137 in memory pointed to by <tt>ptr</tt> and compares it with the
7138 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
7139 memory. The loaded value is yielded in all cases. This provides the
7140 equivalent of an atomic compare-and-swap operation within the SSA
7141 framework.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7142
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7143<h5>Examples:</h5>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7144<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7145%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7146%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7147 store i32 4, %ptr
7148
7149%val1 = add i32 4, 4
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7150%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 4, %val1)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7151 <i>; yields {i32}:result1 = 4</i>
7152%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7153%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7154
7155%val2 = add i32 1, 1
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7156%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32(i32* %ptr, i32 5, %val2)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7157 <i>; yields {i32}:result2 = 8</i>
7158%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
7159
7160%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
7161</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7162
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7163</div>
7164
7165<!-- _______________________________________________________________________ -->
7166<div class="doc_subsubsection">
7167 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
7168</div>
7169<div class="doc_text">
7170<h5>Syntax:</h5>
7171
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7172<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
7173 integer bit width. Not all targets support all bit widths however.</p>
7174
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7175<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7176 declare i8 @llvm.atomic.swap.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;val&gt;)
7177 declare i16 @llvm.atomic.swap.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;val&gt;)
7178 declare i32 @llvm.atomic.swap.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;val&gt;)
7179 declare i64 @llvm.atomic.swap.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;val&gt;)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7180</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7181
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7182<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7183<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
7184 the value from memory. It then stores the value in <tt>val</tt> in the memory
7185 at <tt>ptr</tt>.</p>
7186
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7187<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7188<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
7189 the <tt>val</tt> argument and the result must be integers of the same bit
7190 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
7191 integer type. The targets may only lower integer representations they
7192 support.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7193
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7194<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7195<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
7196 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
7197 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7198
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7199<h5>Examples:</h5>
7200<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7201%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7202%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7203 store i32 4, %ptr
7204
7205%val1 = add i32 4, 4
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7206%result1 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val1)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7207 <i>; yields {i32}:result1 = 4</i>
7208%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
7209%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
7210
7211%val2 = add i32 1, 1
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7212%result2 = call i32 @llvm.atomic.swap.i32.p0i32(i32* %ptr, i32 %val2)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7213 <i>; yields {i32}:result2 = 8</i>
7214
7215%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
7216%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
7217</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7218
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7219</div>
7220
7221<!-- _______________________________________________________________________ -->
7222<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7223 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7224
7225</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7226
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7227<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7228
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7229<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7230<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
7231 any integer bit width. Not all targets support all bit widths however.</p>
7232
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7233<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7234 declare i8 @llvm.atomic.load.add.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7235 declare i16 @llvm.atomic.load.add.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7236 declare i32 @llvm.atomic.load.add.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7237 declare i64 @llvm.atomic.load.add.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7238</pre>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7239
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7240<h5>Overview:</h5>
7241<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
7242 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7243
7244<h5>Arguments:</h5>
7245<p>The intrinsic takes two arguments, the first a pointer to an integer value
7246 and the second an integer value. The result is also an integer value. These
7247 integer types can have any bit width, but they must all have the same bit
7248 width. The targets may only lower integer representations they support.</p>
7249
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7250<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7251<p>This intrinsic does a series of operations atomically. It first loads the
7252 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
7253 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7254
7255<h5>Examples:</h5>
