Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / d07c83760d225f8bfe864939ce7d8d3fbdde1ea4 / . / docs / LangRef.html
blob: 916826aa462e924f5c833e03aa17bf3d041bc02a [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">
8 <meta name="description"
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>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]46 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]47 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]48 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]49 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
50 <li><a href="#datalayout">Data Layout</a></li>
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]51 <li><a href="#pointeraliasing">Pointer Aliasing Rules</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]52 </ol>
53 </li>
54 <li><a href="#typesystem">Type System</a>
55 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]56 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]57 <li><a href="#t_primitive">Primitive Types</a>
58 <ol>
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]59 <li><a href="#t_integer">Integer Type</a></li>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]60 <li><a href="#t_floating">Floating Point Types</a></li>
61 <li><a href="#t_void">Void Type</a></li>
62 <li><a href="#t_label">Label Type</a></li>
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]63 <li><a href="#t_metadata">Metadata Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]64 </ol>
65 </li>
66 <li><a href="#t_derived">Derived Types</a>
67 <ol>
68 <li><a href="#t_array">Array Type</a></li>
69 <li><a href="#t_function">Function Type</a></li>
70 <li><a href="#t_pointer">Pointer Type</a></li>
71 <li><a href="#t_struct">Structure Type</a></li>
72 <li><a href="#t_pstruct">Packed Structure Type</a></li>
73 <li><a href="#t_vector">Vector Type</a></li>
74 <li><a href="#t_opaque">Opaque Type</a></li>
75 </ol>
76 </li>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]77 <li><a href="#t_uprefs">Type Up-references</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]78 </ol>
79 </li>
80 <li><a href="#constants">Constants</a>
81 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]82 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]83 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]84 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
85 <li><a href="#undefvalues">Undefined Values</a></li>
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame^]86 <li><a href="#blockaddress">Address of Basic Block</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]87 <li><a href="#constantexprs">Constant Expressions</a></li>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]88 <li><a href="#metadata">Embedded Metadata</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]89 </ol>
90 </li>
91 <li><a href="#othervalues">Other Values</a>
92 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]93 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]94 </ol>
95 </li>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]96 <li><a href="#intrinsic_globals">Intrinsic Global Variables</a>
97 <ol>
98 <li><a href="#intg_used">The '<tt>llvm.used</tt>' Global Variable</a></li>
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]99 <li><a href="#intg_compiler_used">The '<tt>llvm.compiler.used</tt>'
100 Global Variable</a></li>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]101 <li><a href="#intg_global_ctors">The '<tt>llvm.global_ctors</tt>'
102 Global Variable</a></li>
103 <li><a href="#intg_global_dtors">The '<tt>llvm.global_dtors</tt>'
104 Global Variable</a></li>
105 </ol>
106 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]107 <li><a href="#instref">Instruction Reference</a>
108 <ol>
109 <li><a href="#terminators">Terminator Instructions</a>
110 <ol>
111 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
112 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
113 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]114 <li><a href="#i_indbr">'<tt>indbr</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]115 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
116 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
117 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
118 </ol>
119 </li>
120 <li><a href="#binaryops">Binary Operations</a>
121 <ol>
122 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]123 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]124 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]125 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]126 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]127 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]128 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
129 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
130 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
131 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
132 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
133 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
134 </ol>
135 </li>
136 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
137 <ol>
138 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
139 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
140 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
141 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
142 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
143 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
144 </ol>
145 </li>
146 <li><a href="#vectorops">Vector Operations</a>
147 <ol>
148 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
149 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
150 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
151 </ol>
152 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]153 <li><a href="#aggregateops">Aggregate Operations</a>
154 <ol>
155 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
156 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
157 </ol>
158 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]159 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
160 <ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]161 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
162 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
163 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
164 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
165 </ol>
166 </li>
167 <li><a href="#convertops">Conversion Operations</a>
168 <ol>
169 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
170 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
171 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
172 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
173 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
174 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
175 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
176 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
177 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
178 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
179 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
180 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
181 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]182 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]183 <li><a href="#otherops">Other Operations</a>
184 <ol>
185 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
186 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
187 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
188 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
189 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
190 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
191 </ol>
192 </li>
193 </ol>
194 </li>
195 <li><a href="#intrinsics">Intrinsic Functions</a>
196 <ol>
197 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
198 <ol>
199 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
200 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
201 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
202 </ol>
203 </li>
204 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
205 <ol>
206 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
207 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
208 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
209 </ol>
210 </li>
211 <li><a href="#int_codegen">Code Generator Intrinsics</a>
212 <ol>
213 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
214 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
215 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
216 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
217 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
218 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
219 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
220 </ol>
221 </li>
222 <li><a href="#int_libc">Standard C Library Intrinsics</a>
223 <ol>
224 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
225 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
226 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
227 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
228 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]229 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
230 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
231 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]232 </ol>
233 </li>
234 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
235 <ol>
236 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
237 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
238 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
239 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]240 </ol>
241 </li>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]242 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
243 <ol>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]244 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
245 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
246 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
247 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
248 <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]249 <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]250 </ol>
251 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]252 <li><a href="#int_debugger">Debugger intrinsics</a></li>
253 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]254 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]255 <ol>
256 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]257 </ol>
258 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]259 <li><a href="#int_atomics">Atomic intrinsics</a>
260 <ol>
261 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
262 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
263 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
264 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
265 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
266 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
267 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
268 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
269 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
270 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
271 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
272 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
273 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
274 </ol>
275 </li>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]276 <li><a href="#int_memorymarkers">Memory Use Markers</a>
277 <ol>
278 <li><a href="#int_lifetime_start"><tt>llvm.lifetime.start</tt></a></li>
279 <li><a href="#int_lifetime_end"><tt>llvm.lifetime.end</tt></a></li>
280 <li><a href="#int_invariant_start"><tt>llvm.invariant.start</tt></a></li>
281 <li><a href="#int_invariant_end"><tt>llvm.invariant.end</tt></a></li>
282 </ol>
283 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]284 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]285 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]286 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]287 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]288 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]289 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]290 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]291 '<tt>llvm.trap</tt>' Intrinsic</a></li>
292 <li><a href="#int_stackprotector">
293 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]294 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]295 </li>
296 </ol>
297 </li>
298</ol>
299
300<div class="doc_author">
301 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
302 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
303</div>
304
305<!-- *********************************************************************** -->
306<div class="doc_section"> <a name="abstract">Abstract </a></div>
307<!-- *********************************************************************** -->
308
309<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]310
311<p>This document is a reference manual for the LLVM assembly language. LLVM is
312 a Static Single Assignment (SSA) based representation that provides type
313 safety, low-level operations, flexibility, and the capability of representing
314 'all' high-level languages cleanly. It is the common code representation
315 used throughout all phases of the LLVM compilation strategy.</p>
316
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]317</div>
318
319<!-- *********************************************************************** -->
320<div class="doc_section"> <a name="introduction">Introduction</a> </div>
321<!-- *********************************************************************** -->
322
323<div class="doc_text">
324
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]325<p>The LLVM code representation is designed to be used in three different forms:
326 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
327 for fast loading by a Just-In-Time compiler), and as a human readable
328 assembly language representation. This allows LLVM to provide a powerful
329 intermediate representation for efficient compiler transformations and
330 analysis, while providing a natural means to debug and visualize the
331 transformations. The three different forms of LLVM are all equivalent. This
332 document describes the human readable representation and notation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]333
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]334<p>The LLVM representation aims to be light-weight and low-level while being
335 expressive, typed, and extensible at the same time. It aims to be a
336 "universal IR" of sorts, by being at a low enough level that high-level ideas
337 may be cleanly mapped to it (similar to how microprocessors are "universal
338 IR's", allowing many source languages to be mapped to them). By providing
339 type information, LLVM can be used as the target of optimizations: for
340 example, through pointer analysis, it can be proven that a C automatic
341 variable is never accessed outside of the current function... allowing it to
342 be promoted to a simple SSA value instead of a memory location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]343
344</div>
345
346<!-- _______________________________________________________________________ -->
347<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
348
349<div class="doc_text">
350
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]351<p>It is important to note that this document describes 'well formed' LLVM
352 assembly language. There is a difference between what the parser accepts and
353 what is considered 'well formed'. For example, the following instruction is
354 syntactically okay, but not well formed:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]355
356<div class="doc_code">
357<pre>
358%x = <a href="#i_add">add</a> i32 1, %x
359</pre>
360</div>
361
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]362<p>...because the definition of <tt>%x</tt> does not dominate all of its
363 uses. The LLVM infrastructure provides a verification pass that may be used
364 to verify that an LLVM module is well formed. This pass is automatically run
365 by the parser after parsing input assembly and by the optimizer before it
366 outputs bitcode. The violations pointed out by the verifier pass indicate
367 bugs in transformation passes or input to the parser.</p>
368
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]369</div>
370
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]371<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]372
373<!-- *********************************************************************** -->
374<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
375<!-- *********************************************************************** -->
376
377<div class="doc_text">
378
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]379<p>LLVM identifiers come in two basic types: global and local. Global
380 identifiers (functions, global variables) begin with the <tt>'@'</tt>
381 character. Local identifiers (register names, types) begin with
382 the <tt>'%'</tt> character. Additionally, there are three different formats
383 for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]384
385<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]386 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]387 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
388 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
389 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
390 other characters in their names can be surrounded with quotes. Special
391 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
392 ASCII code for the character in hexadecimal. In this way, any character
393 can be used in a name value, even quotes themselves.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]394
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]395 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]396 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]397
398 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]399 constants</a>, below.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]400</ol>
401
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]402<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]403 don't need to worry about name clashes with reserved words, and the set of
404 reserved words may be expanded in the future without penalty. Additionally,
405 unnamed identifiers allow a compiler to quickly come up with a temporary
406 variable without having to avoid symbol table conflicts.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]407
408<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]409 languages. There are keywords for different opcodes
410 ('<tt><a href="#i_add">add</a></tt>',
411 '<tt><a href="#i_bitcast">bitcast</a></tt>',
412 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
413 ('<tt><a href="#t_void">void</a></tt>',
414 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
415 reserved words cannot conflict with variable names, because none of them
416 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]417
418<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]419 '<tt>%X</tt>' by 8:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]420
421<p>The easy way:</p>
422
423<div class="doc_code">
424<pre>
425%result = <a href="#i_mul">mul</a> i32 %X, 8
426</pre>
427</div>
428
429<p>After strength reduction:</p>
430
431<div class="doc_code">
432<pre>
433%result = <a href="#i_shl">shl</a> i32 %X, i8 3
434</pre>
435</div>
436
437<p>And the hard way:</p>
438
439<div class="doc_code">
440<pre>
441<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
442<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
443%result = <a href="#i_add">add</a> i32 %1, %1
444</pre>
445</div>
446
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]447<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
448 lexical features of LLVM:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]449
450<ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]451 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]452 line.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]453
454 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]455 assigned to a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]456
457 <li>Unnamed temporaries are numbered sequentially</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]458</ol>
459
460<p>...and it also shows a convention that we follow in this document. When
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]461 demonstrating instructions, we will follow an instruction with a comment that
462 defines the type and name of value produced. Comments are shown in italic
463 text.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]464
465</div>
466
467<!-- *********************************************************************** -->
468<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
469<!-- *********************************************************************** -->
470
471<!-- ======================================================================= -->
472<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
473</div>
474
475<div class="doc_text">
476
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]477<p>LLVM programs are composed of "Module"s, each of which is a translation unit
478 of the input programs. Each module consists of functions, global variables,
479 and symbol table entries. Modules may be combined together with the LLVM
480 linker, which merges function (and global variable) definitions, resolves
481 forward declarations, and merges symbol table entries. Here is an example of
482 the "hello world" module:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]483
484<div class="doc_code">
485<pre><i>; Declare the string constant as a global constant...</i>
486<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
487 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
488
489<i>; External declaration of the puts function</i>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]490<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]491
492<i>; Definition of main function</i>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]493define i32 @main() { <i>; i32()* </i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]494 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]495 %cast210 = <a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]496 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]497
498 <i>; Call puts function to write out the string to stdout...</i>
499 <a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]500 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]501 <a
502 href="#i_ret">ret</a> i32 0<br>}<br>
503</pre>
504</div>
505
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]506<p>This example is made up of a <a href="#globalvars">global variable</a> named
507 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function, and
508 a <a href="#functionstructure">function definition</a> for
509 "<tt>main</tt>".</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]510
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]511<p>In general, a module is made up of a list of global values, where both
512 functions and global variables are global values. Global values are
513 represented by a pointer to a memory location (in this case, a pointer to an
514 array of char, and a pointer to a function), and have one of the
515 following <a href="#linkage">linkage types</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]516
517</div>
518
519<!-- ======================================================================= -->
520<div class="doc_subsection">
521 <a name="linkage">Linkage Types</a>
522</div>
523
524<div class="doc_text">
525
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]526<p>All Global Variables and Functions have one of the following types of
527 linkage:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]528
529<dl>
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]530 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]531 <dd>Global values with private linkage are only directly accessible by objects
532 in the current module. In particular, linking code into a module with an
533 private global value may cause the private to be renamed as necessary to
534 avoid collisions. Because the symbol is private to the module, all
535 references can be updated. This doesn't show up in any symbol table in the
536 object file.</dd>
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]537
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]538 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt>: </dt>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]539 <dd>Similar to private, but the symbol is passed through the assembler and
Chris Lattner06c1ecc2009-08-24 04:32:16 +0000[diff] [blame]540 removed by the linker after evaluation. Note that (unlike private
541 symbols) linker_private symbols are subject to coalescing by the linker:
542 weak symbols get merged and redefinitions are rejected. However, unlike
543 normal strong symbols, they are removed by the linker from the final
544 linked image (executable or dynamic library).</dd>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]545
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]546 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]547 <dd>Similar to private, but the value shows as a local symbol
548 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
549 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]550
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]551 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt>: </dt>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]552 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]553 into the object file corresponding to the LLVM module. They exist to
554 allow inlining and other optimizations to take place given knowledge of
555 the definition of the global, which is known to be somewhere outside the
556 module. Globals with <tt>available_externally</tt> linkage are allowed to
557 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
558 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]559
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]560 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]561 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]562 the same name when linkage occurs. This is typically used to implement
563 inline functions, templates, or other code which must be generated in each
564 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
565 allowed to be discarded.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]566
567 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]568 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
569 <tt>linkonce</tt> linkage, except that unreferenced globals with
570 <tt>weak</tt> linkage may not be discarded. This is used for globals that
571 are declared "weak" in C source code.</dd>
572
573 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
574 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
575 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
576 global scope.
577 Symbols with "<tt>common</tt>" linkage are merged in the same way as
578 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattner83c4a712009-08-05 05:41:44 +0000[diff] [blame]579 <tt>common</tt> symbols may not have an explicit section,
580 must have a zero initializer, and may not be marked '<a
581 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
582 have common linkage.</dd>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]583
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]584
585 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]586 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]587 pointer to array type. When two global variables with appending linkage
588 are linked together, the two global arrays are appended together. This is
589 the LLVM, typesafe, equivalent of having the system linker append together
590 "sections" with identical names when .o files are linked.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]591
592 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]593 <dd>The semantics of this linkage follow the ELF object file model: the symbol
594 is weak until linked, if not linked, the symbol becomes null instead of
595 being an undefined reference.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]596
Chris Lattner0fee5c22009-10-10 18:26:06 +0000[diff] [blame]597 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt>: </dt>
598 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt>: </dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]599 <dd>Some languages allow differing globals to be merged, such as two functions
600 with different semantics. Other languages, such as <tt>C++</tt>, ensure
601 that only equivalent globals are ever merged (the "one definition rule" -
602 "ODR"). Such languages can use the <tt>linkonce_odr</tt>
603 and <tt>weak_odr</tt> linkage types to indicate that the global will only
604 be merged with equivalent globals. These linkage types are otherwise the
605 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]606
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]607 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]608 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]609 visible, meaning that it participates in linkage and can be used to
610 resolve external symbol references.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]611</dl>
612
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]613<p>The next two types of linkage are targeted for Microsoft Windows platform
614 only. They are designed to support importing (exporting) symbols from (to)
615 DLLs (Dynamic Link Libraries).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]616
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]617<dl>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]618 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]619 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]620 or variable via a global pointer to a pointer that is set up by the DLL
621 exporting the symbol. On Microsoft Windows targets, the pointer name is
622 formed by combining <code>__imp_</code> and the function or variable
623 name.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]624
625 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]626 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]627 pointer to a pointer in a DLL, so that it can be referenced with the
628 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
629 name is formed by combining <code>__imp_</code> and the function or
630 variable name.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]631</dl>
632
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]633<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
634 another module defined a "<tt>.LC0</tt>" variable and was linked with this
635 one, one of the two would be renamed, preventing a collision. Since
636 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
637 declarations), they are accessible outside of the current module.</p>
638
639<p>It is illegal for a function <i>declaration</i> to have any linkage type
640 other than "externally visible", <tt>dllimport</tt>
641 or <tt>extern_weak</tt>.</p>
642
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]643<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]644 or <tt>weak_odr</tt> linkages.</p>
645
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]646</div>
647
648<!-- ======================================================================= -->
649<div class="doc_subsection">
650 <a name="callingconv">Calling Conventions</a>
651</div>
652
653<div class="doc_text">
654
655<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]656 and <a href="#i_invoke">invokes</a> can all have an optional calling
657 convention specified for the call. The calling convention of any pair of
658 dynamic caller/callee must match, or the behavior of the program is
659 undefined. The following calling conventions are supported by LLVM, and more
660 may be added in the future:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]661
662<dl>
663 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]664 <dd>This calling convention (the default if no other calling convention is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]665 specified) matches the target C calling conventions. This calling
666 convention supports varargs function calls and tolerates some mismatch in
667 the declared prototype and implemented declaration of the function (as
668 does normal C).</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]669
670 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]671 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]672 (e.g. by passing things in registers). This calling convention allows the
673 target to use whatever tricks it wants to produce fast code for the
674 target, without having to conform to an externally specified ABI
675 (Application Binary Interface). Implementations of this convention should
676 allow arbitrary <a href="CodeGenerator.html#tailcallopt">tail call
677 optimization</a> to be supported. This calling convention does not
678 support varargs and requires the prototype of all callees to exactly match
679 the prototype of the function definition.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]680
681 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]682 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]683 as possible under the assumption that the call is not commonly executed.
684 As such, these calls often preserve all registers so that the call does
685 not break any live ranges in the caller side. This calling convention
686 does not support varargs and requires the prototype of all callees to
687 exactly match the prototype of the function definition.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]688
689 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]690 <dd>Any calling convention may be specified by number, allowing
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]691 target-specific calling conventions to be used. Target specific calling
692 conventions start at 64.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]693</dl>
694
695<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]696 support Pascal conventions or any other well-known target-independent
697 convention.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]698
699</div>
700
701<!-- ======================================================================= -->
702<div class="doc_subsection">
703 <a name="visibility">Visibility Styles</a>
704</div>
705
706<div class="doc_text">
707
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]708<p>All Global Variables and Functions have one of the following visibility
709 styles:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]710
711<dl>
712 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]713 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]714 that the declaration is visible to other modules and, in shared libraries,
715 means that the declared entity may be overridden. On Darwin, default
716 visibility means that the declaration is visible to other modules. Default
717 visibility corresponds to "external linkage" in the language.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]718
719 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]720 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]721 object if they are in the same shared object. Usually, hidden visibility
722 indicates that the symbol will not be placed into the dynamic symbol
723 table, so no other module (executable or shared library) can reference it
724 directly.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]725
726 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]727 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]728 the dynamic symbol table, but that references within the defining module
729 will bind to the local symbol. That is, the symbol cannot be overridden by
730 another module.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]731</dl>
732
733</div>
734
735<!-- ======================================================================= -->
736<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]737 <a name="namedtypes">Named Types</a>
738</div>
739
740<div class="doc_text">
741
742<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]743 it easier to read the IR and make the IR more condensed (particularly when
744 recursive types are involved). An example of a name specification is:</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]745
746<div class="doc_code">
747<pre>
748%mytype = type { %mytype*, i32 }
749</pre>
750</div>
751
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]752<p>You may give a name to any <a href="#typesystem">type</a> except
753 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
754 is expected with the syntax "%mytype".</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]755
756<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]757 and that you can therefore specify multiple names for the same type. This
758 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
759 uses structural typing, the name is not part of the type. When printing out
760 LLVM IR, the printer will pick <em>one name</em> to render all types of a
761 particular shape. This means that if you have code where two different
762 source types end up having the same LLVM type, that the dumper will sometimes
763 print the "wrong" or unexpected type. This is an important design point and
764 isn't going to change.</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]765
766</div>
767
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]768<!-- ======================================================================= -->
769<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]770 <a name="globalvars">Global Variables</a>
771</div>
772
773<div class="doc_text">
774
775<p>Global variables define regions of memory allocated at compilation time
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]776 instead of run-time. Global variables may optionally be initialized, may
777 have an explicit section to be placed in, and may have an optional explicit
778 alignment specified. A variable may be defined as "thread_local", which
779 means that it will not be shared by threads (each thread will have a
780 separated copy of the variable). A variable may be defined as a global
781 "constant," which indicates that the contents of the variable
782 will <b>never</b> be modified (enabling better optimization, allowing the
783 global data to be placed in the read-only section of an executable, etc).
