Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / e05d773333c76663af591beeba8a024b8879206b / . / docs / LangRef.html
blob: bc276ebe928b62271c618b57c9264ed37e081e1f [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]8 <meta name="description"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
12
13<body>
14
15<div class="doc_title"> LLVM Language Reference Manual </div>
16<ol>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]23 <li><a href="#linkage">Linkage Types</a>
24 <ol>
Bill Wendlinge2753242009-07-20 02:41:50 +0000[diff] [blame]25 <li><a href="#linkage_private">'<tt>private</tt>' Linkage</a></li>
26 <li><a href="#linkage_linker_private">'<tt>linker_private</tt>' Linkage</a></li>
27 <li><a href="#linkage_internal">'<tt>internal</tt>' Linkage</a></li>
28 <li><a href="#linkage_available_externally">'<tt>available_externally</tt>' Linkage</a></li>
29 <li><a href="#linkage_linkonce">'<tt>linkonce</tt>' Linkage</a></li>
30 <li><a href="#linkage_common">'<tt>common</tt>' Linkage</a></li>
31 <li><a href="#linkage_weak">'<tt>weak</tt>' Linkage</a></li>
32 <li><a href="#linkage_appending">'<tt>appending</tt>' Linkage</a></li>
33 <li><a href="#linkage_externweak">'<tt>extern_weak</tt>' Linkage</a></li>
Chris Lattner0fee5c22009-10-10 18:26:06 +0000[diff] [blame]34 <li><a href="#linkage_linkonce_odr">'<tt>linkonce_odr</tt>' Linkage</a></li>
Bill Wendlinge2753242009-07-20 02:41:50 +0000[diff] [blame]35 <li><a href="#linkage_weak">'<tt>weak_odr</tt>' Linkage</a></li>
36 <li><a href="#linkage_external">'<tt>externally visible</tt>' Linkage</a></li>
37 <li><a href="#linkage_dllimport">'<tt>dllimport</tt>' Linkage</a></li>
38 <li><a href="#linkage_dllexport">'<tt>dllexport</tt>' Linkage</a></li>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]39 </ol>
40 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]41 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]42 <li><a href="#namedtypes">Named Types</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]43 <li><a href="#globalvars">Global Variables</a></li>
44 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]45 <li><a href="#aliasstructure">Aliases</a></li>
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>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]57 <li><a href="#t_primitive">Primitive Types</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]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 Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]86 <li><a href="#blockaddress">Addresses of Basic Blocks</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 Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]114 <li><a href="#i_indirectbr">'<tt>indirectbr</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>
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]294 <li><a href="#int_objectsize">
295 '<tt>llvm.objectsize</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]296 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]297 </li>
298 </ol>
299 </li>
300</ol>
301
302<div class="doc_author">
303 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
304 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
305</div>
306
307<!-- *********************************************************************** -->
308<div class="doc_section"> <a name="abstract">Abstract </a></div>
309<!-- *********************************************************************** -->
310
311<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]312
313<p>This document is a reference manual for the LLVM assembly language. LLVM is
314 a Static Single Assignment (SSA) based representation that provides type
315 safety, low-level operations, flexibility, and the capability of representing
316 'all' high-level languages cleanly. It is the common code representation
317 used throughout all phases of the LLVM compilation strategy.</p>
318
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]319</div>
320
321<!-- *********************************************************************** -->
322<div class="doc_section"> <a name="introduction">Introduction</a> </div>
323<!-- *********************************************************************** -->
324
325<div class="doc_text">
326
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]327<p>The LLVM code representation is designed to be used in three different forms:
328 as an in-memory compiler IR, as an on-disk bitcode representation (suitable
329 for fast loading by a Just-In-Time compiler), and as a human readable
330 assembly language representation. This allows LLVM to provide a powerful
331 intermediate representation for efficient compiler transformations and
332 analysis, while providing a natural means to debug and visualize the
333 transformations. The three different forms of LLVM are all equivalent. This
334 document describes the human readable representation and notation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]335
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]336<p>The LLVM representation aims to be light-weight and low-level while being
337 expressive, typed, and extensible at the same time. It aims to be a
338 "universal IR" of sorts, by being at a low enough level that high-level ideas
339 may be cleanly mapped to it (similar to how microprocessors are "universal
340 IR's", allowing many source languages to be mapped to them). By providing
341 type information, LLVM can be used as the target of optimizations: for
342 example, through pointer analysis, it can be proven that a C automatic
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]343 variable is never accessed outside of the current function, allowing it to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]344 be promoted to a simple SSA value instead of a memory location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]345
346</div>
347
348<!-- _______________________________________________________________________ -->
349<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
350
351<div class="doc_text">
352
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]353<p>It is important to note that this document describes 'well formed' LLVM
354 assembly language. There is a difference between what the parser accepts and
355 what is considered 'well formed'. For example, the following instruction is
356 syntactically okay, but not well formed:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]357
358<div class="doc_code">
359<pre>
360%x = <a href="#i_add">add</a> i32 1, %x
361</pre>
362</div>
363
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]364<p>because the definition of <tt>%x</tt> does not dominate all of its uses. The
365 LLVM infrastructure provides a verification pass that may be used to verify
366 that an LLVM module is well formed. This pass is automatically run by the
367 parser after parsing input assembly and by the optimizer before it outputs
368 bitcode. The violations pointed out by the verifier pass indicate bugs in
369 transformation passes or input to the parser.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]370
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]371</div>
372
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]373<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]374
375<!-- *********************************************************************** -->
376<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
377<!-- *********************************************************************** -->
378
379<div class="doc_text">
380