7256<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7257%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7258%ptr = bitcast i8* %mallocP to i32*
7259 store i32 4, %ptr
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7260%result1 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 4)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7261 <i>; yields {i32}:result1 = 4</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7262%result2 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 2)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7263 <i>; yields {i32}:result2 = 8</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7264%result3 = call i32 @llvm.atomic.load.add.i32.p0i32(i32* %ptr, i32 5)
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7265 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7266%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7267</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7268
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]7269</div>
7270
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7271<!-- _______________________________________________________________________ -->
7272<div class="doc_subsubsection">
7273 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
7274
7275</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7276
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7277<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7278
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7279<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7280<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
7281 any integer bit width and for different address spaces. Not all targets
7282 support all bit widths however.</p>
7283
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7284<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7285 declare i8 @llvm.atomic.load.sub.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7286 declare i16 @llvm.atomic.load.sub.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7287 declare i32 @llvm.atomic.load.sub.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7288 declare i64 @llvm.atomic.load.sub.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7289</pre>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7290
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7291<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7292<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7293 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
7294
7295<h5>Arguments:</h5>
7296<p>The intrinsic takes two arguments, the first a pointer to an integer value
7297 and the second an integer value. The result is also an integer value. These
7298 integer types can have any bit width, but they must all have the same bit
7299 width. The targets may only lower integer representations they support.</p>
7300
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7301<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7302<p>This intrinsic does a series of operations atomically. It first loads the
7303 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
7304 result to <tt>ptr</tt>. It yields the original value stored
7305 at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7306
7307<h5>Examples:</h5>
7308<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7309%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7310%ptr = bitcast i8* %mallocP to i32*
7311 store i32 8, %ptr
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7312%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 4)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7313 <i>; yields {i32}:result1 = 8</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7314%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 2)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7315 <i>; yields {i32}:result2 = 4</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7316%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %ptr, i32 5)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7317 <i>; yields {i32}:result3 = 2</i>
7318%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
7319</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7320
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7321</div>
7322
7323<!-- _______________________________________________________________________ -->
7324<div class="doc_subsubsection">
7325 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
7326 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
7327 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
7328 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7329</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7330
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7331<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7332
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7333<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7334<p>These are overloaded intrinsics. You can
7335 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
7336 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
7337 bit width and for different address spaces. Not all targets support all bit
7338 widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7339
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7340<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7341 declare i8 @llvm.atomic.load.and.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7342 declare i16 @llvm.atomic.load.and.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7343 declare i32 @llvm.atomic.load.and.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7344 declare i64 @llvm.atomic.load.and.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7345</pre>
7346
7347<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7348 declare i8 @llvm.atomic.load.or.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7349 declare i16 @llvm.atomic.load.or.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7350 declare i32 @llvm.atomic.load.or.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7351 declare i64 @llvm.atomic.load.or.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7352</pre>
7353
7354<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7355 declare i8 @llvm.atomic.load.nand.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7356 declare i16 @llvm.atomic.load.nand.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7357 declare i32 @llvm.atomic.load.nand.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7358 declare i64 @llvm.atomic.load.nand.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7359</pre>
7360
7361<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7362 declare i8 @llvm.atomic.load.xor.i8.p0i32(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7363 declare i16 @llvm.atomic.load.xor.i16.p0i32(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7364 declare i32 @llvm.atomic.load.xor.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7365 declare i64 @llvm.atomic.load.xor.i64.p0i32(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7366</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7367