784 Note that variables that need runtime initialization cannot be marked
785 "constant" as there is a store to the variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]786
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]787<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
788 constant, even if the final definition of the global is not. This capability
789 can be used to enable slightly better optimization of the program, but
790 requires the language definition to guarantee that optimizations based on the
791 'constantness' are valid for the translation units that do not include the
792 definition.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]793
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]794<p>As SSA values, global variables define pointer values that are in scope
795 (i.e. they dominate) all basic blocks in the program. Global variables
796 always define a pointer to their "content" type because they describe a
797 region of memory, and all memory objects in LLVM are accessed through
798 pointers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]799
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]800<p>A global variable may be declared to reside in a target-specific numbered
801 address space. For targets that support them, address spaces may affect how
802 optimizations are performed and/or what target instructions are used to
803 access the variable. The default address space is zero. The address space
804 qualifier must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]805
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]806<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]807 supports it, it will emit globals to the section specified.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]808
809<p>An explicit alignment may be specified for a global. If not present, or if
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]810 the alignment is set to zero, the alignment of the global is set by the
811 target to whatever it feels convenient. If an explicit alignment is
812 specified, the global is forced to have at least that much alignment. All
813 alignments must be a power of 2.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]814
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]815<p>For example, the following defines a global in a numbered address space with
816 an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]817
818<div class="doc_code">
819<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]820@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]821</pre>
822</div>
823
824</div>
825
826
827<!-- ======================================================================= -->
828<div class="doc_subsection">
829 <a name="functionstructure">Functions</a>
830</div>
831
832<div class="doc_text">
833
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]834<p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an
835 optional <a href="#linkage">linkage type</a>, an optional
836 <a href="#visibility">visibility style</a>, an optional
837 <a href="#callingconv">calling convention</a>, a return type, an optional
838 <a href="#paramattrs">parameter attribute</a> for the return type, a function
839 name, a (possibly empty) argument list (each with optional
840 <a href="#paramattrs">parameter attributes</a>), optional
841 <a href="#fnattrs">function attributes</a>, an optional section, an optional
842 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
843 curly brace, a list of basic blocks, and a closing curly brace.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]844
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]845<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
846 optional <a href="#linkage">linkage type</a>, an optional
847 <a href="#visibility">visibility style</a>, an optional
848 <a href="#callingconv">calling convention</a>, a return type, an optional
849 <a href="#paramattrs">parameter attribute</a> for the return type, a function
850 name, a possibly empty list of arguments, an optional alignment, and an
851 optional <a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]852
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]853<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]854 (Control Flow Graph) for the function. Each basic block may optionally start
855 with a label (giving the basic block a symbol table entry), contains a list
856 of instructions, and ends with a <a href="#terminators">terminator</a>
857 instruction (such as a branch or function return).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]858
859<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]860 executed on entrance to the function, and it is not allowed to have
861 predecessor basic blocks (i.e. there can not be any branches to the entry
862 block of a function). Because the block can have no predecessors, it also
863 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]864
865<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]866 supports it, it will emit functions to the section specified.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]867
868<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]869 the alignment is set to zero, the alignment of the function is set by the
870 target to whatever it feels convenient. If an explicit alignment is
871 specified, the function is forced to have at least that much alignment. All
872 alignments must be a power of 2.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]873
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]874<h5>Syntax:</h5>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]875<div class="doc_code">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]876<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]877define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]878 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
879 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
880 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
881 [<a href="#gc">gc</a>] { ... }
882</pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]883</div>
884
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]885</div>
886
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]887<!-- ======================================================================= -->
888<div class="doc_subsection">
889 <a name="aliasstructure">Aliases</a>
890</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]891
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]892<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]893
894<p>Aliases act as "second name" for the aliasee value (which can be either
895 function, global variable, another alias or bitcast of global value). Aliases
896 may have an optional <a href="#linkage">linkage type</a>, and an
897 optional <a href="#visibility">visibility style</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]898
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]899<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]900<div class="doc_code">
901<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]902@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]903</pre>
904</div>
905
906</div>
907
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]908<!-- ======================================================================= -->
909<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]910
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]911<div class="doc_text">
912
913<p>The return type and each parameter of a function type may have a set of
914 <i>parameter attributes</i> associated with them. Parameter attributes are
915 used to communicate additional information about the result or parameters of
916 a function. Parameter attributes are considered to be part of the function,
917 not of the function type, so functions with different parameter attributes
918 can have the same function type.</p>
919
920<p>Parameter attributes are simple keywords that follow the type specified. If
921 multiple parameter attributes are needed, they are space separated. For
922 example:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]923
924<div class="doc_code">
925<pre>
Nick Lewycky3022a742009-02-15 23:06:14 +0000[diff] [blame]926declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]927declare i32 @atoi(i8 zeroext)
928declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]929</pre>
930</div>
931
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]932<p>Note that any attributes for the function result (<tt>nounwind</tt>,
933 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]934
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]935<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]936
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]937<dl>
938 <dt><tt>zeroext</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]939 <dd>This indicates to the code generator that the parameter or return value
940 should be zero-extended to a 32-bit value by the caller (for a parameter)
941 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]942
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]943 <dt><tt>signext</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]944 <dd>This indicates to the code generator that the parameter or return value
945 should be sign-extended to a 32-bit value by the caller (for a parameter)
946 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]947
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]948 <dt><tt>inreg</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]949 <dd>This indicates that this parameter or return value should be treated in a
950 special target-dependent fashion during while emitting code for a function
951 call or return (usually, by putting it in a register as opposed to memory,
952 though some targets use it to distinguish between two different kinds of
953 registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]954
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]955 <dt><tt><a name="byval">byval</a></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]956 <dd>This indicates that the pointer parameter should really be passed by value
957 to the function. The attribute implies that a hidden copy of the pointee
958 is made between the caller and the callee, so the callee is unable to
959 modify the value in the callee. This attribute is only valid on LLVM
960 pointer arguments. It is generally used to pass structs and arrays by
961 value, but is also valid on pointers to scalars. The copy is considered
962 to belong to the caller not the callee (for example,
963 <tt><a href="#readonly">readonly</a></tt> functions should not write to
964 <tt>byval</tt> parameters). This is not a valid attribute for return
965 values. The byval attribute also supports specifying an alignment with
966 the align attribute. This has a target-specific effect on the code
967 generator that usually indicates a desired alignment for the synthesized
968 stack slot.</dd>
969
970 <dt><tt>sret</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]971 <dd>This indicates that the pointer parameter specifies the address of a
972 structure that is the return value of the function in the source program.
973 This pointer must be guaranteed by the caller to be valid: loads and
974 stores to the structure may be assumed by the callee to not to trap. This
975 may only be applied to the first parameter. This is not a valid attribute
976 for return values. </dd>
977
978 <dt><tt>noalias</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]979 <dd>This indicates that the pointer does not alias any global or any other
980 parameter. The caller is responsible for ensuring that this is the
981 case. On a function return value, <tt>noalias</tt> additionally indicates
982 that the pointer does not alias any other pointers visible to the
983 caller. For further details, please see the discussion of the NoAlias
984 response in
985 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
986 analysis</a>.</dd>
987
988 <dt><tt>nocapture</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]989 <dd>This indicates that the callee does not make any copies of the pointer
990 that outlive the callee itself. This is not a valid attribute for return
991 values.</dd>
992
993 <dt><tt>nest</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]994 <dd>This indicates that the pointer parameter can be excised using the
995 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
996 attribute for return values.</dd>
997</dl>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]998
999</div>
1000
1001<!-- ======================================================================= -->
1002<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1003 <a name="gc">Garbage Collector Names</a>
1004</div>
1005
1006<div class="doc_text">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1007
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1008<p>Each function may specify a garbage collector name, which is simply a
1009 string:</p>
1010
1011<div class="doc_code">
1012<pre>
1013define void @f() gc "name" { ...
1014</pre>
1015</div>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1016
1017<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1018 collector which will cause the compiler to alter its output in order to
1019 support the named garbage collection algorithm.</p>
1020
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1021</div>
1022
1023<!-- ======================================================================= -->
1024<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1025 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1026</div>
1027
1028<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1029
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1030<p>Function attributes are set to communicate additional information about a
1031 function. Function attributes are considered to be part of the function, not
1032 of the function type, so functions with different parameter attributes can
1033 have the same function type.</p>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1034
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1035<p>Function attributes are simple keywords that follow the type specified. If
1036 multiple attributes are needed, they are space separated. For example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1037
1038<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1039<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1040define void @f() noinline { ... }
1041define void @f() alwaysinline { ... }
1042define void @f() alwaysinline optsize { ... }
1043define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1044</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1045</div>
1046
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1047<dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1048 <dt><tt>alwaysinline</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1049 <dd>This attribute indicates that the inliner should attempt to inline this
1050 function into callers whenever possible, ignoring any active inlining size
1051 threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1052
Dale Johannesen7fd6b0f2009-08-26 01:08:21 +0000[diff] [blame]1053 <dt><tt>inlinehint</tt></dt>
1054 <dd>This attribute indicates that the source code contained a hint that inlining
1055 this function is desirable (such as the "inline" keyword in C/C++). It
1056 is just a hint; it imposes no requirements on the inliner.</dd>
1057
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1058 <dt><tt>noinline</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1059 <dd>This attribute indicates that the inliner should never inline this
1060 function in any situation. This attribute may not be used together with
1061 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1062
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1063 <dt><tt>optsize</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1064 <dd>This attribute suggests that optimization passes and code generator passes
1065 make choices that keep the code size of this function low, and otherwise
1066 do optimizations specifically to reduce code size.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1067
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1068 <dt><tt>noreturn</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1069 <dd>This function attribute indicates that the function never returns
1070 normally. This produces undefined behavior at runtime if the function
1071 ever does dynamically return.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1072
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1073 <dt><tt>nounwind</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1074 <dd>This function attribute indicates that the function never returns with an
1075 unwind or exceptional control flow. If the function does unwind, its
1076 runtime behavior is undefined.</dd>
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1077
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1078 <dt><tt>readnone</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1079 <dd>This attribute indicates that the function computes its result (or decides
1080 to unwind an exception) based strictly on its arguments, without
1081 dereferencing any pointer arguments or otherwise accessing any mutable
1082 state (e.g. memory, control registers, etc) visible to caller functions.
1083 It does not write through any pointer arguments
1084 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1085 changes any state visible to callers. This means that it cannot unwind
1086 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1087 could use the <tt>unwind</tt> instruction.</dd>
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1088
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1089 <dt><tt><a name="readonly">readonly</a></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1090 <dd>This attribute indicates that the function does not write through any
1091 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1092 arguments) or otherwise modify any state (e.g. memory, control registers,
1093 etc) visible to caller functions. It may dereference pointer arguments
1094 and read state that may be set in the caller. A readonly function always
1095 returns the same value (or unwinds an exception identically) when called
1096 with the same set of arguments and global state. It cannot unwind an
1097 exception by calling the <tt>C++</tt> exception throwing methods, but may
1098 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovedd7d112009-07-17 18:07:26 +0000[diff] [blame]1099
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1100 <dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1101 <dd>This attribute indicates that the function should emit a stack smashing
1102 protector. It is in the form of a "canary"&mdash;a random value placed on
1103 the stack before the local variables that's checked upon return from the
1104 function to see if it has been overwritten. A heuristic is used to
1105 determine if a function needs stack protectors or not.<br>
1106<br>
1107 If a function that has an <tt>ssp</tt> attribute is inlined into a
1108 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1109 function will have an <tt>ssp</tt> attribute.</dd>
1110
1111 <dt><tt>sspreq</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1112 <dd>This attribute indicates that the function should <em>always</em> emit a
1113 stack smashing protector. This overrides
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1114 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1115<br>
1116 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1117 function that doesn't have an <tt>sspreq</tt> attribute or which has
1118 an <tt>ssp</tt> attribute, then the resulting function will have
1119 an <tt>sspreq</tt> attribute.</dd>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1120
1121 <dt><tt>noredzone</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1122 <dd>This attribute indicates that the code generator should not use a red
1123 zone, even if the target-specific ABI normally permits it.</dd>
1124
1125 <dt><tt>noimplicitfloat</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1126 <dd>This attributes disables implicit floating point instructions.</dd>
1127
1128 <dt><tt>naked</tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1129 <dd>This attribute disables prologue / epilogue emission for the function.
1130 This can have very system-specific consequences.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1131</dl>
1132
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1133</div>
1134
1135<!-- ======================================================================= -->
1136<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1137 <a name="moduleasm">Module-Level Inline Assembly</a>
1138</div>
1139
1140<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1141
1142<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1143 the GCC "file scope inline asm" blocks. These blocks are internally
1144 concatenated by LLVM and treated as a single unit, but may be separated in
1145 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1146
1147<div class="doc_code">
1148<pre>
1149module asm "inline asm code goes here"
1150module asm "more can go here"
1151</pre>
1152</div>
1153
1154<p>The strings can contain any character by escaping non-printable characters.
1155 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]1156 for the number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1157
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1158<p>The inline asm code is simply printed to the machine code .s file when
1159 assembly code is generated.</p>
1160
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1161</div>
1162
1163<!-- ======================================================================= -->
1164<div class="doc_subsection">
1165 <a name="datalayout">Data Layout</a>
1166</div>
1167
1168<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1169
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1170<p>A module may specify a target specific data layout string that specifies how
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1171 data is to be laid out in memory. The syntax for the data layout is
1172 simply:</p>
1173
1174<div class="doc_code">
1175<pre>
1176target datalayout = "<i>layout specification</i>"
1177</pre>
1178</div>
1179
1180<p>The <i>layout specification</i> consists of a list of specifications
1181 separated by the minus sign character ('-'). Each specification starts with
1182 a letter and may include other information after the letter to define some
1183 aspect of the data layout. The specifications accepted are as follows:</p>
1184
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1185<dl>
1186 <dt><tt>E</tt></dt>
1187 <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]1188 bits with the most significance have the lowest address location.</dd>
1189
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1190 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1191 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1192 the bits with the least significance have the lowest address
1193 location.</dd>
1194
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1195 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1196 <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]1197 <i>preferred</i> alignments. All sizes are in bits. Specifying
1198 the <i>pref</i> alignment is optional. If omitted, the
1199 preceding <tt>:</tt> should be omitted too.</dd>
1200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1201 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1202 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1203 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1204
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1205 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1206 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1207 <i>size</i>.</dd>
1208
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1209 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1210 <dd>This specifies the alignment for a floating point type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1211 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1212 (double).</dd>
1213
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1214 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1215 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1216 <i>size</i>.</dd>
1217
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1218 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1219 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1220 <i>size</i>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1221</dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1222
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1223<p>When constructing the data layout for a given target, LLVM starts with a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1224 default set of specifications which are then (possibly) overriden by the
1225 specifications in the <tt>datalayout</tt> keyword. The default specifications
1226 are given in this list:</p>
1227
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1228<ul>
1229 <li><tt>E</tt> - big endian</li>
1230 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1231 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1232 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1233 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1234 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1235 <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]1236 alignment of 64-bits</li>
1237 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1238 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1239 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1240 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1241 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1242 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1243</ul>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1244
1245<p>When LLVM is determining the alignment for a given type, it uses the
1246 following rules:</p>
1247
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1248<ol>
1249 <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]1250 specification is used.</li>
1251
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1252 <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]1253 smallest integer type that is larger than the bitwidth of the sought type
1254 is used. If none of the specifications are larger than the bitwidth then
1255 the the largest integer type is used. For example, given the default
1256 specifications above, the i7 type will use the alignment of i8 (next
1257 largest) while both i65 and i256 will use the alignment of i64 (largest
1258 specified).</li>
1259
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1260 <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]1261 largest vector type that is smaller than the sought vector type will be
1262 used as a fall back. This happens because &lt;128 x double&gt; can be
1263 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1264</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1265
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1266</div>
1267
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1268<!-- ======================================================================= -->
1269<div class="doc_subsection">
1270 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1271</div>
1272
1273<div class="doc_text">
1274
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1275<p>Any memory access must be done through a pointer value associated
Andreas Bolka23bece42009-07-27 20:37:10 +0000[diff] [blame]1276with an address range of the memory access, otherwise the behavior
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1277is undefined. Pointer values are associated with address ranges
1278according to the following rules:</p>
1279
1280<ul>
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1281 <li>A pointer value formed from a
1282 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1283 is associated with the addresses associated with the first operand
1284 of the <tt>getelementptr</tt>.</li>
1285 <li>An address of a global variable is associated with the address
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1286 range of the variable's storage.</li>
1287 <li>The result value of an allocation instruction is associated with
1288 the address range of the allocated storage.</li>
1289 <li>A null pointer in the default address-space is associated with
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1290 no address.</li>
1291 <li>A pointer value formed by an
1292 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1293 address ranges of all pointer values that contribute (directly or
1294 indirectly) to the computation of the pointer's value.</li>
1295 <li>The result value of a
1296 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1297 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1298 <li>An integer constant other than zero or a pointer value returned
1299 from a function not defined within LLVM may be associated with address
1300 ranges allocated through mechanisms other than those provided by
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1301 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1302 allocated by mechanisms provided by LLVM.</li>
1303 </ul>
1304
1305<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1306<tt><a href="#i_load">load</a></tt> merely indicates the size and
1307alignment of the memory from which to load, as well as the
1308interpretation of the value. The first operand of a
1309<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1310and alignment of the store.</p>
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1311
1312<p>Consequently, type-based alias analysis, aka TBAA, aka
1313<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1314LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1315additional information which specialized optimization passes may use
1316to implement type-based alias analysis.</p>
1317
1318</div>
1319
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1320<!-- *********************************************************************** -->
1321<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1322<!-- *********************************************************************** -->
1323
1324<div class="doc_text">
1325
1326<p>The LLVM type system is one of the most important features of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1327 intermediate representation. Being typed enables a number of optimizations
1328 to be performed on the intermediate representation directly, without having
1329 to do extra analyses on the side before the transformation. A strong type
1330 system makes it easier to read the generated code and enables novel analyses
1331 and transformations that are not feasible to perform on normal three address
1332 code representations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1333
1334</div>
1335
1336<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1337<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1338Classifications</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1339
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1340<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1341
1342<p>The types fall into a few useful classifications:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1343
1344<table border="1" cellspacing="0" cellpadding="4">
1345 <tbody>
1346 <tr><th>Classification</th><th>Types</th></tr>
1347 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1348 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1349 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1350 </tr>
1351 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1352 <td><a href="#t_floating">floating point</a></td>
1353 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1354 </tr>
1355 <tr>
1356 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1357 <td><a href="#t_integer">integer</a>,
1358 <a href="#t_floating">floating point</a>,
1359 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1360 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1361 <a href="#t_struct">structure</a>,
1362 <a href="#t_array">array</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1363 <a href="#t_label">label</a>,
1364 <a href="#t_metadata">metadata</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1365 </td>
1366 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1367 <tr>
1368 <td><a href="#t_primitive">primitive</a></td>
1369 <td><a href="#t_label">label</a>,
1370 <a href="#t_void">void</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1371 <a href="#t_floating">floating point</a>,
1372 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1373 </tr>
1374 <tr>
1375 <td><a href="#t_derived">derived</a></td>
1376 <td><a href="#t_integer">integer</a>,
1377 <a href="#t_array">array</a>,
1378 <a href="#t_function">function</a>,
1379 <a href="#t_pointer">pointer</a>,
1380 <a href="#t_struct">structure</a>,
1381 <a href="#t_pstruct">packed structure</a>,
1382 <a href="#t_vector">vector</a>,
1383 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1384 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1385 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1386 </tbody>
1387</table>
1388
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1389<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1390 important. Values of these types are the only ones which can be produced by
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1391 instructions.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1392
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1393</div>
1394
1395<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1396<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1397
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1398<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1399
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1400<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1401 system.</p>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1402
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1403</div>
1404
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1405<!-- _______________________________________________________________________ -->
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1406<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1407
1408<div class="doc_text">
1409
1410<h5>Overview:</h5>
1411<p>The integer type is a very simple type that simply specifies an arbitrary
1412 bit width for the integer type desired. Any bit width from 1 bit to
1413 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1414
1415<h5>Syntax:</h5>
1416<pre>
1417 iN
1418</pre>
1419
1420<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1421 value.</p>
1422
1423<h5>Examples:</h5>
1424<table class="layout">
1425 <tr class="layout">
1426 <td class="left"><tt>i1</tt></td>
1427 <td class="left">a single-bit integer.</td>
1428 </tr>
1429 <tr class="layout">
1430 <td class="left"><tt>i32</tt></td>
1431 <td class="left">a 32-bit integer.</td>
1432 </tr>
1433 <tr class="layout">
1434 <td class="left"><tt>i1942652</tt></td>
1435 <td class="left">a really big integer of over 1 million bits.</td>
1436 </tr>
1437</table>
1438
1439<p>Note that the code generator does not yet support large integer types to be
1440 used as function return types. The specific limit on how large a return type
1441 the code generator can currently handle is target-dependent; currently it's
1442 often 64 bits for 32-bit targets and 128 bits for 64-bit targets.</p>
1443
1444</div>
1445
1446<!-- _______________________________________________________________________ -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1447<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1448
1449<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1450
1451<table>
1452 <tbody>
1453 <tr><th>Type</th><th>Description</th></tr>
1454 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1455 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1456 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1457 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1458 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1459 </tbody>
1460</table>
1461
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1462</div>
1463
1464<!-- _______________________________________________________________________ -->
1465<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1466
1467<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1468
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1469<h5>Overview:</h5>
1470<p>The void type does not represent any value and has no size.</p>
1471
1472<h5>Syntax:</h5>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1473<pre>
1474 void
1475</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1476
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1477</div>
1478
1479<!-- _______________________________________________________________________ -->
1480<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1481
1482<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1483
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1484<h5>Overview:</h5>
1485<p>The label type represents code labels.</p>
1486
1487<h5>Syntax:</h5>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1488<pre>
1489 label
1490</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1491
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1492</div>
1493
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1494<!-- _______________________________________________________________________ -->
1495<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1496
1497<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1498
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1499<h5>Overview:</h5>
Nick Lewyckyeb021882009-09-27 23:27:42 +0000[diff] [blame]1500<p>The metadata type represents embedded metadata. No derived types may be
1501 created from metadata except for <a href="#t_function">function</a>
1502 arguments.
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1503
1504<h5>Syntax:</h5>
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1505<pre>
1506 metadata
1507</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1508
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1509</div>
1510
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1511
1512<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1513<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1514
1515<div class="doc_text">
1516
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1517<p>The real power in LLVM comes from the derived types in the system. This is
1518 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1519 useful types. Each of these types contain one or more element types which
1520 may be a primitive type, or another derived type. For example, it is
1521 possible to have a two dimensional array, using an array as the element type
1522 of another array.</p>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1523
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1524</div>
1525
1526<!-- _______________________________________________________________________ -->
1527<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1528
1529<div class="doc_text">
1530
1531<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1532<p>The array type is a very simple derived type that arranges elements
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1533 sequentially in memory. The array type requires a size (number of elements)
1534 and an underlying data type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1535
1536<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1537<pre>
1538 [&lt;# elements&gt; x &lt;elementtype&gt;]
1539</pre>
1540
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1541<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1542 be any type with a size.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1543
1544<h5>Examples:</h5>
1545<table class="layout">
1546 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1547 <td class="left"><tt>[40 x i32]</tt></td>
1548 <td class="left">Array of 40 32-bit integer values.</td>
1549 </tr>
1550 <tr class="layout">
1551 <td class="left"><tt>[41 x i32]</tt></td>
1552 <td class="left">Array of 41 32-bit integer values.</td>
1553 </tr>
1554 <tr class="layout">
1555 <td class="left"><tt>[4 x i8]</tt></td>
1556 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1557 </tr>
1558</table>
1559<p>Here are some examples of multidimensional arrays:</p>
1560<table class="layout">
1561 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1562 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1563 <td class="left">3x4 array of 32-bit integer values.</td>
1564 </tr>
1565 <tr class="layout">
1566 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1567 <td class="left">12x10 array of single precision floating point values.</td>
1568 </tr>
1569 <tr class="layout">
1570 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1571 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1572 </tr>
1573</table>
1574
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1575<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1576 length array. Normally, accesses past the end of an array are undefined in
1577 LLVM (e.g. it is illegal to access the 5th element of a 3 element array). As
1578 a special case, however, zero length arrays are recognized to be variable
1579 length. This allows implementation of 'pascal style arrays' with the LLVM
1580 type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1581
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1582<p>Note that the code generator does not yet support large aggregate types to be
1583 used as function return types. The specific limit on how large an aggregate
1584 return type the code generator can currently handle is target-dependent, and
1585 also dependent on the aggregate element types.</p>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1586
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1587</div>
1588
1589<!-- _______________________________________________________________________ -->
1590<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1591
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1592<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1593
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1594<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1595<p>The function type can be thought of as a function signature. It consists of
1596 a return type and a list of formal parameter types. The return type of a
1597 function type is a scalar type, a void type, or a struct type. If the return
1598 type is a struct type then all struct elements must be of first class types,
1599 and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1600
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1601<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1602<pre>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1603 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1604</pre>
1605
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1606<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]1607 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1608 which indicates that the function takes a variable number of arguments.
1609 Variable argument functions can access their arguments with
1610 the <a href="#int_varargs">variable argument handling intrinsic</a>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1611 functions. '<tt>&lt;returntype&gt;</tt>' is a any type except
Nick Lewyckyeb021882009-09-27 23:27:42 +0000[diff] [blame]1612 <a href="#t_label">label</a>.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1613
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1614<h5>Examples:</h5>
1615<table class="layout">
1616 <tr class="layout">
1617 <td class="left"><tt>i32 (i32)</tt></td>
1618 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1619 </td>
1620 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1621 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1622 </tt></td>
1623 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1624 an <tt>i16</tt> that should be sign extended and a
1625 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1626 <tt>float</tt>.
1627 </td>
1628 </tr><tr class="layout">
1629 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1630 <td class="left">A vararg function that takes at least one
1631 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1632 which returns an integer. This is the signature for <tt>printf</tt> in
1633 LLVM.
1634 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1635 </tr><tr class="layout">
1636 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1637 <td class="left">A function taking an <tt>i32</tt>, returning a
1638 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1639 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1640 </tr>
1641</table>
1642
1643</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1644
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1645<!-- _______________________________________________________________________ -->
1646<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1647
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1648<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1649
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1650<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1651<p>The structure type is used to represent a collection of data members together
1652 in memory. The packing of the field types is defined to match the ABI of the
1653 underlying processor. The elements of a structure may be any type that has a
1654 size.</p>
1655
1656<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1657 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1658 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1659
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1660<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1661<pre>
1662 { &lt;type list&gt; }
1663</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1664
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1665<h5>Examples:</h5>
1666<table class="layout">
1667 <tr class="layout">
1668 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1669 <td class="left">A triple of three <tt>i32</tt> values</td>
1670 </tr><tr class="layout">
1671 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1672 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1673 second element is a <a href="#t_pointer">pointer</a> to a
1674 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1675 an <tt>i32</tt>.</td>
1676 </tr>
1677</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1678
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1679<p>Note that the code generator does not yet support large aggregate types to be
1680 used as function return types. The specific limit on how large an aggregate
1681 return type the code generator can currently handle is target-dependent, and
1682 also dependent on the aggregate element types.</p>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1683
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1684</div>
1685
1686<!-- _______________________________________________________________________ -->
1687<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1688</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1689
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1690<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1691
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1692<h5>Overview:</h5>
1693<p>The packed structure type is used to represent a collection of data members
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1694 together in memory. There is no padding between fields. Further, the
1695 alignment of a packed structure is 1 byte. The elements of a packed
1696 structure may be any type that has a size.</p>
1697
1698<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt> and
1699 '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field with
1700 the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
1701
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1702<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1703<pre>
1704 &lt; { &lt;type list&gt; } &gt;
1705</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1706
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1707<h5>Examples:</h5>
1708<table class="layout">
1709 <tr class="layout">
1710 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1711 <td class="left">A triple of three <tt>i32</tt> values</td>
1712 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1713 <td class="left">
1714<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1715 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1716 second element is a <a href="#t_pointer">pointer</a> to a
1717 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1718 an <tt>i32</tt>.</td>
1719 </tr>
1720</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1721
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1722</div>
1723
1724<!-- _______________________________________________________________________ -->
1725<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1726
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1727<div class="doc_text">
1728
1729<h5>Overview:</h5>
1730<p>As in many languages, the pointer type represents a pointer or reference to
1731 another object, which must live in memory. Pointer types may have an optional
1732 address space attribute defining the target-specific numbered address space
1733 where the pointed-to object resides. The default address space is zero.</p>
1734
1735<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does it
1736 permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1737
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1738<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1739<pre>
1740 &lt;type&gt; *
1741</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1742
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1743<h5>Examples:</h5>
1744<table class="layout">
1745 <tr class="layout">
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]1746 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1747 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1748 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1749 </tr>
1750 <tr class="layout">
1751 <td class="left"><tt>i32 (i32 *) *</tt></td>
1752 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1753 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1754 <tt>i32</tt>.</td>
1755 </tr>
1756 <tr class="layout">
1757 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1758 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1759 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1760 </tr>
1761</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1762
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1763</div>
1764
1765<!-- _______________________________________________________________________ -->
1766<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1767
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1768<div class="doc_text">
1769
1770<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1771<p>A vector type is a simple derived type that represents a vector of elements.
1772 Vector types are used when multiple primitive data are operated in parallel
1773 using a single instruction (SIMD). A vector type requires a size (number of
1774 elements) and an underlying primitive data type. Vectors must have a power
1775 of two length (1, 2, 4, 8, 16 ...). Vector types are considered
1776 <a href="#t_firstclass">first class</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1777
1778<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1779<pre>
1780 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1781</pre>
1782
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1783<p>The number of elements is a constant integer value; elementtype may be any
1784 integer or floating point type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1785
1786<h5>Examples:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1787<table class="layout">
1788 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1789 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1790 <td class="left">Vector of 4 32-bit integer values.</td>
1791 </tr>
1792 <tr class="layout">
1793 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1794 <td class="left">Vector of 8 32-bit floating-point values.</td>
1795 </tr>
1796 <tr class="layout">
1797 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1798 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1799 </tr>
1800</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1801
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1802<p>Note that the code generator does not yet support large vector types to be
1803 used as function return types. The specific limit on how large a vector
1804 return type codegen can currently handle is target-dependent; currently it's
1805 often a few times longer than a hardware vector register.</p>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1806
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1807</div>
1808
1809<!-- _______________________________________________________________________ -->
1810<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1811<div class="doc_text">
1812
1813<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1814<p>Opaque types are used to represent unknown types in the system. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1815 corresponds (for example) to the C notion of a forward declared structure
1816 type. In LLVM, opaque types can eventually be resolved to any type (not just
1817 a structure type).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1818
1819<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1820<pre>
1821 opaque
1822</pre>
1823
1824<h5>Examples:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1825<table class="layout">
1826 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1827 <td class="left"><tt>opaque</tt></td>
1828 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1829 </tr>
1830</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1831
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1832</div>
1833
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1834<!-- ======================================================================= -->
1835<div class="doc_subsection">
1836 <a name="t_uprefs">Type Up-references</a>
1837</div>
1838
1839<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1840
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1841<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1842<p>An "up reference" allows you to refer to a lexically enclosing type without
1843 requiring it to have a name. For instance, a structure declaration may
1844 contain a pointer to any of the types it is lexically a member of. Example
1845 of up references (with their equivalent as named type declarations)
1846 include:</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1847
1848<pre>
Chris Lattner5ad632d2009-02-09 10:00:56 +0000[diff] [blame]1849 { \2 * } %x = type { %x* }
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1850 { \2 }* %y = type { %y }*
1851 \1* %z = type %z*
1852</pre>
1853
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1854<p>An up reference is needed by the asmprinter for printing out cyclic types
1855 when there is no declared name for a type in the cycle. Because the
1856 asmprinter does not want to print out an infinite type string, it needs a
1857 syntax to handle recursive types that have no names (all names are optional
1858 in llvm IR).</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1859
1860<h5>Syntax:</h5>
1861<pre>
1862 \&lt;level&gt;
1863</pre>
1864
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1865<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]1866
1867<h5>Examples:</h5>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1868<table class="layout">
1869 <tr class="layout">
1870 <td class="left"><tt>\1*</tt></td>
1871 <td class="left">Self-referential pointer.</td>
1872 </tr>
1873 <tr class="layout">
1874 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1875 <td class="left">Recursive structure where the upref refers to the out-most
1876 structure.</td>
1877 </tr>
1878</table>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1879
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1880</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1881
1882<!-- *********************************************************************** -->
1883<div class="doc_section"> <a name="constants">Constants</a> </div>
1884<!-- *********************************************************************** -->
1885
1886<div class="doc_text">
1887
1888<p>LLVM has several different basic types of constants. This section describes
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1889 them all and their syntax.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1890
1891</div>
1892
1893<!-- ======================================================================= -->
1894<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1895
1896<div class="doc_text">
1897
1898<dl>
1899 <dt><b>Boolean constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1900 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1901 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1902
1903 <dt><b>Integer constants</b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1904 <dd>Standard integers (such as '4') are constants of
1905 the <a href="#t_integer">integer</a> type. Negative numbers may be used
1906 with integer types.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1907
1908 <dt><b>Floating point constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1909 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1910 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1911 notation (see below). The assembler requires the exact decimal value of a
1912 floating-point constant. For example, the assembler accepts 1.25 but
1913 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1914 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1915
1916 <dt><b>Null pointer constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1917 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1918 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1919</dl>
1920
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1921<p>The one non-intuitive notation for constants is the hexadecimal form of
1922 floating point constants. For example, the form '<tt>double
1923 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
1924 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
1925 constants are required (and the only time that they are generated by the
1926 disassembler) is when a floating point constant must be emitted but it cannot
1927 be represented as a decimal floating point number in a reasonable number of
1928 digits. For example, NaN's, infinities, and other special values are
1929 represented in their IEEE hexadecimal format so that assembly and disassembly
1930 do not cause any bits to change in the constants.</p>
1931
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1932<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1933 represented using the 16-digit form shown above (which matches the IEEE754
1934 representation for double); float values must, however, be exactly
1935 representable as IEE754 single precision. Hexadecimal format is always used
1936 for long double, and there are three forms of long double. The 80-bit format
1937 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
1938 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1939 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
1940 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
1941 currently supported target uses this format. Long doubles will only work if
1942 they match the long double format on your target. All hexadecimal formats
1943 are big-endian (sign bit at the left).</p>
1944
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1945</div>
1946
1947<!-- ======================================================================= -->
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1948<div class="doc_subsection">
Bill Wendling1a2630a2009-07-20 02:32:41 +0000[diff] [blame]1949<a name="aggregateconstants"></a> <!-- old anchor -->
1950<a name="complexconstants">Complex Constants</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1951</div>
1952
1953<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1954
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1955<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1956 constants and smaller complex constants.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1957
1958<dl>
1959 <dt><b>Structure constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1960 <dd>Structure constants are represented with notation similar to structure
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1961 type definitions (a comma separated list of elements, surrounded by braces
1962 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1963 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
1964 Structure constants must have <a href="#t_struct">structure type</a>, and
1965 the number and types of elements must match those specified by the
1966 type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1967
1968 <dt><b>Array constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1969 <dd>Array constants are represented with notation similar to array type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1970 definitions (a comma separated list of elements, surrounded by square
1971 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
1972 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
1973 the number and types of elements must match those specified by the
1974 type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1975
1976 <dt><b>Vector constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1977 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1978 definitions (a comma separated list of elements, surrounded by
1979 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
1980 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
1981 have <a href="#t_vector">vector type</a>, and the number and types of
1982 elements must match those specified by the type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1983
1984 <dt><b>Zero initialization</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1985 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1986 value to zero of <em>any</em> type, including scalar and aggregate types.
1987 This is often used to avoid having to print large zero initializers
1988 (e.g. for large arrays) and is always exactly equivalent to using explicit
1989 zero initializers.</dd>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]1990
1991 <dt><b>Metadata node</b></dt>
Nick Lewyckyf122c7e2009-05-30 16:08:30 +0000[diff] [blame]1992 <dd>A metadata node is a structure-like constant with
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1993 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
1994 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
1995 be interpreted as part of the instruction stream, metadata is a place to
1996 attach additional information such as debug info.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1997</dl>
1998
1999</div>
2000
2001<!-- ======================================================================= -->
2002<div class="doc_subsection">
2003 <a name="globalconstants">Global Variable and Function Addresses</a>
2004</div>
2005
2006<div class="doc_text">
2007
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2008<p>The addresses of <a href="#globalvars">global variables</a>
2009 and <a href="#functionstructure">functions</a> are always implicitly valid
2010 (link-time) constants. These constants are explicitly referenced when
2011 the <a href="#identifiers">identifier for the global</a> is used and always
2012 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2013 legal LLVM file:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2014
2015<div class="doc_code">
2016<pre>
2017@X = global i32 17
2018@Y = global i32 42
2019@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2020</pre>
2021</div>
2022
2023</div>
2024
2025<!-- ======================================================================= -->
2026<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2027<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2028
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2029<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]2030 indicates that the user of the value may receive an unspecified bit-pattern.
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2031 Undefined values may be of any type (other than label or void) and be used
2032 anywhere a constant is permitted.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2033
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2034<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2035 program is well defined no matter what value is used. This gives the
2036 compiler more freedom to optimize. Here are some examples of (potentially
2037 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2038
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2039
2040<div class="doc_code">
2041<pre>
2042 %A = add %X, undef
2043 %B = sub %X, undef
2044 %C = xor %X, undef
2045Safe:
2046 %A = undef
2047 %B = undef
2048 %C = undef
2049</pre>
2050</div>
2051
2052<p>This is safe because all of the output bits are affected by the undef bits.
2053Any output bit can have a zero or one depending on the input bits.</p>
2054
2055<div class="doc_code">
2056<pre>
2057 %A = or %X, undef
2058 %B = and %X, undef
2059Safe:
2060 %A = -1
2061 %B = 0
2062Unsafe:
2063 %A = undef
2064 %B = undef
2065</pre>
2066</div>
2067
2068<p>These logical operations have bits that are not always affected by the input.
2069For example, if "%X" has a zero bit, then the output of the 'and' operation will
2070always be a zero, no matter what the corresponding bit from the undef is. As
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2071such, it is unsafe to optimize or assume that the result of the and is undef.
2072However, it is safe to assume that all bits of the undef could be 0, and
2073optimize the and to 0. Likewise, it is safe to assume that all the bits of
2074the undef operand to the or could be set, allowing the or to be folded to
2075-1.</p>
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2076
2077<div class="doc_code">
2078<pre>
2079 %A = select undef, %X, %Y
2080 %B = select undef, 42, %Y
2081 %C = select %X, %Y, undef
2082Safe:
2083 %A = %X (or %Y)
2084 %B = 42 (or %Y)
2085 %C = %Y
2086Unsafe:
2087 %A = undef
2088 %B = undef
2089 %C = undef
2090</pre>
2091</div>
2092
2093<p>This set of examples show that undefined select (and conditional branch)
2094conditions can go "either way" but they have to come from one of the two
2095operands. In the %A example, if %X and %Y were both known to have a clear low
2096bit, then %A would have to have a cleared low bit. However, in the %C example,
2097the optimizer is allowed to assume that the undef operand could be the same as
2098%Y, allowing the whole select to be eliminated.</p>
2099
2100
2101<div class="doc_code">
2102<pre>
2103 %A = xor undef, undef
2104
2105 %B = undef
2106 %C = xor %B, %B
2107
2108 %D = undef
2109 %E = icmp lt %D, 4
2110 %F = icmp gte %D, 4
2111
2112Safe:
2113 %A = undef
2114 %B = undef
2115 %C = undef
2116 %D = undef
2117 %E = undef
2118 %F = undef
2119</pre>
2120</div>
2121
2122<p>This example points out that two undef operands are not necessarily the same.
2123This can be surprising to people (and also matches C semantics) where they
2124assume that "X^X" is always zero, even if X is undef. This isn't true for a
2125number of reasons, but the short answer is that an undef "variable" can
2126arbitrarily change its value over its "live range". This is true because the
2127"variable" doesn't actually <em>have a live range</em>. Instead, the value is
2128logically read from arbitrary registers that happen to be around when needed,
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2129so the value is not necessarily consistent over time. In fact, %A and %C need
Chris Lattnerc802e482009-09-08 15:13:16 +0000[diff] [blame]2130to have the same semantics or the core LLVM "replace all uses with" concept
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2131would not hold.</p>
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2132
2133<div class="doc_code">
2134<pre>
2135 %A = fdiv undef, %X
2136 %B = fdiv %X, undef
2137Safe:
2138 %A = undef
2139b: unreachable
2140</pre>
2141</div>
2142
2143<p>These examples show the crucial difference between an <em>undefined
2144value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2145allowed to have an arbitrary bit-pattern. This means that the %A operation
2146can be constant folded to undef because the undef could be an SNaN, and fdiv is
2147not (currently) defined on SNaN's. However, in the second example, we can make
2148a more aggressive assumption: because the undef is allowed to be an arbitrary
2149value, 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]2150has <em>undefined behavior</em>, we are allowed to assume that the operation
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2151does not execute at all. This allows us to delete the divide and all code after
2152it: since the undefined operation "can't happen", the optimizer can assume that
2153it occurs in dead code.
2154</p>
2155
2156<div class="doc_code">
2157<pre>
2158a: store undef -> %X
2159b: store %X -> undef
2160Safe:
2161a: &lt;deleted&gt;
2162b: unreachable
2163</pre>
2164</div>
2165
2166<p>These examples reiterate the fdiv example: a store "of" an undefined value
2167can be assumed to not have any effect: we can assume that the value is
2168overwritten with bits that happen to match what was already there. However, a
2169store "to" an undefined location could clobber arbitrary memory, therefore, it
2170has undefined behavior.</p>
2171
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2172</div>
2173
2174<!-- ======================================================================= -->
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame^]2175<div class="doc_subsection"><a name="blockaddress">Address of Basic
2176 Block</a></div>
2177<div class="doc_text">
2178
2179<p><b><tt>blockaddress(@function, %block)</tt></b></p>
2180
2181<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
2182 basic block in the specified function, and always has an i8* type.</p>
2183
2184
2185<p>This value only has defined behavior when used as an operand to the
2186 '<a href="#i_indbr"><tt>indbr</tt></a>' instruction or for comparisons
2187 against null. Pointer equality tests between labels addresses is undefined
2188 behavior - though, again, comparison against null is ok, and no label is
2189 equal to the null pointer. Some targets may provide defined semantics when
2190 using the value as the operand to an inline assembly, but that is target
2191 specific.
2192 </p>
2193
2194</div>
2195
2196
2197<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2198<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2199</div>
2200
2201<div class="doc_text">
2202
2203<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2204 to be used as constants. Constant expressions may be of
2205 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2206 operation that does not have side effects (e.g. load and call are not
2207 supported). The following is the syntax for constant expressions:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2208
2209<dl>
2210 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2211 <dd>Truncate a constant to another type. The bit size of CST must be larger
2212 than the bit size of TYPE. Both types must be integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2213
2214 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2215 <dd>Zero extend a constant to another type. The bit size of CST must be
2216 smaller or equal to the bit size of TYPE. Both types must be
2217 integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2218
2219 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2220 <dd>Sign extend a constant to another type. The bit size of CST must be
2221 smaller or equal to the bit size of TYPE. Both types must be
2222 integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2223
2224 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2225 <dd>Truncate a floating point constant to another floating point type. The
2226 size of CST must be larger than the size of TYPE. Both types must be
2227 floating point.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2228
2229 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2230 <dd>Floating point extend a constant to another type. The size of CST must be
2231 smaller or equal to the size of TYPE. Both types must be floating
2232 point.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2233
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]2234 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2235 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2236 constant. TYPE must be a scalar or vector integer type. CST must be of
2237 scalar or vector floating point type. Both CST and TYPE must be scalars,
2238 or vectors of the same number of elements. If the value won't fit in the
2239 integer type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2240
2241 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
2242 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2243 constant. TYPE must be a scalar or vector integer type. CST must be of
2244 scalar or vector floating point type. Both CST and TYPE must be scalars,
2245 or vectors of the same number of elements. If the value won't fit in the
2246 integer type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2247
2248 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
2249 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2250 constant. TYPE must be a scalar or vector floating point type. CST must be
2251 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2252 vectors of the same number of elements. If the value won't fit in the
2253 floating point type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2254
2255 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2256 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2257 constant. TYPE must be a scalar or vector floating point type. CST must be
2258 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2259 vectors of the same number of elements. If the value won't fit in the
2260 floating point type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2261
2262 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2263 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2264 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2265 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2266 make it fit in <tt>TYPE</tt>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2267
2268 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2269 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2270 type. CST must be of integer type. The CST value is zero extended,
2271 truncated, or unchanged to make it fit in a pointer size. This one is
2272 <i>really</i> dangerous!</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2273
2274 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Chris Lattner557bc5d2009-02-28 18:27:03 +0000[diff] [blame]2275 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2276 are the same as those for the <a href="#i_bitcast">bitcast
2277 instruction</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2278
2279 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]2280 <dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2281 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2282 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2283 instruction, the index list may have zero or more indexes, which are
2284 required to make sense for the type of "CSTPTR".</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2285
2286 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2287 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2288
2289 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2290 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2291
2292 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2293 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2294
2295 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2296 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2297 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2298
2299 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2300 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2301 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2302
2303 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2304 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2305 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2306
2307 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2308 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2309 be any of the <a href="#binaryops">binary</a>
2310 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2311 on operands are the same as those for the corresponding instruction
2312 (e.g. no bitwise operations on floating point values are allowed).</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2313</dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2314
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2315</div>
2316
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2317<!-- ======================================================================= -->
2318<div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
2319</div>
2320
2321<div class="doc_text">
2322
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2323<p>Embedded metadata provides a way to attach arbitrary data to the instruction
2324 stream without affecting the behaviour of the program. There are two
2325 metadata primitives, strings and nodes. All metadata has the
2326 <tt>metadata</tt> type and is identified in syntax by a preceding exclamation
2327 point ('<tt>!</tt>').</p>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2328
2329<p>A metadata string is a string surrounded by double quotes. It can contain
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2330 any character by escaping non-printable characters with "\xx" where "xx" is
2331 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2332
2333<p>Metadata nodes are represented with notation similar to structure constants
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2334 (a comma separated list of elements, surrounded by braces and preceded by an
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2335 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2336 10}</tt>".</p>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2337
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2338<p>A metadata node will attempt to track changes to the values it holds. In the
2339 event that a value is deleted, it will be replaced with a typeless
2340 "<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
Nick Lewycky117f4382009-05-10 20:57:05 +0000[diff] [blame]2341
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2342<p>Optimizations may rely on metadata to provide additional information about
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2343 the program that isn't available in the instructions, or that isn't easily
2344 computable. Similarly, the code generator may expect a certain metadata
2345 format to be used to express debugging information.</p>
2346
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2347</div>
2348
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2349<!-- *********************************************************************** -->
2350<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2351<!-- *********************************************************************** -->
2352
2353<!-- ======================================================================= -->
2354<div class="doc_subsection">
2355<a name="inlineasm">Inline Assembler Expressions</a>
2356</div>
2357
2358<div class="doc_text">
2359
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2360<p>LLVM supports inline assembler expressions (as opposed
2361 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2362 a special value. This value represents the inline assembler as a string
2363 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2364 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2365 expression has side effects, and a flag indicating whether the function
2366 containing the asm needs to align its stack conservatively. An example
2367 inline assembler expression is:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2368
2369<div class="doc_code">
2370<pre>
2371i32 (i32) asm "bswap $0", "=r,r"
2372</pre>
2373</div>
2374
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2375<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2376 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2377 have:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2378
2379<div class="doc_code">
2380<pre>
2381%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2382</pre>
2383</div>
2384
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2385<p>Inline asms with side effects not visible in the constraint list must be
2386 marked as having side effects. This is done through the use of the
2387 '<tt>sideeffect</tt>' keyword, like so:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2388
2389<div class="doc_code">
2390<pre>
2391call void asm sideeffect "eieio", ""()
2392</pre>
2393</div>
2394
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2395<p>In some cases inline asms will contain code that will not work unless the
2396 stack is aligned in some way, such as calls or SSE instructions on x86,
2397 yet will not contain code that does that alignment within the asm.