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7368<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7369<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
7370 the value stored in memory at <tt>ptr</tt>. It yields the original value
7371 at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7372
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7373<h5>Arguments:</h5>
7374<p>These intrinsics take two arguments, the first a pointer to an integer value
7375 and the second an integer value. The result is also an integer value. These
7376 integer types can have any bit width, but they must all have the same bit
7377 width. The targets may only lower integer representations they support.</p>
7378
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7379<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7380<p>These intrinsics does a series of operations atomically. They first load the
7381 value stored at <tt>ptr</tt>. They then do the bitwise
7382 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
7383 original value stored at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7384
7385<h5>Examples:</h5>
7386<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7387%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7388%ptr = bitcast i8* %mallocP to i32*
7389 store i32 0x0F0F, %ptr
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7390%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7391 <i>; yields {i32}:result0 = 0x0F0F</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7392%result1 = call i32 @llvm.atomic.load.and.i32.p0i32(i32* %ptr, i32 0xFF)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7393 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7394%result2 = call i32 @llvm.atomic.load.or.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7395 <i>; yields {i32}:result2 = 0xF0</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7396%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32(i32* %ptr, i32 0F)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7397 <i>; yields {i32}:result3 = FF</i>
7398%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
7399</pre>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7400
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7401</div>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7402
7403<!-- _______________________________________________________________________ -->
7404<div class="doc_subsubsection">
7405 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
7406 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
7407 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
7408 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7409</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7410
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7411<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7412
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7413<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7414<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
7415 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
7416 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
7417 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7418
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7419<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7420 declare i8 @llvm.atomic.load.max.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7421 declare i16 @llvm.atomic.load.max.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7422 declare i32 @llvm.atomic.load.max.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7423 declare i64 @llvm.atomic.load.max.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7424</pre>
7425
7426<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7427 declare i8 @llvm.atomic.load.min.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7428 declare i16 @llvm.atomic.load.min.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7429 declare i32 @llvm.atomic.load.min.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7430 declare i64 @llvm.atomic.load.min.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7431</pre>
7432
7433<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7434 declare i8 @llvm.atomic.load.umax.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7435 declare i16 @llvm.atomic.load.umax.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7436 declare i32 @llvm.atomic.load.umax.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7437 declare i64 @llvm.atomic.load.umax.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7438</pre>
7439
7440<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7441 declare i8 @llvm.atomic.load.umin.i8.p0i8(i8* &lt;ptr&gt;, i8 &lt;delta&gt;)
7442 declare i16 @llvm.atomic.load.umin.i16.p0i16(i16* &lt;ptr&gt;, i16 &lt;delta&gt;)
7443 declare i32 @llvm.atomic.load.umin.i32.p0i32(i32* &lt;ptr&gt;, i32 &lt;delta&gt;)
7444 declare i64 @llvm.atomic.load.umin.i64.p0i64(i64* &lt;ptr&gt;, i64 &lt;delta&gt;)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7445</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7446
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7447<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7448<p>These intrinsics takes the signed or unsigned minimum or maximum of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7449 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
7450 original value at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7451
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7452<h5>Arguments:</h5>
7453<p>These intrinsics take two arguments, the first a pointer to an integer value
7454 and the second an integer value. The result is also an integer value. These
7455 integer types can have any bit width, but they must all have the same bit
7456 width. The targets may only lower integer representations they support.</p>
7457
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7458<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7459<p>These intrinsics does a series of operations atomically. They first load the
7460 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
7461 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
7462 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7463
7464<h5>Examples:</h5>
7465<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7466%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7467%ptr = bitcast i8* %mallocP to i32*
7468 store i32 7, %ptr