2398 The compiler should make conservative assumptions about what the asm might
2399 contain and should generate its usual stack alignment code in the prologue
2400 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2401
2402<div class="doc_code">
2403<pre>
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2404call void asm alignstack "eieio", ""()
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2405</pre>
2406</div>
2407
2408<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2409 first.</p>
2410
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2411<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2412 documented here. Constraints on what can be done (e.g. duplication, moving,
2413 etc need to be documented). This is probably best done by reference to
2414 another document that covers inline asm from a holistic perspective.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2415
2416</div>
2417
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2418
2419<!-- *********************************************************************** -->
2420<div class="doc_section">
2421 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2422</div>
2423<!-- *********************************************************************** -->
2424
2425<p>LLVM has a number of "magic" global variables that contain data that affect
2426code generation or other IR semantics. These are documented here. All globals
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]2427of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2428section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2429by LLVM.</p>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2430
2431<!-- ======================================================================= -->
2432<div class="doc_subsection">
2433<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2434</div>
2435
2436<div class="doc_text">
2437
2438<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2439href="#linkage_appending">appending linkage</a>. This array contains a list of
2440pointers to global variables and functions which may optionally have a pointer
2441cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2442
2443<pre>
2444 @X = global i8 4
2445 @Y = global i32 123
2446
2447 @llvm.used = appending global [2 x i8*] [
2448 i8* @X,
2449 i8* bitcast (i32* @Y to i8*)
2450 ], section "llvm.metadata"
2451</pre>
2452
2453<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2454compiler, assembler, and linker are required to treat the symbol as if there is
2455a reference to the global that it cannot see. For example, if a variable has
2456internal linkage and no references other than that from the <tt>@llvm.used</tt>
2457list, it cannot be deleted. This is commonly used to represent references from
2458inline asms and other things the compiler cannot "see", and corresponds to
2459"attribute((used))" in GNU C.</p>
2460
2461<p>On some targets, the code generator must emit a directive to the assembler or
2462object file to prevent the assembler and linker from molesting the symbol.</p>
2463
2464</div>
2465
2466<!-- ======================================================================= -->
2467<div class="doc_subsection">
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]2468<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2469</div>
2470
2471<div class="doc_text">
2472
2473<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2474<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2475touching the symbol. On targets that support it, this allows an intelligent
2476linker to optimize references to the symbol without being impeded as it would be
2477by <tt>@llvm.used</tt>.</p>
2478
2479<p>This is a rare construct that should only be used in rare circumstances, and
2480should not be exposed to source languages.</p>
2481
2482</div>
2483
2484<!-- ======================================================================= -->
2485<div class="doc_subsection">
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2486<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2487</div>
2488
2489<div class="doc_text">
2490
2491<p>TODO: Describe this.</p>
2492
2493</div>
2494
2495<!-- ======================================================================= -->
2496<div class="doc_subsection">
2497<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2498</div>
2499
2500<div class="doc_text">
2501
2502<p>TODO: Describe this.</p>
2503
2504</div>
2505
2506
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2507<!-- *********************************************************************** -->
2508<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2509<!-- *********************************************************************** -->
2510
2511<div class="doc_text">
2512
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2513<p>The LLVM instruction set consists of several different classifications of
2514 instructions: <a href="#terminators">terminator
2515 instructions</a>, <a href="#binaryops">binary instructions</a>,
2516 <a href="#bitwiseops">bitwise binary instructions</a>,
2517 <a href="#memoryops">memory instructions</a>, and
2518 <a href="#otherops">other instructions</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2519
2520</div>
2521
2522<!-- ======================================================================= -->
2523<div class="doc_subsection"> <a name="terminators">Terminator
2524Instructions</a> </div>
2525
2526<div class="doc_text">
2527
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2528<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2529 in a program ends with a "Terminator" instruction, which indicates which
2530 block should be executed after the current block is finished. These
2531 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2532 control flow, not values (the one exception being the
2533 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2534
2535<p>There are six different terminator instructions: the
2536 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2537 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2538 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2539 '<a href="#i_indbr">'<tt>indbr</tt>' Instruction, the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2540 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2541 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2542 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2543
2544</div>
2545
2546<!-- _______________________________________________________________________ -->
2547<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2548Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2549
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2550<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2551
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2552<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2553<pre>
2554 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]2555 ret void <i>; Return from void function</i>
2556</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2557
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2558<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2559<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2560 a value) from a function back to the caller.</p>
2561
2562<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2563 value and then causes control flow, and one that just causes control flow to
2564 occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2565
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2566<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2567<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2568 return value. The type of the return value must be a
2569 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2570
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2571<p>A function is not <a href="#wellformed">well formed</a> if it it has a
2572 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2573 value or a return value with a type that does not match its type, or if it
2574 has a void return type and contains a '<tt>ret</tt>' instruction with a
2575 return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2576
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2577<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2578<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2579 the calling function's context. If the caller is a
2580 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2581 instruction after the call. If the caller was an
2582 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2583 the beginning of the "normal" destination block. If the instruction returns
2584 a value, that value shall set the call or invoke instruction's return
2585 value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2586
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2587<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2588<pre>
2589 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2590 ret void <i>; Return from a void function</i>
Bill Wendlingd163e2d2009-02-28 22:12:54 +0000[diff] [blame]2591 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]2592</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2593
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]2594<p>Note that the code generator does not yet fully support large
2595 return values. The specific sizes that are currently supported are
2596 dependent on the target. For integers, on 32-bit targets the limit
2597 is often 64 bits, and on 64-bit targets the limit is often 128 bits.
2598 For aggregate types, the current limits are dependent on the element
2599 types; for example targets are often limited to 2 total integer
2600 elements and 2 total floating-point elements.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2601
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2602</div>
2603<!-- _______________________________________________________________________ -->
2604<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2605
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2606<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2607
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2608<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2609<pre>
2610 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]2611</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2612
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2613<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2614<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2615 different basic block in the current function. There are two forms of this
2616 instruction, corresponding to a conditional branch and an unconditional
2617 branch.</p>
2618
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2619<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2620<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2621 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2622 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2623 target.</p>
2624
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2625<h5>Semantics:</h5>
2626<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2627 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2628 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2629 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2630
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2631<h5>Example:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]2632<pre>
2633Test:
2634 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2635 br i1 %cond, label %IfEqual, label %IfUnequal
2636IfEqual:
2637 <a href="#i_ret">ret</a> i32 1
2638IfUnequal:
2639 <a href="#i_ret">ret</a> i32 0
2640</pre>
2641
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2642</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2643
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2644<!-- _______________________________________________________________________ -->
2645<div class="doc_subsubsection">
2646 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2647</div>
2648
2649<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2650
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2651<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2652<pre>
2653 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2654</pre>
2655
2656<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2657<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]2658 several different places. It is a generalization of the '<tt>br</tt>'
2659 instruction, allowing a branch to occur to one of many possible
2660 destinations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2661
2662<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2663<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2664 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2665 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2666 The table is not allowed to contain duplicate constant entries.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2667
2668<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2669<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2670 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2671 is searched for the given value. If the value is found, control flow is
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2672 transferred to the corresponding destination; otherwise, control flow is
2673 transferred to the default destination.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2674
2675<h5>Implementation:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2676<p>Depending on properties of the target machine and the particular
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2677 <tt>switch</tt> instruction, this instruction may be code generated in
2678 different ways. For example, it could be generated as a series of chained
2679 conditional branches or with a lookup table.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2680
2681<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2682<pre>
2683 <i>; Emulate a conditional br instruction</i>
2684 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2685 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2686
2687 <i>; Emulate an unconditional br instruction</i>
2688 switch i32 0, label %dest [ ]
2689
2690 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2691 switch i32 %val, label %otherwise [ i32 0, label %onzero
2692 i32 1, label %onone
2693 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2694</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2695
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2696</div>
2697
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2698
2699<!-- _______________________________________________________________________ -->
2700<div class="doc_subsubsection">
2701 <a name="i_indbr">'<tt>indbr</tt>' Instruction</a>
2702</div>
2703
2704<div class="doc_text">
2705
2706<h5>Syntax:</h5>
2707<pre>
2708 indbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
2709</pre>
2710
2711<h5>Overview:</h5>
2712
2713<p>The '<tt>indbr</tt>' instruction implements an indirect branch to a label
2714 within the current function, whose address is specified by
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame^]2715 "<tt>address</tt>". Address must be derived from a <a
2716 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2717
2718<h5>Arguments:</h5>
2719
2720<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
2721 rest of the arguments indicate the full set of possible destinations that the
2722 address may point to. Blocks are allowed to occur multiple times in the
2723 destination list, though this isn't particularly useful.</p>
2724
2725<p>This destination list is required so that dataflow analysis has an accurate
2726 understanding of the CFG.</p>
2727
2728<h5>Semantics:</h5>
2729
2730<p>Control transfers to the block specified in the address argument. All
2731 possible destination blocks must be listed in the label list, otherwise this
2732 instruction has undefined behavior. This implies that jumps to labels
2733 defined in other functions have undefined behavior as well.</p>
2734
2735<h5>Implementation:</h5>
2736
2737<p>This is typically implemented with a jump through a register.</p>
2738
2739<h5>Example:</h5>
2740<pre>
Chris Lattner1e6c6eb2009-10-27 20:27:24 +0000[diff] [blame]2741 indbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2742</pre>
2743
2744</div>
2745
2746
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2747<!-- _______________________________________________________________________ -->
2748<div class="doc_subsubsection">
2749 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2750</div>
2751
2752<div class="doc_text">
2753
2754<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2755<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2756 &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]2757 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2758</pre>
2759
2760<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2761<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2762 function, with the possibility of control flow transfer to either the
2763 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
2764 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
2765 control flow will return to the "normal" label. If the callee (or any
2766 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
2767 instruction, control is interrupted and continued at the dynamically nearest
2768 "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2769
2770<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2771<p>This instruction requires several arguments:</p>
2772
2773<ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2774 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
2775 convention</a> the call should use. If none is specified, the call
2776 defaults to using C calling conventions.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2777
2778 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2779 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
2780 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2781
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2782 <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]2783 function value being invoked. In most cases, this is a direct function
2784 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
2785 off an arbitrary pointer to function value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2786
2787 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2788 function to be invoked. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2789
2790 <li>'<tt>function args</tt>': argument list whose types match the function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2791 signature argument types. If the function signature indicates the
2792 function accepts a variable number of arguments, the extra arguments can
2793 be specified.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2794
2795 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2796 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2797
2798 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2799 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2800
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2801 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2802 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2803 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2804</ol>
2805
2806<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2807<p>This instruction is designed to operate as a standard
2808 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
2809 primary difference is that it establishes an association with a label, which
2810 is used by the runtime library to unwind the stack.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2811
2812<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2813 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2814 exception. Additionally, this is important for implementation of
2815 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2816
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2817<p>For the purposes of the SSA form, the definition of the value returned by the
2818 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
2819 block to the "normal" label. If the callee unwinds then no return value is
2820 available.</p>
Dan Gohman140ba5d2009-05-22 21:47:08 +0000[diff] [blame]2821
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2822<h5>Example:</h5>
2823<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2824 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2825 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2826 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2827 unwind label %TestCleanup <i>; {i32}:retval set</i>
2828</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2829
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2830</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2831
2832<!-- _______________________________________________________________________ -->
2833
2834<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2835Instruction</a> </div>
2836
2837<div class="doc_text">
2838
2839<h5>Syntax:</h5>
2840<pre>
2841 unwind
2842</pre>
2843
2844<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2845<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2846 at the first callee in the dynamic call stack which used
2847 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
2848 This is primarily used to implement exception handling.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2849
2850<h5>Semantics:</h5>
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2851<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2852 immediately halt. The dynamic call stack is then searched for the
2853 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
2854 Once found, execution continues at the "exceptional" destination block
2855 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
2856 instruction in the dynamic call chain, undefined behavior results.</p>
2857
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2858</div>
2859
2860<!-- _______________________________________________________________________ -->
2861
2862<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2863Instruction</a> </div>
2864
2865<div class="doc_text">
2866
2867<h5>Syntax:</h5>
2868<pre>
2869 unreachable
2870</pre>
2871
2872<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2873<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2874 instruction is used to inform the optimizer that a particular portion of the
2875 code is not reachable. This can be used to indicate that the code after a
2876 no-return function cannot be reached, and other facts.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2877
2878<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2879<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2880
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2881</div>
2882
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2883<!-- ======================================================================= -->
2884<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2885
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2886<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2887
2888<p>Binary operators are used to do most of the computation in a program. They
2889 require two operands of the same type, execute an operation on them, and
2890 produce a single value. The operands might represent multiple data, as is
2891 the case with the <a href="#t_vector">vector</a> data type. The result value
2892 has the same type as its operands.</p>
2893
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2894<p>There are several different binary operators:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2895
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2896</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2897
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2898<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2899<div class="doc_subsubsection">
2900 <a name="i_add">'<tt>add</tt>' Instruction</a>
2901</div>
2902
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2903<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2904
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2905<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2906<pre>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]2907 &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]2908 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2909 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2910 &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]2911</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2912
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2913<h5>Overview:</h5>
2914<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2915
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2916<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2917<p>The two arguments to the '<tt>add</tt>' instruction must
2918 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2919 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2920
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2921<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2922<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2923
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2924<p>If the sum has unsigned overflow, the result returned is the mathematical
2925 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]2926
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2927<p>Because LLVM integers use a two's complement representation, this instruction
2928 is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2929
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]2930<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
2931 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
2932 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
2933 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]2934
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2935<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2936<pre>
2937 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2938</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2939
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2940</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2941
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2942<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2943<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2944 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2945</div>
2946
2947<div class="doc_text">
2948
2949<h5>Syntax:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2950<pre>
2951 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2952</pre>
2953
2954<h5>Overview:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2955<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2956
2957<h5>Arguments:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2958<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2959 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2960 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2961
2962<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2963<p>The value produced is the floating point sum of the two operands.</p>
2964
2965<h5>Example:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2966<pre>
2967 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2968</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2969
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2970</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2971
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2972<!-- _______________________________________________________________________ -->
2973<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2974 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2975</div>
2976
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2977<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2978
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2979<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2980<pre>
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]2981 &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]2982 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2983 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2984 &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]2985</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2986
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2987<h5>Overview:</h5>
2988<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2989 operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2990
2991<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2992 '<tt>neg</tt>' instruction present in most other intermediate
2993 representations.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2994
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2995<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2996<p>The two arguments to the '<tt>sub</tt>' instruction must
2997 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2998 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2999
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3000<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3001<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3002
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3003<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3004 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
3005 result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3006
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3007<p>Because LLVM integers use a two's complement representation, this instruction
3008 is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3009
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3010<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3011 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3012 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3013 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3014
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3015<h5>Example:</h5>
3016<pre>
3017 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3018 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3019</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3020
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3021</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3022
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3023<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3024<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3025 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3026</div>
3027
3028<div class="doc_text">
3029
3030<h5>Syntax:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3031<pre>
3032 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3033</pre>
3034
3035<h5>Overview:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3036<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3037 operands.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3038
3039<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3040 '<tt>fneg</tt>' instruction present in most other intermediate
3041 representations.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3042
3043<h5>Arguments:</h5>
Bill Wendling1a2630a2009-07-20 02:32:41 +0000[diff] [blame]3044<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3045 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3046 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3047
3048<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3049<p>The value produced is the floating point difference of the two operands.</p>
3050
3051<h5>Example:</h5>
3052<pre>
3053 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3054 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3055</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3056
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3057</div>
3058
3059<!-- _______________________________________________________________________ -->
3060<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3061 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3062</div>
3063
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3064<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3065
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3066<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3067<pre>
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3068 &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]3069 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3070 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3071 &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]3072</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3073
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3074<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3075<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3076
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3077<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3078<p>The two arguments to the '<tt>mul</tt>' instruction must
3079 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3080 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3081
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3082<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3083<p>The value produced is the integer product of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3084
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3085<p>If the result of the multiplication has unsigned overflow, the result
3086 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3087 width of the result.</p>
3088
3089<p>Because LLVM integers use a two's complement representation, and the result
3090 is the same width as the operands, this instruction returns the correct
3091 result for both signed and unsigned integers. If a full product
3092 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3093 be sign-extended or zero-extended as appropriate to the width of the full
3094 product.</p>
3095
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3096<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3097 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3098 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3099 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3100
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3101<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3102<pre>
3103 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3104</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3105
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3106</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3107
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3108<!-- _______________________________________________________________________ -->
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3109<div class="doc_subsubsection">
3110 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3111</div>
3112
3113<div class="doc_text">
3114
3115<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3116<pre>
3117 &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]3118</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3119
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3120<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3121<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3122
3123<h5>Arguments:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3124<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3125 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3126 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3127
3128<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3129<p>The value produced is the floating point product of the two operands.</p>
3130
3131<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3132<pre>
3133 &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]3134</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3135
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3136</div>
3137
3138<!-- _______________________________________________________________________ -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3139<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3140</a></div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3141
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3142<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3143
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3144<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3145<pre>
3146 &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]3147</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3148
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3149<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3150<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3151
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3152<h5>Arguments:</h5>
3153<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3154 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3155 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3156
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3157<h5>Semantics:</h5>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3158<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3159
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3160<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3161 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3162
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3163<p>Division by zero leads to undefined behavior.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3164
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3165<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3166<pre>
3167 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3168</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3169
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3170</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3171
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3172<!-- _______________________________________________________________________ -->
3173<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3174</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3175
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3176<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3177
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3178<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3179<pre>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3180 &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]3181 &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]3182</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3183
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3184<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3185<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3186
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3187<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3188<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3189 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3190 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3191
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3192<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3193<p>The value produced is the signed integer quotient of the two operands rounded
3194 towards zero.</p>
3195
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3196<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3197 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3198
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3199<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3200 undefined behavior; this is a rare case, but can occur, for example, by doing
3201 a 32-bit division of -2147483648 by -1.</p>
3202
Dan Gohman67fa48e2009-07-22 00:04:19 +0000[diff] [blame]3203<p>If the <tt>exact</tt> keyword is present, the result value of the
3204 <tt>sdiv</tt> is undefined if the result would be rounded or if overflow
3205 would occur.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3206
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3207<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3208<pre>
3209 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3210</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3211
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3212</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3213
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3214<!-- _______________________________________________________________________ -->
3215<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3216Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3217
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3218<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3219
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3220<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3221<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3222 &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]3223</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3224
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3225<h5>Overview:</h5>
3226<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3227
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3228<h5>Arguments:</h5>
3229<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3230 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3231 floating point values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3232
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3233<h5>Semantics:</h5>
3234<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3235
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3236<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3237<pre>
3238 &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]3239</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3240
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3241</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3242
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3243<!-- _______________________________________________________________________ -->
3244<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3245</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3246
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3247<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3248
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3249<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3250<pre>
3251 &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]3252</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3253
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3254<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3255<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3256 division of its two arguments.</p>
3257
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3258<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3259<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3260 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3261 values. Both arguments must have identical types.</p>
3262
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3263<h5>Semantics:</h5>
3264<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3265 This instruction always performs an unsigned division to get the
3266 remainder.</p>
3267
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3268<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3269 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3270
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3271<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3272
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3273<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3274<pre>
3275 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3276</pre>
3277
3278</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3279
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3280<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3281<div class="doc_subsubsection">
3282 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3283</div>
3284
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3285<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3286
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3287<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3288<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3289 &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]3290</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3291
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3292<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3293<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3294 division of its two operands. This instruction can also take
3295 <a href="#t_vector">vector</a> versions of the values in which case the
3296 elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]3297
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3298<h5>Arguments:</h5>
3299<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3300 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3301 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3302
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3303<h5>Semantics:</h5>
3304<p>This instruction returns the <i>remainder</i> of a division (where the result
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3305 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3306 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3307 a value. For more information about the difference,
3308 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3309 Math Forum</a>. For a table of how this is implemented in various languages,
3310 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3311 Wikipedia: modulo operation</a>.</p>
3312
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3313<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3314 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3315
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3316<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3317 Overflow also leads to undefined behavior; this is a rare case, but can
3318 occur, for example, by taking the remainder of a 32-bit division of
3319 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3320 lets srem be implemented using instructions that return both the result of
3321 the division and the remainder.)</p>
3322
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3323<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3324<pre>
3325 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3326</pre>
3327
3328</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3329
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3330<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3331<div class="doc_subsubsection">
3332 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3333
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3334<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3335
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3336<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3337<pre>
3338 &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]3339</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3340
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3341<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3342<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3343 its two operands.</p>
3344
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3345<h5>Arguments:</h5>
3346<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3347 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3348 floating point values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3349
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3350<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3351<p>This instruction returns the <i>remainder</i> of a division. The remainder
3352 has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3353
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3354<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3355<pre>
3356 &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]3357</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3358
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3359</div>
3360
3361<!-- ======================================================================= -->
3362<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3363Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3364
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3365<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3366
3367<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3368 program. They are generally very efficient instructions and can commonly be
3369 strength reduced from other instructions. They require two operands of the
3370 same type, execute an operation on them, and produce a single value. The
3371 resulting value is the same type as its operands.</p>
3372
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3373</div>
3374
3375<!-- _______________________________________________________________________ -->
3376<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3377Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3378
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3379<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3380
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3381<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3382<pre>
3383 &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]3384</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3385
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3386<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3387<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3388 a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3389
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3390<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3391<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3392 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3393 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3394
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3395<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3396<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3397 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3398 is (statically or dynamically) negative or equal to or larger than the number
3399 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3400 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3401 shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3402
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3403<h5>Example:</h5>
3404<pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3405 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3406 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3407 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3408 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3409 &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]3410</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3411
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3412</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3413
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3414<!-- _______________________________________________________________________ -->
3415<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3416Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3417
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3418<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3419
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3420<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3421<pre>
3422 &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]3423</pre>
3424
3425<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3426<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3427 operand shifted to the right a specified number of bits with zero fill.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3428
3429<h5>Arguments:</h5>
3430<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3431 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3432 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3433
3434<h5>Semantics:</h5>
3435<p>This instruction always performs a logical shift right operation. The most
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3436 significant bits of the result will be filled with zero bits after the shift.
3437 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3438 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3439 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3440 shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3441
3442<h5>Example:</h5>
3443<pre>
3444 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3445 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3446 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3447 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3448 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3449 &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]3450</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3451
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3452</div>
3453
3454<!-- _______________________________________________________________________ -->
3455<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3456Instruction</a> </div>
3457<div class="doc_text">
3458
3459<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3460<pre>
3461 &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]3462</pre>
3463
3464<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3465<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3466 operand shifted to the right a specified number of bits with sign
3467 extension.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3468
3469<h5>Arguments:</h5>
3470<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3471 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3472 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3473
3474<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3475<p>This instruction always performs an arithmetic shift right operation, The
3476 most significant bits of the result will be filled with the sign bit
3477 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3478 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3479 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3480 the corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3481
3482<h5>Example:</h5>
3483<pre>
3484 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3485 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3486 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3487 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3488 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3489 &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]3490</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3491
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3492</div>
3493
3494<!-- _______________________________________________________________________ -->
3495<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3496Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3497
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3498<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3499
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3500<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3501<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3502 &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]3503</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3504
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3505<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3506<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3507 operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3508
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3509<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3510<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3511 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3512 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3513
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3514<h5>Semantics:</h5>
3515<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3516
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3517<table border="1" cellspacing="0" cellpadding="4">
3518 <tbody>
3519 <tr>
3520 <td>In0</td>
3521 <td>In1</td>
3522 <td>Out</td>
3523 </tr>
3524 <tr>
3525 <td>0</td>
3526 <td>0</td>
3527 <td>0</td>
3528 </tr>
3529 <tr>
3530 <td>0</td>
3531 <td>1</td>
3532 <td>0</td>
3533 </tr>
3534 <tr>
3535 <td>1</td>
3536 <td>0</td>
3537 <td>0</td>
3538 </tr>
3539 <tr>
3540 <td>1</td>
3541 <td>1</td>
3542 <td>1</td>
3543 </tr>
3544 </tbody>
3545</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3546
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3547<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3548<pre>
3549 &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]3550 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3551 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3552</pre>
3553</div>
3554<!-- _______________________________________________________________________ -->
3555<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3556
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3557<div class="doc_text">
3558
3559<h5>Syntax:</h5>
3560<pre>
3561 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3562</pre>
3563
3564<h5>Overview:</h5>
3565<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3566 two operands.</p>
3567
3568<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3569<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3570 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3571 values. Both arguments must have identical types.</p>
3572
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3573<h5>Semantics:</h5>
3574<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3575
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3576<table border="1" cellspacing="0" cellpadding="4">
3577 <tbody>
3578 <tr>
3579 <td>In0</td>
3580 <td>In1</td>
3581 <td>Out</td>
3582 </tr>
3583 <tr>
3584 <td>0</td>
3585 <td>0</td>
3586 <td>0</td>
3587 </tr>
3588 <tr>
3589 <td>0</td>
3590 <td>1</td>
3591 <td>1</td>
3592 </tr>
3593 <tr>
3594 <td>1</td>
3595 <td>0</td>
3596 <td>1</td>
3597 </tr>
3598 <tr>
3599 <td>1</td>
3600 <td>1</td>
3601 <td>1</td>
3602 </tr>
3603 </tbody>
3604</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3605
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3606<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3607<pre>
3608 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3609 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3610 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3611</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3612
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3613</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3614
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3615<!-- _______________________________________________________________________ -->
3616<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3617Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3618
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3619<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3620
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3621<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3622<pre>
3623 &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]3624</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3625
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3626<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3627<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3628 its two operands. The <tt>xor</tt> is used to implement the "one's
3629 complement" operation, which is the "~" operator in C.</p>
3630
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3631<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3632<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3633 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3634 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3635
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3636<h5>Semantics:</h5>
3637<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3638
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3639<table border="1" cellspacing="0" cellpadding="4">
3640 <tbody>
3641 <tr>
3642 <td>In0</td>
3643 <td>In1</td>
3644 <td>Out</td>
3645 </tr>
3646 <tr>
3647 <td>0</td>
3648 <td>0</td>
3649 <td>0</td>
3650 </tr>
3651 <tr>
3652 <td>0</td>
3653 <td>1</td>
3654 <td>1</td>
3655 </tr>
3656 <tr>
3657 <td>1</td>
3658 <td>0</td>
3659 <td>1</td>
3660 </tr>
3661 <tr>
3662 <td>1</td>
3663 <td>1</td>
3664 <td>0</td>
3665 </tr>
3666 </tbody>
3667</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3668
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3669<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3670<pre>
3671 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3672 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3673 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3674 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3675</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3676
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3677</div>
3678
3679<!-- ======================================================================= -->
3680<div class="doc_subsection">
3681 <a name="vectorops">Vector Operations</a>
3682</div>
3683
3684<div class="doc_text">
3685
3686<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3687 target-independent manner. These instructions cover the element-access and
3688 vector-specific operations needed to process vectors effectively. While LLVM
3689 does directly support these vector operations, many sophisticated algorithms
3690 will want to use target-specific intrinsics to take full advantage of a
3691 specific target.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3692
3693</div>
3694
3695<!-- _______________________________________________________________________ -->
3696<div class="doc_subsubsection">
3697 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3698</div>
3699
3700<div class="doc_text">
3701
3702<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3703<pre>
3704 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
3705</pre>
3706
3707<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3708<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3709 from a vector at a specified index.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3710
3711
3712<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3713<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3714 of <a href="#t_vector">vector</a> type. The second operand is an index
3715 indicating the position from which to extract the element. The index may be
3716 a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3717
3718<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3719<p>The result is a scalar of the same type as the element type of
3720 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3721 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3722 results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3723
3724<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3725<pre>
3726 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
3727</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3728
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3729</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3730
3731<!-- _______________________________________________________________________ -->
3732<div class="doc_subsubsection">
3733 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3734</div>
3735
3736<div class="doc_text">
3737
3738<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3739<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]3740 &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]3741</pre>
3742
3743<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3744<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
3745 vector at a specified index.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3746
3747<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3748<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
3749 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
3750 whose type must equal the element type of the first operand. The third
3751 operand is an index indicating the position at which to insert the value.
3752 The index may be a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3753
3754<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3755<p>The result is a vector of the same type as <tt>val</tt>. Its element values
3756 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
3757 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3758 results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3759
3760<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3761<pre>
3762 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
3763</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3764
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3765</div>
3766
3767<!-- _______________________________________________________________________ -->
3768<div class="doc_subsubsection">
3769 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3770</div>
3771
3772<div class="doc_text">
3773
3774<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3775<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3776 &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]3777</pre>
3778
3779<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3780<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3781 from two input vectors, returning a vector with the same element type as the
3782 input and length that is the same as the shuffle mask.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3783
3784<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3785<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3786 with types that match each other. The third argument is a shuffle mask whose
3787 element type is always 'i32'. The result of the instruction is a vector
3788 whose length is the same as the shuffle mask and whose element type is the
3789 same as the element type of the first two operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3790
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3791<p>The shuffle mask operand is required to be a constant vector with either
3792 constant integer or undef values.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3793
3794<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3795<p>The elements of the two input vectors are numbered from left to right across
3796 both of the vectors. The shuffle mask operand specifies, for each element of
3797 the result vector, which element of the two input vectors the result element
3798 gets. The element selector may be undef (meaning "don't care") and the
3799 second operand may be undef if performing a shuffle from only one vector.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3800
3801<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3802<pre>
3803 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3804 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
3805 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
3806 &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.
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3807 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
3808 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
3809 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3810 &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]3811</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3812
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3813</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3814
3815<!-- ======================================================================= -->
3816<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3817 <a name="aggregateops">Aggregate Operations</a>
3818</div>
3819
3820<div class="doc_text">
3821
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3822<p>LLVM supports several instructions for working with aggregate values.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3823
3824</div>
3825
3826<!-- _______________________________________________________________________ -->
3827<div class="doc_subsubsection">
3828 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3829</div>
3830
3831<div class="doc_text">
3832
3833<h5>Syntax:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3834<pre>
3835 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3836</pre>
3837
3838<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3839<p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3840 or array element from an aggregate value.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3841
3842<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3843<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
3844 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3845 operands are constant indices to specify which value to extract in a similar
3846 manner as indices in a
3847 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3848
3849<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3850<p>The result is the value at the position in the aggregate specified by the
3851 index operands.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3852
3853<h5>Example:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3854<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3855 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3856</pre>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3857
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3858</div>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3859
3860<!-- _______________________________________________________________________ -->
3861<div class="doc_subsubsection">
3862 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3863</div>
3864
3865<div class="doc_text">
3866
3867<h5>Syntax:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3868<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3869 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;val&gt;, &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3870</pre>
3871
3872<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3873<p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
3874 array element in an aggregate.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3875
3876
3877<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3878<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
3879 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3880 second operand is a first-class value to insert. The following operands are
3881 constant indices indicating the position at which to insert the value in a
3882 similar manner as indices in a
3883 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
3884 value to insert must have the same type as the value identified by the
3885 indices.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3886
3887<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3888<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
3889 that of <tt>val</tt> except that the value at the position specified by the
3890 indices is that of <tt>elt</tt>.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3891
3892<h5>Example:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3893<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3894 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3895</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3896
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3897</div>
3898
3899
3900<!-- ======================================================================= -->
3901<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3902 <a name="memoryops">Memory Access and Addressing Operations</a>
3903</div>
3904
3905<div class="doc_text">
3906
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3907<p>A key design point of an SSA-based representation is how it represents
3908 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]3909 very simple. This section describes how to read, write, and allocate
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3910 memory in LLVM.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3911
3912</div>
3913
3914<!-- _______________________________________________________________________ -->
3915<div class="doc_subsubsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3916 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3917</div>
3918
3919<div class="doc_text">
3920
3921<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3922<pre>
3923 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3924</pre>
3925
3926<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3927<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3928 currently executing function, to be automatically released when this function
3929 returns to its caller. The object is always allocated in the generic address
3930 space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3931
3932<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3933<p>The '<tt>alloca</tt>' instruction
3934 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
3935 runtime stack, returning a pointer of the appropriate type to the program.
3936 If "NumElements" is specified, it is the number of elements allocated,
3937 otherwise "NumElements" is defaulted to be one. If a constant alignment is
3938 specified, the value result of the allocation is guaranteed to be aligned to
3939 at least that boundary. If not specified, or if zero, the target can choose
3940 to align the allocation on any convenient boundary compatible with the
3941 type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3942
3943<p>'<tt>type</tt>' may be any sized type.</p>
3944
3945<h5>Semantics:</h5>
Bill Wendling2a454572009-05-08 20:49:29 +0000[diff] [blame]3946<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3947 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3948 memory is automatically released when the function returns. The
3949 '<tt>alloca</tt>' instruction is commonly used to represent automatic
3950 variables that must have an address available. When the function returns
3951 (either with the <tt><a href="#i_ret">ret</a></tt>
3952 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
3953 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3954
3955<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3956<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3957 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3958 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3959 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3960 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3961</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3962
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3963</div>
3964
3965<!-- _______________________________________________________________________ -->
3966<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3967Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3968
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3969<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3970
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3971<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3972<pre>
3973 &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]
3974 &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]
3975</pre>
3976
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3977<h5>Overview:</h5>
3978<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3979
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3980<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3981<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
3982 from which to load. The pointer must point to
3983 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3984 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
3985 number or order of execution of this <tt>load</tt> with other
3986 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3987 instructions. </p>
3988
3989<p>The optional constant "align" argument specifies the alignment of the
3990 operation (that is, the alignment of the memory address). A value of 0 or an
3991 omitted "align" argument means that the operation has the preferential
3992 alignment for the target. It is the responsibility of the code emitter to
3993 ensure that the alignment information is correct. Overestimating the
3994 alignment results in an undefined behavior. Underestimating the alignment may
3995 produce less efficient code. An alignment of 1 is always safe.</p>
3996
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3997<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3998<p>The location of memory pointed to is loaded. If the value being loaded is of
3999 scalar type then the number of bytes read does not exceed the minimum number
4000 of bytes needed to hold all bits of the type. For example, loading an
4001 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
4002 <tt>i20</tt> with a size that is not an integral number of bytes, the result
4003 is undefined if the value was not originally written using a store of the
4004 same type.</p>
4005
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4006<h5>Examples:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4007<pre>
4008 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4009 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4010 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4011</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4012
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4013</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4014
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4015<!-- _______________________________________________________________________ -->
4016<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4017Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4018
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4019<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4020
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4021<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4022<pre>
4023 store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4024 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
4025</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4026
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4027<h5>Overview:</h5>
4028<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4029
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4030<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4031<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4032 and an address at which to store it. The type of the
4033 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4034 the <a href="#t_firstclass">first class</a> type of the
4035 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked
4036 as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
4037 or order of execution of this <tt>store</tt> with other
4038 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4039 instructions.</p>
4040
4041<p>The optional constant "align" argument specifies the alignment of the
4042 operation (that is, the alignment of the memory address). A value of 0 or an
4043 omitted "align" argument means that the operation has the preferential
4044 alignment for the target. It is the responsibility of the code emitter to
4045 ensure that the alignment information is correct. Overestimating the
4046 alignment results in an undefined behavior. Underestimating the alignment may
4047 produce less efficient code. An alignment of 1 is always safe.</p>
4048
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4049<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4050<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4051 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4052 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4053 does not exceed the minimum number of bytes needed to hold all bits of the
4054 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4055 writing a value of a type like <tt>i20</tt> with a size that is not an
4056 integral number of bytes, it is unspecified what happens to the extra bits
4057 that do not belong to the type, but they will typically be overwritten.</p>
4058
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4059<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4060<pre>
4061 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]4062 store i32 3, i32* %ptr <i>; yields {void}</i>
4063 %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]4064</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4065
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4066</div>
4067
4068<!-- _______________________________________________________________________ -->
4069<div class="doc_subsubsection">
4070 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4071</div>
4072
4073<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4074
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4075<h5>Syntax:</h5>
4076<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4077 &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]4078 &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]4079</pre>
4080
4081<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4082<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4083 subelement of an aggregate data structure. It performs address calculation
4084 only and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4085
4086<h5>Arguments:</h5>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4087<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4088 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4089 elements of the aggregate object are indexed. The interpretation of each
4090 index is dependent on the type being indexed into. The first index always
4091 indexes the pointer value given as the first argument, the second index
4092 indexes a value of the type pointed to (not necessarily the value directly
4093 pointed to, since the first index can be non-zero), etc. The first type
4094 indexed into must be a pointer value, subsequent types can be arrays, vectors
4095 and structs. Note that subsequent types being indexed into can never be
4096 pointers, since that would require loading the pointer before continuing
4097 calculation.</p>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4098
4099<p>The type of each index argument depends on the type it is indexing into.
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4100 When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4101 <b>constants</b> are allowed. When indexing into an array, pointer or
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4102 vector, integers of any width are allowed, and they are not required to be
4103 constant.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4104
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4105<p>For example, let's consider a C code fragment and how it gets compiled to
4106 LLVM:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4107
4108<div class="doc_code">
4109<pre>
4110struct RT {
4111 char A;
4112 int B[10][20];
4113 char C;
4114};
4115struct ST {
4116 int X;
4117 double Y;
4118 struct RT Z;
4119};
4120
4121int *foo(struct ST *s) {
4122 return &amp;s[1].Z.B[5][13];
4123}
4124</pre>
4125</div>
4126
4127<p>The LLVM code generated by the GCC frontend is:</p>
4128
4129<div class="doc_code">
4130<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]4131%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4132%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4133
Dan Gohman47360842009-07-25 02:23:48 +0000[diff] [blame]4134define i32* @foo(%ST* %s) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4135entry:
4136 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4137 ret i32* %reg
4138}
4139</pre>
4140</div>
4141
4142<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4143<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]4144 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4145 }</tt>' type, a structure. The second index indexes into the third element
4146 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4147 i8 }</tt>' type, another structure. The third index indexes into the second
4148 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4149 array. The two dimensions of the array are subscripted into, yielding an
4150 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4151 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4152
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4153<p>Note that it is perfectly legal to index partially through a structure,
4154 returning a pointer to an inner element. Because of this, the LLVM code for
4155 the given testcase is equivalent to:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4156
4157<pre>
Dan Gohman47360842009-07-25 02:23:48 +0000[diff] [blame]4158 define i32* @foo(%ST* %s) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4159 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4160 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4161 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4162 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4163 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4164 ret i32* %t5
4165 }
4166</pre>
4167
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4168<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman4cba1562009-07-29 16:00:30 +0000[diff] [blame]4169 <tt>getelementptr</tt> is undefined if the base pointer is not an
4170 <i>in bounds</i> address of an allocated object, or if any of the addresses
Dan Gohmanbc4c87f2009-08-20 17:08:17 +0000[diff] [blame]4171 that would be formed by successive addition of the offsets implied by the
4172 indices to the base address with infinitely precise arithmetic are not an
4173 <i>in bounds</i> address of that allocated object.