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7469%result0 = call i32 @llvm.atomic.load.min.i32.p0i32(i32* %ptr, i32 -2)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7470 <i>; yields {i32}:result0 = 7</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7471%result1 = call i32 @llvm.atomic.load.max.i32.p0i32(i32* %ptr, i32 8)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7472 <i>; yields {i32}:result1 = -2</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7473%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32(i32* %ptr, i32 10)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7474 <i>; yields {i32}:result2 = 8</i>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7475%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32(i32* %ptr, i32 30)
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7476 <i>; yields {i32}:result3 = 8</i>
7477%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7478</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7479
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7480</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7481
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7482
7483<!-- ======================================================================= -->
7484<div class="doc_subsection">
7485 <a name="int_memorymarkers">Memory Use Markers</a>
7486</div>
7487
7488<div class="doc_text">
7489
7490<p>This class of intrinsics exists to information about the lifetime of memory
7491 objects and ranges where variables are immutable.</p>
7492
7493</div>
7494
7495<!-- _______________________________________________________________________ -->
7496<div class="doc_subsubsection">
7497 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7498</div>
7499
7500<div class="doc_text">
7501
7502<h5>Syntax:</h5>
7503<pre>
7504 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7505</pre>
7506
7507<h5>Overview:</h5>
7508<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7509 object's lifetime.</p>
7510
7511<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7512<p>The first argument is a constant integer representing the size of the
7513 object, or -1 if it is variable sized. The second argument is a pointer to
7514 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7515
7516<h5>Semantics:</h5>
7517<p>This intrinsic indicates that before this point in the code, the value of the
7518 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
Nick Lewyckyf8a5f302009-10-27 16:56:58 +0000[diff] [blame]7519 never be used and has an undefined value. A load from the pointer that
7520 precedes this intrinsic can be replaced with
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7521 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7522
7523</div>
7524
7525<!-- _______________________________________________________________________ -->
7526<div class="doc_subsubsection">
7527 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7528</div>
7529
7530<div class="doc_text">
7531
7532<h5>Syntax:</h5>
7533<pre>
7534 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7535</pre>
7536
7537<h5>Overview:</h5>
7538<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7539 object's lifetime.</p>
7540
7541<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7542<p>The first argument is a constant integer representing the size of the
7543 object, or -1 if it is variable sized. The second argument is a pointer to
7544 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7545
7546<h5>Semantics:</h5>
7547<p>This intrinsic indicates that after this point in the code, the value of the
7548 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7549 never be used and has an undefined value. Any stores into the memory object
7550 following this intrinsic may be removed as dead.
7551
7552</div>
7553
7554<!-- _______________________________________________________________________ -->
7555<div class="doc_subsubsection">
7556 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7557</div>
7558
7559<div class="doc_text">
7560
7561<h5>Syntax:</h5>
7562<pre>
7563 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;) readonly
7564</pre>
7565
7566<h5>Overview:</h5>
7567<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7568 a memory object will not change.</p>
7569
7570<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7571<p>The first argument is a constant integer representing the size of the
7572 object, or -1 if it is variable sized. The second argument is a pointer to
7573 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7574
7575<h5>Semantics:</h5>
7576<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7577 the return value, the referenced memory location is constant and
7578 unchanging.</p>
7579
7580</div>
7581
7582<!-- _______________________________________________________________________ -->
7583<div class="doc_subsubsection">
7584 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7585</div>
7586
7587<div class="doc_text">
7588
7589<h5>Syntax:</h5>
7590<pre>
7591 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7592</pre>
7593
7594<h5>Overview:</h5>
7595<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7596 a memory object are mutable.</p>
7597
7598<h5>Arguments:</h5>
7599<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7600 The second argument is a constant integer representing the size of the
7601 object, or -1 if it is variable sized and the third argument is a pointer
7602 to the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7603
7604<h5>Semantics:</h5>
7605<p>This intrinsic indicates that the memory is mutable again.</p>
7606
7607</div>
7608
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7609<!-- ======================================================================= -->
7610<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7611 <a name="int_general">General Intrinsics</a>
7612</div>
7613
7614<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7615
7616<p>This class of intrinsics is designed to be generic and has no specific
7617 purpose.</p>
7618
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7619</div>
7620
7621<!-- _______________________________________________________________________ -->
7622<div class="doc_subsubsection">
7623 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7624</div>
7625
7626<div class="doc_text">
7627
7628<h5>Syntax:</h5>
7629<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7630 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7631</pre>
7632
7633<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7634<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7635
7636<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7637<p>The first argument is a pointer to a value, the second is a pointer to a
7638 global string, the third is a pointer to a global string which is the source
7639 file name, and the last argument is the line number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7640
7641<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7642<p>This intrinsic allows annotation of local variables with arbitrary strings.