Dan Gohman4cba1562009-07-29 16:00:30 +0000[diff] [blame]4174 The <i>in bounds</i> addresses for an allocated object are all the addresses
Dan Gohmanbc4c87f2009-08-20 17:08:17 +0000[diff] [blame]4175 that point into the object, plus the address one byte past the end.</p>
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4176
4177<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4178 the base address with silently-wrapping two's complement arithmetic, and
4179 the result value of the <tt>getelementptr</tt> may be outside the object
4180 pointed to by the base pointer. The result value may not necessarily be
4181 used to access memory though, even if it happens to point into allocated
4182 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4183 section for more information.</p>
4184
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4185<p>The getelementptr instruction is often confusing. For some more insight into
4186 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4187
4188<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4189<pre>
4190 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4191 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4192 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4193 %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]4194 <i>; yields i8*:eptr</i>
4195 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta4f9a0dc2009-04-25 07:27:44 +0000[diff] [blame]4196 <i>; yields i32*:iptr</i>
Sanjiv Gupta1e46c582009-04-24 16:38:13 +0000[diff] [blame]4197 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4198</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4199
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4200</div>
4201
4202<!-- ======================================================================= -->
4203<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4204</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4205
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4206<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4207
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4208<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4209 which all take a single operand and a type. They perform various bit
4210 conversions on the operand.</p>
4211
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4212</div>
4213
4214<!-- _______________________________________________________________________ -->
4215<div class="doc_subsubsection">
4216 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4217</div>
4218<div class="doc_text">
4219
4220<h5>Syntax:</h5>
4221<pre>
4222 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4223</pre>
4224
4225<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4226<p>The '<tt>trunc</tt>' instruction truncates its operand to the
4227 type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4228
4229<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4230<p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4231 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4232 size and type of the result, which must be
4233 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4234 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4235 allowed.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4236
4237<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4238<p>The '<tt>trunc</tt>' instruction truncates the high order bits
4239 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4240 source size must be larger than the destination size, <tt>trunc</tt> cannot
4241 be a <i>no-op cast</i>. It will always truncate bits.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4242
4243<h5>Example:</h5>
4244<pre>
4245 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4246 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
4247 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
4248</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4249
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4250</div>
4251
4252<!-- _______________________________________________________________________ -->
4253<div class="doc_subsubsection">
4254 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4255</div>
4256<div class="doc_text">
4257
4258<h5>Syntax:</h5>
4259<pre>
4260 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4261</pre>
4262
4263<h5>Overview:</h5>
4264<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4265 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4266
4267
4268<h5>Arguments:</h5>
4269<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]4270 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4271 also be of <a href="#t_integer">integer</a> type. The bit size of the
4272 <tt>value</tt> must be smaller than the bit size of the destination type,
4273 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4274
4275<h5>Semantics:</h5>
4276<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]4277 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4278
4279<p>When zero extending from i1, the result will always be either 0 or 1.</p>
4280
4281<h5>Example:</h5>
4282<pre>
4283 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4284 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4285</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4286
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4287</div>
4288
4289<!-- _______________________________________________________________________ -->
4290<div class="doc_subsubsection">
4291 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4292</div>
4293<div class="doc_text">
4294
4295<h5>Syntax:</h5>
4296<pre>
4297 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4298</pre>
4299
4300<h5>Overview:</h5>
4301<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4302
4303<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4304<p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
4305 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4306 also be of <a href="#t_integer">integer</a> type. The bit size of the
4307 <tt>value</tt> must be smaller than the bit size of the destination type,
4308 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4309
4310<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4311<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4312 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4313 of the type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4314
4315<p>When sign extending from i1, the extension always results in -1 or 0.</p>
4316
4317<h5>Example:</h5>
4318<pre>
4319 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4320 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4321</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4322
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4323</div>
4324
4325<!-- _______________________________________________________________________ -->
4326<div class="doc_subsubsection">
4327 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4328</div>
4329
4330<div class="doc_text">
4331
4332<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4333<pre>
4334 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4335</pre>
4336
4337<h5>Overview:</h5>
4338<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4339 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4340
4341<h5>Arguments:</h5>
4342<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4343 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4344 to cast it to. The size of <tt>value</tt> must be larger than the size of
4345 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4346 <i>no-op cast</i>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4347
4348<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4349<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4350 <a href="#t_floating">floating point</a> type to a smaller
4351 <a href="#t_floating">floating point</a> type. If the value cannot fit
4352 within the destination type, <tt>ty2</tt>, then the results are
4353 undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4354
4355<h5>Example:</h5>
4356<pre>
4357 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4358 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4359</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4360
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4361</div>
4362
4363<!-- _______________________________________________________________________ -->
4364<div class="doc_subsubsection">
4365 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4366</div>
4367<div class="doc_text">
4368
4369<h5>Syntax:</h5>
4370<pre>
4371 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4372</pre>
4373
4374<h5>Overview:</h5>
4375<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]4376 floating point value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4377
4378<h5>Arguments:</h5>
4379<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4380 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4381 a <a href="#t_floating">floating point</a> type to cast it to. The source
4382 type must be smaller than the destination type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4383
4384<h5>Semantics:</h5>
4385<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4386 <a href="#t_floating">floating point</a> type to a larger
4387 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4388 used to make a <i>no-op cast</i> because it always changes bits. Use
4389 <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]4390
4391<h5>Example:</h5>
4392<pre>
4393 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4394 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4395</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4396
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4397</div>
4398
4399<!-- _______________________________________________________________________ -->
4400<div class="doc_subsubsection">
4401 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4402</div>
4403<div class="doc_text">
4404
4405<h5>Syntax:</h5>
4406<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4407 &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]4408</pre>
4409
4410<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4411<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4412 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4413
4414<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4415<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4416 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4417 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4418 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4419 vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4420
4421<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4422<p>The '<tt>fptoui</tt>' instruction converts its
4423 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4424 towards zero) unsigned integer value. If the value cannot fit
4425 in <tt>ty2</tt>, the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4426
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4427<h5>Example:</h5>
4428<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4429 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4430 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4431 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4432</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4433
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4434</div>
4435
4436<!-- _______________________________________________________________________ -->
4437<div class="doc_subsubsection">
4438 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4439</div>
4440<div class="doc_text">
4441
4442<h5>Syntax:</h5>
4443<pre>
4444 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4445</pre>
4446
4447<h5>Overview:</h5>
4448<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4449 <a href="#t_floating">floating point</a> <tt>value</tt> to
4450 type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4451
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4452<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4453<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4454 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4455 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4456 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4457 vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4458
4459<h5>Semantics:</h5>
4460<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4461 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4462 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4463 the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4464
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4465<h5>Example:</h5>
4466<pre>
4467 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4468 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4469 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4470</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4471
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4472</div>
4473
4474<!-- _______________________________________________________________________ -->
4475<div class="doc_subsubsection">
4476 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4477</div>
4478<div class="doc_text">
4479
4480<h5>Syntax:</h5>
4481<pre>
4482 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4483</pre>
4484
4485<h5>Overview:</h5>
4486<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4487 integer and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4489<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4490<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]4491 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4492 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4493 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4494 floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4495
4496<h5>Semantics:</h5>
4497<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4498 integer quantity and converts it to the corresponding floating point
4499 value. If the value cannot fit in the floating point value, the results are
4500 undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4501
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4502<h5>Example:</h5>
4503<pre>
4504 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4505 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4506</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4508</div>
4509
4510<!-- _______________________________________________________________________ -->
4511<div class="doc_subsubsection">
4512 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4513</div>
4514<div class="doc_text">
4515
4516<h5>Syntax:</h5>
4517<pre>
4518 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4519</pre>
4520
4521<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4522<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4523 and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4524
4525<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4526<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]4527 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4528 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4529 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4530 floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4531
4532<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4533<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4534 quantity and converts it to the corresponding floating point value. If the
4535 value cannot fit in the floating point value, the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4536
4537<h5>Example:</h5>
4538<pre>
4539 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4540 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4541</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4542
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4543</div>
4544
4545<!-- _______________________________________________________________________ -->
4546<div class="doc_subsubsection">
4547 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4548</div>
4549<div class="doc_text">
4550
4551<h5>Syntax:</h5>
4552<pre>
4553 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4554</pre>
4555
4556<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4557<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4558 the integer type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4559
4560<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4561<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4562 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4563 <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]4564
4565<h5>Semantics:</h5>
4566<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4567 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4568 truncating or zero extending that value to the size of the integer type. If
4569 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4570 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4571 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4572 change.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4573
4574<h5>Example:</h5>
4575<pre>
4576 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4577 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4578</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4579
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4580</div>
4581
4582<!-- _______________________________________________________________________ -->
4583<div class="doc_subsubsection">
4584 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4585</div>
4586<div class="doc_text">
4587
4588<h5>Syntax:</h5>
4589<pre>
4590 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4591</pre>
4592
4593<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4594<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4595 pointer type, <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4596
4597<h5>Arguments:</h5>
4598<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4599 value to cast, and a type to cast it to, which must be a
4600 <a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4601
4602<h5>Semantics:</h5>
4603<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4604 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4605 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4606 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4607 than the size of a pointer then a zero extension is done. If they are the
4608 same size, nothing is done (<i>no-op cast</i>).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4609
4610<h5>Example:</h5>
4611<pre>
4612 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4613 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4614 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4615</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4616
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4617</div>
4618
4619<!-- _______________________________________________________________________ -->
4620<div class="doc_subsubsection">
4621 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4622</div>
4623<div class="doc_text">
4624
4625<h5>Syntax:</h5>
4626<pre>
4627 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4628</pre>
4629
4630<h5>Overview:</h5>
4631<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4632 <tt>ty2</tt> without changing any bits.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4633
4634<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4635<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4636 non-aggregate first class value, and a type to cast it to, which must also be
4637 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4638 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4639 identical. If the source type is a pointer, the destination type must also be
4640 a pointer. This instruction supports bitwise conversion of vectors to
4641 integers and to vectors of other types (as long as they have the same
4642 size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4643
4644<h5>Semantics:</h5>
4645<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4646 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4647 this conversion. The conversion is done as if the <tt>value</tt> had been
4648 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4649 be converted to other pointer types with this instruction. To convert
4650 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4651 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4652
4653<h5>Example:</h5>
4654<pre>
4655 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4656 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4657 %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]4658</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4659
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4660</div>
4661
4662<!-- ======================================================================= -->
4663<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4664
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4665<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4666
4667<p>The instructions in this category are the "miscellaneous" instructions, which
4668 defy better classification.</p>
4669
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4670</div>
4671
4672<!-- _______________________________________________________________________ -->
4673<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4674</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4675
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4676<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4677
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4678<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4679<pre>
4680 &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]4681</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4682
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4683<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4684<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4685 boolean values based on comparison of its two integer, integer vector, or
4686 pointer operands.</p>
4687
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4688<h5>Arguments:</h5>
4689<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4690 the condition code indicating the kind of comparison to perform. It is not a
4691 value, just a keyword. The possible condition code are:</p>
4692
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4693<ol>
4694 <li><tt>eq</tt>: equal</li>
4695 <li><tt>ne</tt>: not equal </li>
4696 <li><tt>ugt</tt>: unsigned greater than</li>
4697 <li><tt>uge</tt>: unsigned greater or equal</li>
4698 <li><tt>ult</tt>: unsigned less than</li>
4699 <li><tt>ule</tt>: unsigned less or equal</li>
4700 <li><tt>sgt</tt>: signed greater than</li>
4701 <li><tt>sge</tt>: signed greater or equal</li>
4702 <li><tt>slt</tt>: signed less than</li>
4703 <li><tt>sle</tt>: signed less or equal</li>
4704</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4705
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4706<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4707 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4708 typed. They must also be identical types.</p>
4709
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4710<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4711<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
4712 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]4713 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]4714 result, as follows:</p>
4715
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4716<ol>
4717 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4718 <tt>false</tt> otherwise. No sign interpretation is necessary or
4719 performed.</li>
4720
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4721 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4722 <tt>false</tt> otherwise. No sign interpretation is necessary or
4723 performed.</li>
4724
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4725 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4726 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4727
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4728 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4729 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4730 to <tt>op2</tt>.</li>
4731
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4732 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4733 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4734
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4735 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4736 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4737
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4738 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4739 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4740
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4741 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4742 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4743 to <tt>op2</tt>.</li>
4744
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4745 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4746 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4747
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4748 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4749 <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]4750</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4751
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4752<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4753 values are compared as if they were integers.</p>
4754
4755<p>If the operands are integer vectors, then they are compared element by
4756 element. The result is an <tt>i1</tt> vector with the same number of elements
4757 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4758
4759<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4760<pre>
4761 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4762 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4763 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4764 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4765 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4766 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4767</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4768
4769<p>Note that the code generator does not yet support vector types with
4770 the <tt>icmp</tt> instruction.</p>
4771
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4772</div>
4773
4774<!-- _______________________________________________________________________ -->
4775<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4776</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4777
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4778<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4779
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4780<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4781<pre>
4782 &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]4783</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4784
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4785<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4786<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
4787 values based on comparison of its operands.</p>
4788
4789<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]4790(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4791
4792<p>If the operands are floating point vectors, then the result type is a vector
4793 of boolean with the same number of elements as the operands being
4794 compared.</p>
4795
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4796<h5>Arguments:</h5>
4797<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4798 the condition code indicating the kind of comparison to perform. It is not a
4799 value, just a keyword. The possible condition code are:</p>
4800
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4801<ol>
4802 <li><tt>false</tt>: no comparison, always returns false</li>
4803 <li><tt>oeq</tt>: ordered and equal</li>
4804 <li><tt>ogt</tt>: ordered and greater than </li>
4805 <li><tt>oge</tt>: ordered and greater than or equal</li>
4806 <li><tt>olt</tt>: ordered and less than </li>
4807 <li><tt>ole</tt>: ordered and less than or equal</li>
4808 <li><tt>one</tt>: ordered and not equal</li>
4809 <li><tt>ord</tt>: ordered (no nans)</li>
4810 <li><tt>ueq</tt>: unordered or equal</li>
4811 <li><tt>ugt</tt>: unordered or greater than </li>
4812 <li><tt>uge</tt>: unordered or greater than or equal</li>
4813 <li><tt>ult</tt>: unordered or less than </li>
4814 <li><tt>ule</tt>: unordered or less than or equal</li>
4815 <li><tt>une</tt>: unordered or not equal</li>
4816 <li><tt>uno</tt>: unordered (either nans)</li>
4817 <li><tt>true</tt>: no comparison, always returns true</li>
4818</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4819
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4820<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4821 <i>unordered</i> means that either operand may be a QNAN.</p>
4822
4823<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
4824 a <a href="#t_floating">floating point</a> type or
4825 a <a href="#t_vector">vector</a> of floating point type. They must have
4826 identical types.</p>
4827
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4828<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4829<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4830 according to the condition code given as <tt>cond</tt>. If the operands are
4831 vectors, then the vectors are compared element by element. Each comparison
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]4832 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4833 follows:</p>
4834
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4835<ol>
4836 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4837
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4838 <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]4839 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4840
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4841 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4842 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4843
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4844 <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]4845 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4846
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4847 <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]4848 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4849
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4850 <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]4851 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4853 <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]4854 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4855
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4856 <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]4857
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4858 <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]4859 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4860
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4861 <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]4862 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4863
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4864 <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]4865 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4866
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4867 <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]4868 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4869
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4870 <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]4871 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4872
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4873 <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]4874 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4875
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4876 <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]4877
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4878 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4879</ol>
4880
4881<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4882<pre>
4883 &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]4884 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4885 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4886 &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]4887</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4888
4889<p>Note that the code generator does not yet support vector types with
4890 the <tt>fcmp</tt> instruction.</p>
4891
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4892</div>
4893
4894<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4895<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4896 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4897</div>
4898
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4899<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4900
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4901<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4902<pre>
4903 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
4904</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4905
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4906<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4907<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
4908 SSA graph representing the function.</p>
4909
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4910<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4911<p>The type of the incoming values is specified with the first type field. After
4912 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
4913 one pair for each predecessor basic block of the current block. Only values
4914 of <a href="#t_firstclass">first class</a> type may be used as the value
4915 arguments to the PHI node. Only labels may be used as the label
4916 arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4917
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4918<p>There must be no non-phi instructions between the start of a basic block and
4919 the PHI instructions: i.e. PHI instructions must be first in a basic
4920 block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4921
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4922<p>For the purposes of the SSA form, the use of each incoming value is deemed to
4923 occur on the edge from the corresponding predecessor block to the current
4924 block (but after any definition of an '<tt>invoke</tt>' instruction's return
4925 value on the same edge).</p>
Jay Foad8e2fd2c2009-06-03 10:20:10 +0000[diff] [blame]4926
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4927<h5>Semantics:</h5>
4928<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4929 specified by the pair corresponding to the predecessor basic block that
4930 executed just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4931
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4932<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4933<pre>
4934Loop: ; Infinite loop that counts from 0 on up...
4935 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4936 %nextindvar = add i32 %indvar, 1
4937 br label %Loop
4938</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4939
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4940</div>
4941
4942<!-- _______________________________________________________________________ -->
4943<div class="doc_subsubsection">
4944 <a name="i_select">'<tt>select</tt>' Instruction</a>
4945</div>
4946
4947<div class="doc_text">
4948
4949<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4950<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4951 &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>
4952
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4953 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4954</pre>
4955
4956<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4957<p>The '<tt>select</tt>' instruction is used to choose one value based on a
4958 condition, without branching.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4959
4960
4961<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4962<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
4963 values indicating the condition, and two values of the
4964 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
4965 vectors and the condition is a scalar, then entire vectors are selected, not
4966 individual elements.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4967
4968<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4969<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
4970 first value argument; otherwise, it returns the second value argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4971
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4972<p>If the condition is a vector of i1, then the value arguments must be vectors
4973 of the same size, and the selection is done element by element.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4974
4975<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4976<pre>
4977 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4978</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4979
4980<p>Note that the code generator does not yet support conditions
4981 with vector type.</p>
4982
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4983</div>
4984
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4985<!-- _______________________________________________________________________ -->
4986<div class="doc_subsubsection">
4987 <a name="i_call">'<tt>call</tt>' Instruction</a>
4988</div>
4989
4990<div class="doc_text">
4991
4992<h5>Syntax:</h5>
4993<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4994 &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]4995</pre>
4996
4997<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4998<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4999
5000<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5001<p>This instruction requires several arguments:</p>
5002
5003<ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5004 <li>The optional "tail" marker indicates whether the callee function accesses
5005 any allocas or varargs in the caller. If the "tail" marker is present,
5006 the function call is eligible for tail call optimization. Note that calls
5007 may be marked "tail" even if they do not occur before
5008 a <a href="#i_ret"><tt>ret</tt></a> instruction.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5009
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5010 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5011 convention</a> the call should use. If none is specified, the call
5012 defaults to using C calling conventions.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5013
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5014 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5015 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5016 '<tt>inreg</tt>' attributes are valid here.</li>
5017
5018 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5019 type of the return value. Functions that return no value are marked
5020 <tt><a href="#t_void">void</a></tt>.</li>
5021
5022 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5023 being invoked. The argument types must match the types implied by this
5024 signature. This type can be omitted if the function is not varargs and if
5025 the function type does not return a pointer to a function.</li>
5026
5027 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5028 be invoked. In most cases, this is a direct function invocation, but
5029 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5030 to function value.</li>
5031
5032 <li>'<tt>function args</tt>': argument list whose types match the function
5033 signature argument types. All arguments must be of
5034 <a href="#t_firstclass">first class</a> type. If the function signature
5035 indicates the function accepts a variable number of arguments, the extra
5036 arguments can be specified.</li>
5037
5038 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5039 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5040 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5041</ol>
5042
5043<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5044<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5045 a specified function, with its incoming arguments bound to the specified
5046 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5047 function, control flow continues with the instruction after the function
5048 call, and the return value of the function is bound to the result
5049 argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5050
5051<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5052<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]5053 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]5054 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
5055 %X = tail call i32 @foo() <i>; yields i32</i>
5056 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5057 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]5058
5059 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5060 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]5061 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5062 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]5063 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]5064 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5065</pre>
5066
Dale Johannesen9fcf4a92009-09-24 18:38:21 +0000[diff] [blame]5067<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen4730cd32009-09-25 17:04:42 +0000[diff] [blame]5068standard C99 library as being the C99 library functions, and may perform
5069optimizations or generate code for them under that assumption. This is
5070something we'd like to change in the future to provide better support for
5071freestanding environments and non-C-based langauges.</p>
Dale Johannesen9fcf4a92009-09-24 18:38:21 +0000[diff] [blame]5072
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5073</div>
5074
5075<!-- _______________________________________________________________________ -->
5076<div class="doc_subsubsection">
5077 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5078</div>
5079
5080<div class="doc_text">
5081
5082<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5083<pre>
5084 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
5085</pre>
5086
5087<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5088<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5089 the "variable argument" area of a function call. It is used to implement the
5090 <tt>va_arg</tt> macro in C.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5091
5092<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5093<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5094 argument. It returns a value of the specified argument type and increments
5095 the <tt>va_list</tt> to point to the next argument. The actual type
5096 of <tt>va_list</tt> is target specific.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5097
5098<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5099<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5100 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5101 to the next argument. For more information, see the variable argument
5102 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5103
5104<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]5105 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5106 function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5107
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5108<p><tt>va_arg</tt> is an LLVM instruction instead of
5109 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5110 argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5111
5112<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5113<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5114
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5115<p>Note that the code generator does not yet fully support va_arg on many
5116 targets. Also, it does not currently support va_arg with aggregate types on
5117 any target.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]5118
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5119</div>
5120
5121<!-- *********************************************************************** -->
5122<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5123<!-- *********************************************************************** -->
5124
5125<div class="doc_text">
5126
5127<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5128 well known names and semantics and are required to follow certain
5129 restrictions. Overall, these intrinsics represent an extension mechanism for
5130 the LLVM language that does not require changing all of the transformations
5131 in LLVM when adding to the language (or the bitcode reader/writer, the
5132 parser, etc...).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5133
5134<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5135 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5136 begin with this prefix. Intrinsic functions must always be external
5137 functions: you cannot define the body of intrinsic functions. Intrinsic
5138 functions may only be used in call or invoke instructions: it is illegal to
5139 take the address of an intrinsic function. Additionally, because intrinsic
5140 functions are part of the LLVM language, it is required if any are added that
5141 they be documented here.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5142
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5143<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5144 family of functions that perform the same operation but on different data
5145 types. Because LLVM can represent over 8 million different integer types,
5146 overloading is used commonly to allow an intrinsic function to operate on any
5147 integer type. One or more of the argument types or the result type can be
5148 overloaded to accept any integer type. Argument types may also be defined as
5149 exactly matching a previous argument's type or the result type. This allows
5150 an intrinsic function which accepts multiple arguments, but needs all of them
5151 to be of the same type, to only be overloaded with respect to a single
5152 argument or the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5153
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5154<p>Overloaded intrinsics will have the names of its overloaded argument types
5155 encoded into its function name, each preceded by a period. Only those types
5156 which are overloaded result in a name suffix. Arguments whose type is matched
5157 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5158 can take an integer of any width and returns an integer of exactly the same
5159 integer width. This leads to a family of functions such as
5160 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5161 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5162 suffix is required. Because the argument's type is matched against the return
5163 type, it does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5164
5165<p>To learn how to add an intrinsic function, please see the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5166 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5167
5168</div>
5169
5170<!-- ======================================================================= -->
5171<div class="doc_subsection">
5172 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5173</div>
5174
5175<div class="doc_text">
5176
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5177<p>Variable argument support is defined in LLVM with
5178 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5179 intrinsic functions. These functions are related to the similarly named
5180 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5181
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5182<p>All of these functions operate on arguments that use a target-specific value
5183 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5184 not define what this type is, so all transformations should be prepared to
5185 handle these functions regardless of the type used.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5186
5187<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]5188 instruction and the variable argument handling intrinsic functions are
5189 used.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5190
5191<div class="doc_code">
5192<pre>
5193define i32 @test(i32 %X, ...) {
5194 ; Initialize variable argument processing
5195 %ap = alloca i8*
5196 %ap2 = bitcast i8** %ap to i8*
5197 call void @llvm.va_start(i8* %ap2)
5198
5199 ; Read a single integer argument
5200 %tmp = va_arg i8** %ap, i32
5201
5202 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5203 %aq = alloca i8*
5204 %aq2 = bitcast i8** %aq to i8*
5205 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5206 call void @llvm.va_end(i8* %aq2)
5207
5208 ; Stop processing of arguments.
5209 call void @llvm.va_end(i8* %ap2)
5210 ret i32 %tmp
5211}
5212
5213declare void @llvm.va_start(i8*)
5214declare void @llvm.va_copy(i8*, i8*)
5215declare void @llvm.va_end(i8*)
5216</pre>
5217</div>
5218
5219</div>
5220
5221<!-- _______________________________________________________________________ -->
5222<div class="doc_subsubsection">
5223 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5224</div>
5225
5226
5227<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5228
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5229<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5230<pre>
5231 declare void %llvm.va_start(i8* &lt;arglist&gt;)
5232</pre>
5233
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5234<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5235<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
5236 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]5237
5238<h5>Arguments:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5239<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]5240
5241<h5>Semantics:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5242<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]5243 macro available in C. In a target-dependent way, it initializes
5244 the <tt>va_list</tt> element to which the argument points, so that the next
5245 call to <tt>va_arg</tt> will produce the first variable argument passed to
5246 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5247 need to know the last argument of the function as the compiler can figure
5248 that out.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5249
5250</div>
5251
5252<!-- _______________________________________________________________________ -->
5253<div class="doc_subsubsection">
5254 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5255</div>
5256
5257<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5258
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5259<h5>Syntax:</h5>
5260<pre>
5261 declare void @llvm.va_end(i8* &lt;arglist&gt;)
5262</pre>
5263
5264<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5265<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5266 which has been initialized previously
5267 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5268 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5269
5270<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5271<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5272
5273<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5274<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]5275 macro available in C. In a target-dependent way, it destroys
5276 the <tt>va_list</tt> element to which the argument points. Calls
5277 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5278 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5279 with calls to <tt>llvm.va_end</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5280
5281</div>
5282
5283<!-- _______________________________________________________________________ -->
5284<div class="doc_subsubsection">
5285 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5286</div>
5287
5288<div class="doc_text">
5289
5290<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5291<pre>
5292 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
5293</pre>
5294
5295<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5296<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5297 from the source argument list to the destination argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5298
5299<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5300<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]5301 The second argument is a pointer to a <tt>va_list</tt> element to copy
5302 from.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5303
5304<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5305<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]5306 macro available in C. In a target-dependent way, it copies the
5307 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5308 element. This intrinsic is necessary because
5309 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5310 arbitrarily complex and require, for example, memory allocation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5311
5312</div>
5313
5314<!-- ======================================================================= -->
5315<div class="doc_subsection">
5316 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5317</div>
5318
5319<div class="doc_text">
5320
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5321<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5322Collection</a> (GC) requires the implementation and generation of these
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5323intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5324roots on the stack</a>, as well as garbage collector implementations that
5325require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5326barriers. Front-ends for type-safe garbage collected languages should generate
5327these intrinsics to make use of the LLVM garbage collectors. For more details,
5328see <a href="GarbageCollection.html">Accurate Garbage Collection with
5329LLVM</a>.</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]5330
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5331<p>The garbage collection intrinsics only operate on objects in the generic
5332 address space (address space zero).</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]5333
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5334</div>
5335
5336<!-- _______________________________________________________________________ -->
5337<div class="doc_subsubsection">
5338 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5339</div>
5340
5341<div class="doc_text">
5342
5343<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5344<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5345 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5346</pre>
5347
5348<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5349<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]5350 the code generator, and allows some metadata to be associated with it.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5351
5352<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5353<p>The first argument specifies the address of a stack object that contains the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5354 root pointer. The second pointer (which must be either a constant or a
5355 global value address) contains the meta-data to be associated with the
5356 root.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5357
5358<h5>Semantics:</h5>
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]5359<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]5360 location. At compile-time, the code generator generates information to allow
5361 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5362 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5363 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5364
5365</div>
5366
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5367<!-- _______________________________________________________________________ -->
5368<div class="doc_subsubsection">
5369 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5370</div>
5371
5372<div class="doc_text">
5373
5374<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5375<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5376 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5377</pre>
5378
5379<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5380<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5381 locations, allowing garbage collector implementations that require read
5382 barriers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5383
5384<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5385<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]5386 allocated from the garbage collector. The first object is a pointer to the
5387 start of the referenced object, if needed by the language runtime (otherwise
5388 null).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5389
5390<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5391<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5392 instruction, but may be replaced with substantially more complex code by the
5393 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5394 may only be used in a function which <a href="#gc">specifies a GC
5395 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5396
5397</div>
5398
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5399<!-- _______________________________________________________________________ -->
5400<div class="doc_subsubsection">
5401 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5402</div>
5403
5404<div class="doc_text">
5405
5406<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5407<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5408 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5409</pre>
5410
5411<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5412<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5413 locations, allowing garbage collector implementations that require write
5414 barriers (such as generational or reference counting collectors).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5415
5416<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5417<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]5418 object to store it to, and the third is the address of the field of Obj to
5419 store to. If the runtime does not require a pointer to the object, Obj may
5420 be null.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5421
5422<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5423<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5424 instruction, but may be replaced with substantially more complex code by the
5425 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5426 may only be used in a function which <a href="#gc">specifies a GC
5427 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5428
5429</div>
5430
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5431<!-- ======================================================================= -->
5432<div class="doc_subsection">
5433 <a name="int_codegen">Code Generator Intrinsics</a>
5434</div>
5435
5436<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5437
5438<p>These intrinsics are provided by LLVM to expose special features that may
5439 only be implemented with code generator support.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5440
5441</div>
5442
5443<!-- _______________________________________________________________________ -->
5444<div class="doc_subsubsection">
5445 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5446</div>
5447
5448<div class="doc_text">
5449
5450<h5>Syntax:</h5>
5451<pre>
5452 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
5453</pre>
5454
5455<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5456<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5457 target-specific value indicating the return address of the current function
5458 or one of its callers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5459
5460<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5461<p>The argument to this intrinsic indicates which function to return the address
5462 for. Zero indicates the calling function, one indicates its caller, etc.
5463 The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5464
5465<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5466<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5467 indicating the return address of the specified call frame, or zero if it
5468 cannot be identified. The value returned by this intrinsic is likely to be
5469 incorrect or 0 for arguments other than zero, so it should only be used for
5470 debugging purposes.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5471
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5472<p>Note that calling this intrinsic does not prevent function inlining or other
5473 aggressive transformations, so the value returned may not be that of the
5474 obvious source-language caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5475
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5476</div>
5477
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5478<!-- _______________________________________________________________________ -->
5479<div class="doc_subsubsection">
5480 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5481</div>
5482
5483<div class="doc_text">
5484
5485<h5>Syntax:</h5>
5486<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5487 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5488</pre>
5489
5490<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5491<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5492 target-specific frame pointer value for the specified stack frame.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5493
5494<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5495<p>The argument to this intrinsic indicates which function to return the frame
5496 pointer for. Zero indicates the calling function, one indicates its caller,
5497 etc. The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5498
5499<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5500<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5501 indicating the frame address of the specified call frame, or zero if it
5502 cannot be identified. The value returned by this intrinsic is likely to be
5503 incorrect or 0 for arguments other than zero, so it should only be used for
5504 debugging purposes.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5505
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5506<p>Note that calling this intrinsic does not prevent function inlining or other
5507 aggressive transformations, so the value returned may not be that of the
5508 obvious source-language caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5509
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5510</div>
5511
5512<!-- _______________________________________________________________________ -->
5513<div class="doc_subsubsection">
5514 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5515</div>
5516
5517<div class="doc_text">
5518
5519<h5>Syntax:</h5>
5520<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5521 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5522</pre>
5523
5524<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5525<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5526 of the function stack, for use
5527 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5528 useful for implementing language features like scoped automatic variable
5529 sized arrays in C99.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5530
5531<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5532<p>This intrinsic returns a opaque pointer value that can be passed
5533 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5534 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5535 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5536 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5537 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5538 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5539
5540</div>
5541
5542<!-- _______________________________________________________________________ -->
5543<div class="doc_subsubsection">
5544 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5545</div>
5546
5547<div class="doc_text">
5548
5549<h5>Syntax:</h5>
5550<pre>
5551 declare void @llvm.stackrestore(i8 * %ptr)
5552</pre>
5553
5554<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5555<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5556 the function stack to the state it was in when the
5557 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5558 executed. This is useful for implementing language features like scoped
5559 automatic variable sized arrays in C99.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5560
5561<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5562<p>See the description
5563 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5564
5565</div>
5566
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5567<!-- _______________________________________________________________________ -->
5568<div class="doc_subsubsection">
5569 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5570</div>
5571
5572<div class="doc_text">
5573
5574<h5>Syntax:</h5>
5575<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5576 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]5577</pre>
5578
5579<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5580<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5581 insert a prefetch instruction if supported; otherwise, it is a noop.
5582 Prefetches have no effect on the behavior of the program but can change its
5583 performance characteristics.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5584
5585<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5586<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5587 specifier determining if the fetch should be for a read (0) or write (1),
5588 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5589 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5590 and <tt>locality</tt> arguments must be constant integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5591
5592<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5593<p>This intrinsic does not modify the behavior of the program. In particular,
5594 prefetches cannot trap and do not produce a value. On targets that support
5595 this intrinsic, the prefetch can provide hints to the processor cache for
5596 better performance.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5597
5598</div>
5599
5600<!-- _______________________________________________________________________ -->
5601<div class="doc_subsubsection">
5602 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5603</div>
5604
5605<div class="doc_text">
5606
5607<h5>Syntax:</h5>
5608<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5609 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5610</pre>
5611
5612<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5613<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5614 Counter (PC) in a region of code to simulators and other tools. The method
5615 is target specific, but it is expected that the marker will use exported
5616 symbols to transmit the PC of the marker. The marker makes no guarantees
5617 that it will remain with any specific instruction after optimizations. It is
5618 possible that the presence of a marker will inhibit optimizations. The
5619 intended use is to be inserted after optimizations to allow correlations of
5620 simulation runs.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5621
5622<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5623<p><tt>id</tt> is a numerical id identifying the marker.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5624
5625<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5626<p>This intrinsic does not modify the behavior of the program. Backends that do
5627 not support this intrinisic may ignore it.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5628
5629</div>
5630
5631<!-- _______________________________________________________________________ -->
5632<div class="doc_subsubsection">
5633 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5634</div>
5635
5636<div class="doc_text">
5637
5638<h5>Syntax:</h5>
5639<pre>
5640 declare i64 @llvm.readcyclecounter( )
5641</pre>
5642
5643<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5644<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5645 counter register (or similar low latency, high accuracy clocks) on those
5646 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5647 should map to RPCC. As the backing counters overflow quickly (on the order
5648 of 9 seconds on alpha), this should only be used for small timings.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5649
5650<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5651<p>When directly supported, reading the cycle counter should not modify any
5652 memory. Implementations are allowed to either return a application specific
5653 value or a system wide value. On backends without support, this is lowered
5654 to a constant 0.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5655
5656</div>
5657
5658<!-- ======================================================================= -->
5659<div class="doc_subsection">
5660 <a name="int_libc">Standard C Library Intrinsics</a>
5661</div>
5662
5663<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5664
5665<p>LLVM provides intrinsics for a few important standard C library functions.
5666 These intrinsics allow source-language front-ends to pass information about
5667 the alignment of the pointer arguments to the code generator, providing
5668 opportunity for more efficient code generation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5669
5670</div>
5671
5672<!-- _______________________________________________________________________ -->
5673<div class="doc_subsubsection">
5674 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5675</div>
5676
5677<div class="doc_text">
5678
5679<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5680<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5681 integer bit width. Not all targets support all bit widths however.</p>
5682
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5683<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5684 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5685 i8 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5686 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5687 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5688 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5689 i32 &lt;len&gt;, i32 &lt;align&gt;)
5690 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5691 i64 &lt;len&gt;, i32 &lt;align&gt;)
5692</pre>
5693
5694<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5695<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5696 source location to the destination location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5697
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5698<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5699 intrinsics do not return a value, and takes an extra alignment argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5700
5701<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5702<p>The first argument is a pointer to the destination, the second is a pointer
5703 to the source. The third argument is an integer argument specifying the
5704 number of bytes to copy, and the fourth argument is the alignment of the
5705 source and destination locations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5706
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5707<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5708 then the caller guarantees that both the source and destination pointers are
5709 aligned to that boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5710
5711<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5712<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5713 source location to the destination location, which are not allowed to
5714 overlap. It copies "len" bytes of memory over. If the argument is known to
5715 be aligned to some boundary, this can be specified as the fourth argument,
5716 otherwise it should be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5717
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5718</div>
5719
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5720<!-- _______________________________________________________________________ -->
5721<div class="doc_subsubsection">
5722 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5723</div>
5724
5725<div class="doc_text">
5726
5727<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5728<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5729 width. Not all targets support all bit widths however.</p>
5730
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5731<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5732 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5733 i8 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5734 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5735 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5736 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5737 i32 &lt;len&gt;, i32 &lt;align&gt;)
5738 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5739 i64 &lt;len&gt;, i32 &lt;align&gt;)
5740</pre>
5741
5742<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5743<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
5744 source location to the destination location. It is similar to the
5745 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
5746 overlap.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5747
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5748<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5749 intrinsics do not return a value, and takes an extra alignment argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5750
5751<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5752<p>The first argument is a pointer to the destination, the second is a pointer
5753 to the source. The third argument is an integer argument specifying the
5754 number of bytes to copy, and the fourth argument is the alignment of the
5755 source and destination locations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5756
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5757<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5758 then the caller guarantees that the source and destination pointers are
5759 aligned to that boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5760
5761<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5762<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
5763 source location to the destination location, which may overlap. It copies
5764 "len" bytes of memory over. If the argument is known to be aligned to some
5765 boundary, this can be specified as the fourth argument, otherwise it should
5766 be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5767
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5768</div>
5769
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5770<!-- _______________________________________________________________________ -->
5771<div class="doc_subsubsection">
5772 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5773</div>
5774
5775<div class="doc_text">
5776
5777<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5778<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5779 width. Not all targets support all bit widths however.</p>
5780
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5781<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5782 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5783 i8 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5784 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5785 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5786 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5787 i32 &lt;len&gt;, i32 &lt;align&gt;)
5788 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5789 i64 &lt;len&gt;, i32 &lt;align&gt;)
5790</pre>
5791
5792<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5793<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
5794 particular byte value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5795
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5796<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
5797 intrinsic does not return a value, and takes an extra alignment argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5798
5799<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5800<p>The first argument is a pointer to the destination to fill, the second is the
5801 byte value to fill it with, the third argument is an integer argument
5802 specifying the number of bytes to fill, and the fourth argument is the known
5803 alignment of destination location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5804
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5805<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5806 then the caller guarantees that the destination pointer is aligned to that
5807 boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5808
5809<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5810<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
5811 at the destination location. If the argument is known to be aligned to some
5812 boundary, this can be specified as the fourth argument, otherwise it should
5813 be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5814
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5815</div>
5816
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5817<!-- _______________________________________________________________________ -->
5818<div class="doc_subsubsection">
5819 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5820</div>
5821
5822<div class="doc_text">
5823
5824<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5825<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5826 floating point or vector of floating point type. Not all targets support all
5827 types however.</p>
5828
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5829<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5830 declare float @llvm.sqrt.f32(float %Val)
5831 declare double @llvm.sqrt.f64(double %Val)
5832 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5833 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5834 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5835</pre>
5836
5837<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5838<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5839 returning the same value as the libm '<tt>sqrt</tt>' functions would.
5840 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
5841 behavior for negative numbers other than -0.0 (which allows for better
5842 optimization, because there is no need to worry about errno being
5843 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]5844
5845<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5846<p>The argument and return value are floating point numbers of the same
5847 type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5848
5849<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5850<p>This function returns the sqrt of the specified operand if it is a
5851 nonnegative floating point number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5853</div>
5854
5855<!-- _______________________________________________________________________ -->
5856<div class="doc_subsubsection">
5857 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5858</div>
5859
5860<div class="doc_text">
5861
5862<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5863<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5864 floating point or vector of floating point type. Not all targets support all
5865 types however.</p>
5866
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5867<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5868 declare float @llvm.powi.f32(float %Val, i32 %power)
5869 declare double @llvm.powi.f64(double %Val, i32 %power)
5870 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5871 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5872 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5873</pre>
5874
5875<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5876<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5877 specified (positive or negative) power. The order of evaluation of
5878 multiplications is not defined. When a vector of floating point type is
5879 used, the second argument remains a scalar integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5880
5881<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5882<p>The second argument is an integer power, and the first is a value to raise to
5883 that power.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5884
5885<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5886<p>This function returns the first value raised to the second power with an
5887 unspecified sequence of rounding operations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5888
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5889</div>
5890
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5891<!-- _______________________________________________________________________ -->
5892<div class="doc_subsubsection">
5893 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5894</div>
5895
5896<div class="doc_text">
5897
5898<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5899<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5900 floating point or vector of floating point type. Not all targets support all
5901 types however.</p>
5902
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5903<pre>
5904 declare float @llvm.sin.f32(float %Val)
5905 declare double @llvm.sin.f64(double %Val)
5906 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5907 declare fp128 @llvm.sin.f128(fp128 %Val)
5908 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5909</pre>
5910
5911<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5912<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5913
5914<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5915<p>The argument and return value are floating point numbers of the same
5916 type.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5917
5918<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5919<p>This function returns the sine of the specified operand, returning the same
5920 values as the libm <tt>sin</tt> functions would, and handles error conditions
5921 in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5922
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5923</div>
5924
5925<!-- _______________________________________________________________________ -->
5926<div class="doc_subsubsection">
5927 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5928</div>
5929
5930<div class="doc_text">
5931
5932<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5933<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5934 floating point or vector of floating point type. Not all targets support all
5935 types however.</p>
5936
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5937<pre>
5938 declare float @llvm.cos.f32(float %Val)
5939 declare double @llvm.cos.f64(double %Val)
5940 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5941 declare fp128 @llvm.cos.f128(fp128 %Val)
5942 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5943</pre>
5944
5945<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5946<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5947
5948<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5949<p>The argument and return value are floating point numbers of the same
5950 type.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5951
5952<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5953<p>This function returns the cosine of the specified operand, returning the same
5954 values as the libm <tt>cos</tt> functions would, and handles error conditions
5955 in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5956
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5957</div>
5958
5959<!-- _______________________________________________________________________ -->
5960<div class="doc_subsubsection">
5961 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5962</div>
5963
5964<div class="doc_text">
5965
5966<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5967<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5968 floating point or vector of floating point type. Not all targets support all
5969 types however.</p>
5970
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5971<pre>
5972 declare float @llvm.pow.f32(float %Val, float %Power)
5973 declare double @llvm.pow.f64(double %Val, double %Power)
5974 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5975 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5976 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5977</pre>
5978
5979<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5980<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5981 specified (positive or negative) power.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5982
5983<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5984<p>The second argument is a floating point power, and the first is a value to
5985 raise to that power.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5986
5987<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5988<p>This function returns the first value raised to the second power, returning
5989 the same values as the libm <tt>pow</tt> functions would, and handles error
5990 conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5991
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5992</div>
5993
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5994<!-- ======================================================================= -->
5995<div class="doc_subsection">
5996 <a name="int_manip">Bit Manipulation Intrinsics</a>
5997</div>
5998
5999<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6000
6001<p>LLVM provides intrinsics for a few important bit manipulation operations.
6002 These allow efficient code generation for some algorithms.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6003
6004</div>
6005
6006<!-- _______________________________________________________________________ -->
6007<div class="doc_subsubsection">
6008 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6009</div>
6010
6011<div class="doc_text">
6012
6013<h5>Syntax:</h5>
6014<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6015 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6016
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6017<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6018 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
6019 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
6020 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6021</pre>
6022
6023<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6024<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6025 values with an even number of bytes (positive multiple of 16 bits). These
6026 are useful for performing operations on data that is not in the target's
6027 native byte order.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6028
6029<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6030<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6031 and low byte of the input i16 swapped. Similarly,
6032 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6033 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6034 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6035 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6036 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6037 more, respectively).</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_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6044</div>
6045
6046<div class="doc_text">
6047
6048<h5>Syntax:</h5>
6049<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]6050 width. Not all targets support all bit widths however.</p>
6051
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6052<pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6053 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6054 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6055 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6056 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
6057 declare i256 @llvm.ctpop.i256(i256 &lt;src&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.ctpop</tt>' family of intrinsics counts the number of bits set
6062 in a value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6063
6064<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6065<p>The only argument is the value to be counted. The argument may be of any
6066 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6067
6068<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6069<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]6070
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6071</div>
6072
6073<!-- _______________________________________________________________________ -->
6074<div class="doc_subsubsection">
6075 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6076</div>
6077
6078<div class="doc_text">
6079
6080<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6081<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6082 integer bit width. Not all targets support all bit widths however.</p>
6083
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6084<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6085 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
6086 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6087 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6088 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
6089 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6090</pre>
6091
6092<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6093<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6094 leading zeros in a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6095
6096<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6097<p>The only argument is the value to be counted. The argument may be of any
6098 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6099
6100<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6101<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6102 zeros in a variable. If the src == 0 then the result is the size in bits of
6103 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6104
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6105</div>
6106
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6107<!-- _______________________________________________________________________ -->
6108<div class="doc_subsubsection">
6109 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6110</div>
6111
6112<div class="doc_text">
6113
6114<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6115<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6116 integer bit width. Not all targets support all bit widths however.</p>
6117
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6118<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6119 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
6120 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6121 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6122 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
6123 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6124</pre>
6125
6126<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6127<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6128 trailing zeros.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6129
6130<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6131<p>The only argument is the value to be counted. The argument may be of any
6132 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6133
6134<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6135<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6136 zeros in a variable. If the src == 0 then the result is the size in bits of
6137 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6138
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6139</div>
6140
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6141<!-- ======================================================================= -->
6142<div class="doc_subsection">
6143 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6144</div>
6145
6146<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6147
6148<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6149
6150</div>
6151
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6152<!-- _______________________________________________________________________ -->
6153<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6154 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6155</div>
6156
6157<div class="doc_text">
6158
6159<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6160<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]6161 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6162
6163<pre>
6164 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6165 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6166 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6167</pre>
6168
6169<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6170<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6171 a signed addition of the two arguments, and indicate whether an overflow
6172 occurred during the signed summation.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6173
6174<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6175<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]6176 be of integer types of any bit width, but they must have the same bit
6177 width. The second element of the result structure must be of
6178 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6179 undergo signed addition.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6180
6181<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6182<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6183 a signed addition of the two variables. They return a structure &mdash; the
6184 first element of which is the signed summation, and the second element of
6185 which is a bit specifying if the signed summation resulted in an
6186 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6187
6188<h5>Examples:</h5>
6189<pre>
6190 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6191 %sum = extractvalue {i32, i1} %res, 0
6192 %obit = extractvalue {i32, i1} %res, 1
6193 br i1 %obit, label %overflow, label %normal
6194</pre>
6195
6196</div>
6197
6198<!-- _______________________________________________________________________ -->
6199<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6200 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6201</div>
6202
6203<div class="doc_text">
6204
6205<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6206<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]6207 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6208
6209<pre>
6210 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6211 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6212 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6213</pre>
6214
6215<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6216<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6217 an unsigned addition of the two arguments, and indicate whether a carry
6218 occurred during the unsigned summation.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6219
6220<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6221<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]6222 be of integer types of any bit width, but they must have the same bit
6223 width. The second element of the result structure must be of
6224 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6225 undergo unsigned addition.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6226
6227<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6228<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6229 an unsigned addition of the two arguments. They return a structure &mdash;
6230 the first element of which is the sum, and the second element of which is a
6231 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6232
6233<h5>Examples:</h5>
6234<pre>
6235 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6236 %sum = extractvalue {i32, i1} %res, 0
6237 %obit = extractvalue {i32, i1} %res, 1
6238 br i1 %obit, label %carry, label %normal
6239</pre>
6240
6241</div>
6242
6243<!-- _______________________________________________________________________ -->
6244<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6245 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6246</div>
6247
6248<div class="doc_text">
6249
6250<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6251<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]6252 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6253
6254<pre>
6255 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6256 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6257 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6258</pre>
6259
6260<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6261<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6262 a signed subtraction of the two arguments, and indicate whether an overflow
6263 occurred during the signed subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6264
6265<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6266<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]6267 be of integer types of any bit width, but they must have the same bit
6268 width. The second element of the result structure must be of
6269 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6270 undergo signed subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6271
6272<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6273<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6274 a signed subtraction of the two arguments. They return a structure &mdash;
6275 the first element of which is the subtraction, and the second element of
6276 which is a bit specifying if the signed subtraction resulted in an
6277 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6278
6279<h5>Examples:</h5>
6280<pre>
6281 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6282 %sum = extractvalue {i32, i1} %res, 0
6283 %obit = extractvalue {i32, i1} %res, 1
6284 br i1 %obit, label %overflow, label %normal
6285</pre>
6286
6287</div>
6288
6289<!-- _______________________________________________________________________ -->
6290<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6291 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6292</div>
6293
6294<div class="doc_text">
6295
6296<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6297<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]6298 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6299
6300<pre>
6301 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6302 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6303 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6304</pre>
6305
6306<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6307<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6308 an unsigned subtraction of the two arguments, and indicate whether an
6309 overflow occurred during the unsigned subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6310
6311<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6312<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]6313 be of integer types of any bit width, but they must have the same bit
6314 width. The second element of the result structure must be of
6315 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6316 undergo unsigned subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6317
6318<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6319<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6320 an unsigned subtraction of the two arguments. They return a structure &mdash;
6321 the first element of which is the subtraction, and the second element of
6322 which is a bit specifying if the unsigned subtraction resulted in an
6323 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6324
6325<h5>Examples:</h5>
6326<pre>
6327 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6328 %sum = extractvalue {i32, i1} %res, 0
6329 %obit = extractvalue {i32, i1} %res, 1
6330 br i1 %obit, label %overflow, label %normal
6331</pre>
6332
6333</div>
6334
6335<!-- _______________________________________________________________________ -->
6336<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6337 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6338</div>
6339
6340<div class="doc_text">
6341
6342<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6343<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]6344 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6345
6346<pre>
6347 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6348 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6349 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6350</pre>
6351
6352<h5>Overview:</h5>
6353
6354<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6355 a signed multiplication of the two arguments, and indicate whether an
6356 overflow occurred during the signed multiplication.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6357
6358<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6359<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]6360 be of integer types of any bit width, but they must have the same bit
6361 width. The second element of the result structure must be of
6362 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6363 undergo signed multiplication.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6364
6365<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6366<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6367 a signed multiplication of the two arguments. They return a structure &mdash;
6368 the first element of which is the multiplication, and the second element of
6369 which is a bit specifying if the signed multiplication resulted in an
6370 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6371
6372<h5>Examples:</h5>
6373<pre>
6374 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6375 %sum = extractvalue {i32, i1} %res, 0
6376 %obit = extractvalue {i32, i1} %res, 1
6377 br i1 %obit, label %overflow, label %normal
6378</pre>
6379
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6380</div>
6381
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6382<!-- _______________________________________________________________________ -->
6383<div class="doc_subsubsection">
6384 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6385</div>
6386
6387<div class="doc_text">
6388
6389<h5>Syntax:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6390<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]6391 on any integer bit width.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6392
6393<pre>
6394 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6395 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6396 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6397</pre>
6398
6399<h5>Overview:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6400<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6401 a unsigned multiplication of the two arguments, and indicate whether an
6402 overflow occurred during the unsigned multiplication.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6403
6404<h5>Arguments:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6405<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]6406 be of integer types of any bit width, but they must have the same bit
6407 width. The second element of the result structure must be of
6408 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6409 undergo unsigned multiplication.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6410
6411<h5>Semantics:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6412<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6413 an unsigned multiplication of the two arguments. They return a structure
6414 &mdash; the first element of which is the multiplication, and the second
6415 element of which is a bit specifying if the unsigned multiplication resulted
6416 in an overflow.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6417
6418<h5>Examples:</h5>
6419<pre>
6420 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6421 %sum = extractvalue {i32, i1} %res, 0
6422 %obit = extractvalue {i32, i1} %res, 1
6423 br i1 %obit, label %overflow, label %normal
6424</pre>
6425
6426</div>
6427
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6428<!-- ======================================================================= -->
6429<div class="doc_subsection">
6430 <a name="int_debugger">Debugger Intrinsics</a>
6431</div>
6432
6433<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6434
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6435<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6436 prefix), are described in
6437 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6438 Level Debugging</a> document.</p>
6439
6440</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6441
6442<!-- ======================================================================= -->
6443<div class="doc_subsection">
6444 <a name="int_eh">Exception Handling Intrinsics</a>
6445</div>
6446
6447<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6448
6449<p>The LLVM exception handling intrinsics (which all start with
6450 <tt>llvm.eh.</tt> prefix), are described in
6451 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6452 Handling</a> document.</p>
6453
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6454</div>
6455
6456<!-- ======================================================================= -->
6457<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6458 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6459</div>
6460
6461<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6462
6463<p>This intrinsic makes it possible to excise one parameter, marked with
6464 the <tt>nest</tt> attribute, from a function. The result is a callable
6465 function pointer lacking the nest parameter - the caller does not need to
6466 provide a value for it. Instead, the value to use is stored in advance in a
6467 "trampoline", a block of memory usually allocated on the stack, which also
6468 contains code to splice the nest value into the argument list. This is used
6469 to implement the GCC nested function address extension.</p>
6470
6471<p>For example, if the function is
6472 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6473 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6474 follows:</p>
6475
6476<div class="doc_code">
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6477<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6478 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6479 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6480 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6481 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6482</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6483</div>
6484
6485<p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6486 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6487
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6488</div>
6489
6490<!-- _______________________________________________________________________ -->
6491<div class="doc_subsubsection">
6492 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6493</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6494
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6495<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6496
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6497<h5>Syntax:</h5>
6498<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6499 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]6500</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6501
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6502<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6503<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6504 function pointer suitable for executing it.</p>
6505
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6506<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6507<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6508 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6509 sufficiently aligned block of memory; this memory is written to by the
6510 intrinsic. Note that the size and the alignment are target-specific - LLVM
6511 currently provides no portable way of determining them, so a front-end that
6512 generates this intrinsic needs to have some target-specific knowledge.