7643 This can be useful for special purpose optimizations that want to look for
7644 these annotations. These have no other defined use, they are ignored by code
7645 generation and optimization.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7646
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7647</div>
7648
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7649<!-- _______________________________________________________________________ -->
7650<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]7651 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7652</div>
7653
7654<div class="doc_text">
7655
7656<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7657<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7658 any integer bit width.</p>
7659
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7660<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7661 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7662 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7663 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7664 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
7665 declare i256 @llvm.annotation.i256(i256 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt;)
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7666</pre>
7667
7668<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7669<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7670
7671<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7672<p>The first argument is an integer value (result of some expression), the
7673 second is a pointer to a global string, the third is a pointer to a global
7674 string which is the source file name, and the last argument is the line
7675 number. It returns the value of the first argument.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7676
7677<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7678<p>This intrinsic allows annotations to be put on arbitrary expressions with
7679 arbitrary strings. This can be useful for special purpose optimizations that
7680 want to look for these annotations. These have no other defined use, they
7681 are ignored by code generation and optimization.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7682
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7683</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7684
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7685<!-- _______________________________________________________________________ -->
7686<div class="doc_subsubsection">
7687 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7688</div>
7689
7690<div class="doc_text">
7691
7692<h5>Syntax:</h5>
7693<pre>
7694 declare void @llvm.trap()
7695</pre>
7696
7697<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7698<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7699
7700<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7701<p>None.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7702
7703<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7704<p>This intrinsics is lowered to the target dependent trap instruction. If the
7705 target does not have a trap instruction, this intrinsic will be lowered to
7706 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7707
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7708</div>
7709
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7710<!-- _______________________________________________________________________ -->
7711<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]7712 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7713</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7714
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7715<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7716
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7717<h5>Syntax:</h5>
7718<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7719 declare void @llvm.stackprotector(i8* &lt;guard&gt;, i8** &lt;slot&gt;)
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7720</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7721
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7722<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7723<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7724 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7725 ensure that it is placed on the stack before local variables.</p>
7726
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7727<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7728<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7729 arguments. The first argument is the value loaded from the stack
7730 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7731 that has enough space to hold the value of the guard.</p>
7732
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7733<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7734<p>This intrinsic causes the prologue/epilogue inserter to force the position of
7735 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7736 stack. This is to ensure that if a local variable on the stack is
7737 overwritten, it will destroy the value of the guard. When the function exits,
7738 the guard on the stack is checked against the original guard. If they're
7739 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7740 function.</p>
7741
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7742</div>
7743
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7744<!-- _______________________________________________________________________ -->
7745<div class="doc_subsubsection">
7746 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7747</div>
7748
7749<div class="doc_text">
7750
7751<h5>Syntax:</h5>
7752<pre>
Dan Gohmand1ec0af2010-05-28 17:07:41 +0000[diff] [blame]7753 declare i32 @llvm.objectsize.i32(i8* &lt;object&gt;, i1 &lt;type&gt;)
7754 declare i64 @llvm.objectsize.i64(i8* &lt;object&gt;, i1 &lt;type&gt;)
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7755</pre>
7756
7757<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7758<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
Eric Christopher184f44d2010-01-08 21:42:39 +0000[diff] [blame]7759 to the optimizers to discover at compile time either a) when an
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7760 operation like memcpy will either overflow a buffer that corresponds to
7761 an object, or b) to determine that a runtime check for overflow isn't
7762 necessary. An object in this context means an allocation of a
Eric Christopher0101f9d2009-12-23 00:29:49 +0000[diff] [blame]7763 specific class, structure, array, or other object.</p>
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7764
7765<h5>Arguments:</h5>
7766<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher0101f9d2009-12-23 00:29:49 +0000[diff] [blame]7767 argument is a pointer to or into the <tt>object</tt>. The second argument
7768 is a boolean 0 or 1. This argument determines whether you want the
7769 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7770 1, variables are not allowed.</p>
7771
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7772<h5>Semantics:</h5>
7773<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7774 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7775 (depending on the <tt>type</tt> argument if the size cannot be determined
7776 at compile time.</p>
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7777
7778</div>
7779
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7780<!-- *********************************************************************** -->
7781<hr>
7782<address>
7783 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7784 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7785 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7786 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7787
7788 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7789 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7790 Last modified: $Date$
7791</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]7792
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7793</body>
7794</html>