6513 The <tt>func</tt> argument must hold a function bitcast to
6514 an <tt>i8*</tt>.</p>
6515
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6516<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6517<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6518 dependent code, turning it into a function. A pointer to this function is
6519 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6520 function pointer type</a> before being called. The new function's signature
6521 is the same as that of <tt>func</tt> with any arguments marked with
6522 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6523 is allowed, and it must be of pointer type. Calling the new function is
6524 equivalent to calling <tt>func</tt> with the same argument list, but
6525 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6526 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6527 by <tt>tramp</tt> is modified, then the effect of any later call to the
6528 returned function pointer is undefined.</p>
6529
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6530</div>
6531
6532<!-- ======================================================================= -->
6533<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6534 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6535</div>
6536
6537<div class="doc_text">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6538
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6539<p>These intrinsic functions expand the "universal IR" of LLVM to represent
6540 hardware constructs for atomic operations and memory synchronization. This
6541 provides an interface to the hardware, not an interface to the programmer. It
6542 is aimed at a low enough level to allow any programming models or APIs
6543 (Application Programming Interfaces) which need atomic behaviors to map
6544 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6545 hardware provides a "universal IR" for source languages, it also provides a
6546 starting point for developing a "universal" atomic operation and
6547 synchronization IR.</p>
6548
6549<p>These do <em>not</em> form an API such as high-level threading libraries,
6550 software transaction memory systems, atomic primitives, and intrinsic
6551 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6552 application libraries. The hardware interface provided by LLVM should allow
6553 a clean implementation of all of these APIs and parallel programming models.
6554 No one model or paradigm should be selected above others unless the hardware
6555 itself ubiquitously does so.</p>
6556
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6557</div>
6558
6559<!-- _______________________________________________________________________ -->
6560<div class="doc_subsubsection">
6561 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6562</div>
6563<div class="doc_text">
6564<h5>Syntax:</h5>
6565<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6566 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]6567</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6568
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6569<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6570<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6571 specific pairs of memory access types.</p>
6572
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6573<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6574<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6575 The first four arguments enables a specific barrier as listed below. The
6576 fith argument specifies that the barrier applies to io or device or uncached
6577 memory.</p>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6578
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6579<ul>
6580 <li><tt>ll</tt>: load-load barrier</li>
6581 <li><tt>ls</tt>: load-store barrier</li>
6582 <li><tt>sl</tt>: store-load barrier</li>
6583 <li><tt>ss</tt>: store-store barrier</li>
6584 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6585</ul>
6586
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6587<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6588<p>This intrinsic causes the system to enforce some ordering constraints upon
6589 the loads and stores of the program. This barrier does not
6590 indicate <em>when</em> any events will occur, it only enforces
6591 an <em>order</em> in which they occur. For any of the specified pairs of load
6592 and store operations (f.ex. load-load, or store-load), all of the first
6593 operations preceding the barrier will complete before any of the second
6594 operations succeeding the barrier begin. Specifically the semantics for each
6595 pairing is as follows:</p>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6596
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6597<ul>
6598 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6599 after the barrier begins.</li>
6600 <li><tt>ls</tt>: All loads before the barrier must complete before any
6601 store after the barrier begins.</li>
6602 <li><tt>ss</tt>: All stores before the barrier must complete before any
6603 store after the barrier begins.</li>
6604 <li><tt>sl</tt>: All stores before the barrier must complete before any
6605 load after the barrier begins.</li>
6606</ul>
6607
6608<p>These semantics are applied with a logical "and" behavior when more than one
6609 is enabled in a single memory barrier intrinsic.</p>
6610
6611<p>Backends may implement stronger barriers than those requested when they do
6612 not support as fine grained a barrier as requested. Some architectures do
6613 not need all types of barriers and on such architectures, these become
6614 noops.</p>
6615
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6616<h5>Example:</h5>
6617<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6618%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6619%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6620 store i32 4, %ptr
6621
6622%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6623 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6624 <i>; guarantee the above finishes</i>
6625 store i32 8, %ptr <i>; before this begins</i>
6626</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6627
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6628</div>
6629
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6630<!-- _______________________________________________________________________ -->
6631<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6632 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6633</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6634
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6635<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6636
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6637<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6638<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6639 any integer bit width and for different address spaces. Not all targets
6640 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6641
6642<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6643 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
6644 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
6645 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
6646 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]6647</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6648
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6649<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6650<p>This loads a value in memory and compares it to a given value. If they are
6651 equal, it stores a new value into the memory.</p>
6652
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6653<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6654<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
6655 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6656 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6657 this integer type. While any bit width integer may be used, targets may only
6658 lower representations they support in hardware.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6659
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6660<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6661<p>This entire intrinsic must be executed atomically. It first loads the value
6662 in memory pointed to by <tt>ptr</tt> and compares it with the
6663 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
6664 memory. The loaded value is yielded in all cases. This provides the
6665 equivalent of an atomic compare-and-swap operation within the SSA
6666 framework.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6667
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6668<h5>Examples:</h5>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6669<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6670%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6671%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6672 store i32 4, %ptr
6673
6674%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6675%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6676 <i>; yields {i32}:result1 = 4</i>
6677%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6678%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6679
6680%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6681%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6682 <i>; yields {i32}:result2 = 8</i>
6683%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6684
6685%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6686</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6687
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6688</div>
6689
6690<!-- _______________________________________________________________________ -->
6691<div class="doc_subsubsection">
6692 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6693</div>
6694<div class="doc_text">
6695<h5>Syntax:</h5>
6696
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6697<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6698 integer bit width. Not all targets support all bit widths however.</p>
6699
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6700<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6701 declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6702 declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6703 declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6704 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]6705</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6706
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6707<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6708<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6709 the value from memory. It then stores the value in <tt>val</tt> in the memory
6710 at <tt>ptr</tt>.</p>
6711
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6712<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6713<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
6714 the <tt>val</tt> argument and the result must be integers of the same bit
6715 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6716 integer type. The targets may only lower integer representations they
6717 support.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6718
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6719<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6720<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6721 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6722 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6723
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6724<h5>Examples:</h5>
6725<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6726%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6727%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6728 store i32 4, %ptr
6729
6730%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6731%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6732 <i>; yields {i32}:result1 = 4</i>
6733%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6734%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6735
6736%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6737%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6738 <i>; yields {i32}:result2 = 8</i>
6739
6740%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6741%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6742</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6743
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6744</div>
6745
6746<!-- _______________________________________________________________________ -->
6747<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6748 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6749
6750</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6751
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6752<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6753
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6754<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6755<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
6756 any integer bit width. Not all targets support all bit widths however.</p>
6757
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6758<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6759 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6760 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6761 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6762 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]6763</pre>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6764
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6765<h5>Overview:</h5>
6766<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
6767 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6768
6769<h5>Arguments:</h5>
6770<p>The intrinsic takes two arguments, the first a pointer to an integer value
6771 and the second an integer value. The result is also an integer value. These
6772 integer types can have any bit width, but they must all have the same bit
6773 width. The targets may only lower integer representations they support.</p>
6774
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6775<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6776<p>This intrinsic does a series of operations atomically. It first loads the
6777 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6778 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]6779
6780<h5>Examples:</h5>
6781<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6782%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6783%ptr = bitcast i8* %mallocP to i32*
6784 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6785%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6786 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6787%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6788 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6789%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6790 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6791%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6792</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6793
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6794</div>
6795
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6796<!-- _______________________________________________________________________ -->
6797<div class="doc_subsubsection">
6798 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6799
6800</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6801
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6802<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6803
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6804<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6805<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6806 any integer bit width and for different address spaces. Not all targets
6807 support all bit widths however.</p>
6808
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6809<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6810 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6811 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6812 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6813 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]6814</pre>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6815
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6816<h5>Overview:</h5>
6817<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6818 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6819
6820<h5>Arguments:</h5>
6821<p>The intrinsic takes two arguments, the first a pointer to an integer value
6822 and the second an integer value. The result is also an integer value. These
6823 integer types can have any bit width, but they must all have the same bit
6824 width. The targets may only lower integer representations they support.</p>
6825
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6826<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6827<p>This intrinsic does a series of operations atomically. It first loads the
6828 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6829 result to <tt>ptr</tt>. It yields the original value stored
6830 at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6831
6832<h5>Examples:</h5>
6833<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6834%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6835%ptr = bitcast i8* %mallocP to i32*
6836 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6837%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6838 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6839%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6840 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6841%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6842 <i>; yields {i32}:result3 = 2</i>
6843%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6844</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6845
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6846</div>
6847
6848<!-- _______________________________________________________________________ -->
6849<div class="doc_subsubsection">
6850 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6851 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6852 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6853 <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]6854</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6855
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6856<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6857
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6858<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6859<p>These are overloaded intrinsics. You can
6860 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
6861 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
6862 bit width and for different address spaces. Not all targets support all bit
6863 widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6864
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6865<pre>
6866 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6867 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6868 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6869 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]6870</pre>
6871
6872<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6873 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6874 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6875 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6876 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]6877</pre>
6878
6879<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6880 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6881 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6882 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6883 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]6884</pre>
6885
6886<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6887 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6888 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6889 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6890 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]6891</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6892
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6893<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6894<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6895 the value stored in memory at <tt>ptr</tt>. It yields the original value
6896 at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6897
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6898<h5>Arguments:</h5>
6899<p>These intrinsics take two arguments, the first a pointer to an integer value
6900 and the second an integer value. The result is also an integer value. These
6901 integer types can have any bit width, but they must all have the same bit
6902 width. The targets may only lower integer representations they support.</p>
6903
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6904<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6905<p>These intrinsics does a series of operations atomically. They first load the
6906 value stored at <tt>ptr</tt>. They then do the bitwise
6907 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
6908 original value stored at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6909
6910<h5>Examples:</h5>
6911<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6912%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6913%ptr = bitcast i8* %mallocP to i32*
6914 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6915%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6916 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6917%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6918 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6919%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6920 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6921%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6922 <i>; yields {i32}:result3 = FF</i>
6923%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6924</pre>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6925
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6926</div>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6927
6928<!-- _______________________________________________________________________ -->
6929<div class="doc_subsubsection">
6930 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6931 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6932 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6933 <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]6934</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6935
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6936<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6937
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6938<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6939<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6940 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6941 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6942 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6943
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6944<pre>
6945 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6946 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6947 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6948 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]6949</pre>
6950
6951<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6952 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6953 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6954 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6955 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]6956</pre>
6957
6958<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6959 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6960 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6961 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6962 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]6963</pre>
6964
6965<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6966 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6967 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6968 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6969 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]6970</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6971
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6972<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6973<p>These intrinsics takes the signed or unsigned minimum or maximum of
6974 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6975 original value at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6976
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6977<h5>Arguments:</h5>
6978<p>These intrinsics take two arguments, the first a pointer to an integer value
6979 and the second an integer value. The result is also an integer value. These
6980 integer types can have any bit width, but they must all have the same bit
6981 width. The targets may only lower integer representations they support.</p>
6982
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6983<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6984<p>These intrinsics does a series of operations atomically. They first load the
6985 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
6986 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
6987 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6988
6989<h5>Examples:</h5>
6990<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6991%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6992%ptr = bitcast i8* %mallocP to i32*
6993 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6994%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6995 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6996%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6997 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6998%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6999 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]7000%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7001 <i>; yields {i32}:result3 = 8</i>
7002%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7003</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7004
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7005</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7006
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7007
7008<!-- ======================================================================= -->
7009<div class="doc_subsection">
7010 <a name="int_memorymarkers">Memory Use Markers</a>
7011</div>
7012
7013<div class="doc_text">
7014
7015<p>This class of intrinsics exists to information about the lifetime of memory
7016 objects and ranges where variables are immutable.</p>
7017
7018</div>
7019
7020<!-- _______________________________________________________________________ -->
7021<div class="doc_subsubsection">
7022 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7023</div>
7024
7025<div class="doc_text">
7026
7027<h5>Syntax:</h5>
7028<pre>
7029 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7030</pre>
7031
7032<h5>Overview:</h5>
7033<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7034 object's lifetime.</p>
7035
7036<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7037<p>The first argument is a constant integer representing the size of the
7038 object, or -1 if it is variable sized. The second argument is a pointer to
7039 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7040
7041<h5>Semantics:</h5>
7042<p>This intrinsic indicates that before this point in the code, the value of the
7043 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]7044 never be used and has an undefined value. A load from the pointer that
7045 precedes this intrinsic can be replaced with
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7046 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7047
7048</div>
7049
7050<!-- _______________________________________________________________________ -->
7051<div class="doc_subsubsection">
7052 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7053</div>
7054
7055<div class="doc_text">
7056
7057<h5>Syntax:</h5>
7058<pre>
7059 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7060</pre>
7061
7062<h5>Overview:</h5>
7063<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7064 object's lifetime.</p>
7065
7066<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7067<p>The first argument is a constant integer representing the size of the
7068 object, or -1 if it is variable sized. The second argument is a pointer to
7069 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7070
7071<h5>Semantics:</h5>
7072<p>This intrinsic indicates that after this point in the code, the value of the
7073 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7074 never be used and has an undefined value. Any stores into the memory object
7075 following this intrinsic may be removed as dead.
7076
7077</div>
7078
7079<!-- _______________________________________________________________________ -->
7080<div class="doc_subsubsection">
7081 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7082</div>
7083
7084<div class="doc_text">
7085
7086<h5>Syntax:</h5>
7087<pre>
7088 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;) readonly
7089</pre>
7090
7091<h5>Overview:</h5>
7092<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7093 a memory object will not change.</p>
7094
7095<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7096<p>The first argument is a constant integer representing the size of the
7097 object, or -1 if it is variable sized. The second argument is a pointer to
7098 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7099
7100<h5>Semantics:</h5>
7101<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7102 the return value, the referenced memory location is constant and
7103 unchanging.</p>
7104
7105</div>
7106
7107<!-- _______________________________________________________________________ -->
7108<div class="doc_subsubsection">
7109 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7110</div>
7111
7112<div class="doc_text">
7113
7114<h5>Syntax:</h5>
7115<pre>
7116 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7117</pre>
7118
7119<h5>Overview:</h5>
7120<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7121 a memory object are mutable.</p>
7122
7123<h5>Arguments:</h5>
7124<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7125 The second argument is a constant integer representing the size of the
7126 object, or -1 if it is variable sized and the third argument is a pointer
7127 to the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7128
7129<h5>Semantics:</h5>
7130<p>This intrinsic indicates that the memory is mutable again.</p>
7131
7132</div>
7133
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7134<!-- ======================================================================= -->
7135<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7136 <a name="int_general">General Intrinsics</a>
7137</div>
7138
7139<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7140
7141<p>This class of intrinsics is designed to be generic and has no specific
7142 purpose.</p>
7143
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7144</div>
7145
7146<!-- _______________________________________________________________________ -->
7147<div class="doc_subsubsection">
7148 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7149</div>
7150
7151<div class="doc_text">
7152
7153<h5>Syntax:</h5>
7154<pre>
7155 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7156</pre>
7157
7158<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7159<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7160
7161<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7162<p>The first argument is a pointer to a value, the second is a pointer to a
7163 global string, the third is a pointer to a global string which is the source
7164 file name, and the last argument is the line number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7165
7166<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7167<p>This intrinsic allows annotation of local variables with arbitrary strings.
7168 This can be useful for special purpose optimizations that want to look for
7169 these annotations. These have no other defined use, they are ignored by code
7170 generation and optimization.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7171
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7172</div>
7173
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7174<!-- _______________________________________________________________________ -->
7175<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]7176 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7177</div>
7178
7179<div class="doc_text">
7180
7181<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7182<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7183 any integer bit width.</p>
7184
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7185<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]7186 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7187 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7188 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7189 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7190 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]7191</pre>
7192
7193<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7194<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7195
7196<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7197<p>The first argument is an integer value (result of some expression), the
7198 second is a pointer to a global string, the third is a pointer to a global
7199 string which is the source file name, and the last argument is the line
7200 number. It returns the value of the first argument.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7201
7202<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7203<p>This intrinsic allows annotations to be put on arbitrary expressions with
7204 arbitrary strings. This can be useful for special purpose optimizations that
7205 want to look for these annotations. These have no other defined use, they
7206 are ignored by code generation and optimization.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7207
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7208</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7209
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7210<!-- _______________________________________________________________________ -->
7211<div class="doc_subsubsection">
7212 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7213</div>
7214
7215<div class="doc_text">
7216
7217<h5>Syntax:</h5>
7218<pre>
7219 declare void @llvm.trap()
7220</pre>
7221
7222<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7223<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7224
7225<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7226<p>None.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7227
7228<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7229<p>This intrinsics is lowered to the target dependent trap instruction. If the
7230 target does not have a trap instruction, this intrinsic will be lowered to
7231 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7232
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7233</div>
7234
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7235<!-- _______________________________________________________________________ -->
7236<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]7237 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7238</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7239
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7240<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7241
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7242<h5>Syntax:</h5>
7243<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7244 declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7245</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7246
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7247<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7248<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7249 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7250 ensure that it is placed on the stack before local variables.</p>
7251
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7252<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7253<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7254 arguments. The first argument is the value loaded from the stack
7255 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7256 that has enough space to hold the value of the guard.</p>
7257
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7258<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7259<p>This intrinsic causes the prologue/epilogue inserter to force the position of
7260 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7261 stack. This is to ensure that if a local variable on the stack is
7262 overwritten, it will destroy the value of the guard. When the function exits,
7263 the guard on the stack is checked against the original guard. If they're
7264 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7265 function.</p>
7266
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7267</div>
7268
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7269<!-- *********************************************************************** -->
7270<hr>
7271<address>
7272 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7273 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7274 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7275 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]7276
7277 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7278 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7279 Last modified: $Date$
7280</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]7281
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7282</body>
7283</html>