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]381<p>LLVM identifiers come in two basic types: global and local. Global
382 identifiers (functions, global variables) begin with the <tt>'@'</tt>
383 character. Local identifiers (register names, types) begin with
384 the <tt>'%'</tt> character. Additionally, there are three different formats
385 for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]386
387<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]388 <li>Named values are represented as a string of characters with their prefix.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]389 For example, <tt>%foo</tt>, <tt>@DivisionByZero</tt>,
390 <tt>%a.really.long.identifier</tt>. The actual regular expression used is
391 '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'. Identifiers which require
392 other characters in their names can be surrounded with quotes. Special
393 characters may be escaped using <tt>"\xx"</tt> where <tt>xx</tt> is the
394 ASCII code for the character in hexadecimal. In this way, any character
395 can be used in a name value, even quotes themselves.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]396
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]397 <li>Unnamed values are represented as an unsigned numeric value with their
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]398 prefix. For example, <tt>%12</tt>, <tt>@2</tt>, <tt>%44</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]399
400 <li>Constants, which are described in a <a href="#constants">section about
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]401 constants</a>, below.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]402</ol>
403
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]404<p>LLVM requires that values start with a prefix for two reasons: Compilers
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]405 don't need to worry about name clashes with reserved words, and the set of
406 reserved words may be expanded in the future without penalty. Additionally,
407 unnamed identifiers allow a compiler to quickly come up with a temporary
408 variable without having to avoid symbol table conflicts.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]409
410<p>Reserved words in LLVM are very similar to reserved words in other
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]411 languages. There are keywords for different opcodes
412 ('<tt><a href="#i_add">add</a></tt>',
413 '<tt><a href="#i_bitcast">bitcast</a></tt>',
414 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names
415 ('<tt><a href="#t_void">void</a></tt>',
416 '<tt><a href="#t_primitive">i32</a></tt>', etc...), and others. These
417 reserved words cannot conflict with variable names, because none of them
418 start with a prefix character (<tt>'%'</tt> or <tt>'@'</tt>).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]419
420<p>Here is an example of LLVM code to multiply the integer variable
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]421 '<tt>%X</tt>' by 8:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]422
423<p>The easy way:</p>
424
425<div class="doc_code">
426<pre>
427%result = <a href="#i_mul">mul</a> i32 %X, 8
428</pre>
429</div>
430
431<p>After strength reduction:</p>
432
433<div class="doc_code">
434<pre>
435%result = <a href="#i_shl">shl</a> i32 %X, i8 3
436</pre>
437</div>
438
439<p>And the hard way:</p>
440
441<div class="doc_code">
442<pre>
Gabor Greifc0ea7672009-10-28 13:05:07 +0000[diff] [blame]443%0 = <a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
444%1 = <a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]445%result = <a href="#i_add">add</a> i32 %1, %1
446</pre>
447</div>
448
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]449<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several important
450 lexical features of LLVM:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]451
452<ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]453 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]454 line.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]455
456 <li>Unnamed temporaries are created when the result of a computation is not
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]457 assigned to a named value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]458
459 <li>Unnamed temporaries are numbered sequentially</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]460</ol>
461
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]462<p>It also shows a convention that we follow in this document. When
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]463 demonstrating instructions, we will follow an instruction with a comment that
464 defines the type and name of value produced. Comments are shown in italic
465 text.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]466
467</div>
468
469<!-- *********************************************************************** -->
470<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
471<!-- *********************************************************************** -->
472
473<!-- ======================================================================= -->
474<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
475</div>
476
477<div class="doc_text">
478
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]479<p>LLVM programs are composed of "Module"s, each of which is a translation unit
480 of the input programs. Each module consists of functions, global variables,
481 and symbol table entries. Modules may be combined together with the LLVM
482 linker, which merges function (and global variable) definitions, resolves
483 forward declarations, and merges symbol table entries. Here is an example of
484 the "hello world" module:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]485
486<div class="doc_code">
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]487<pre>
488<i>; Declare the string constant as a global constant.</i>
489<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]490
491<i>; External declaration of the puts function</i>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]492<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]493
494<i>; Definition of main function</i>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]495define i32 @main() { <i>; i32()* </i>
496 <i>; Convert [13 x i8]* to i8 *...</i>
497 %cast210 = <a href="#i_getelementptr">getelementptr</a> [13 x i8]* @.LC0, i64 0, i64 0 <i>; i8 *</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]498
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]499 <i>; Call puts function to write out the string to stdout.</i>
500 <a href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
501 <a href="#i_ret">ret</a> i32 0<br>}<br>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]502</pre>
503</div>
504
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]505<p>This example is made up of a <a href="#globalvars">global variable</a> named
506 "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>" function, and
507 a <a href="#functionstructure">function definition</a> for
508 "<tt>main</tt>".</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]509
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]510<p>In general, a module is made up of a list of global values, where both
511 functions and global variables are global values. Global values are
512 represented by a pointer to a memory location (in this case, a pointer to an
513 array of char, and a pointer to a function), and have one of the
514 following <a href="#linkage">linkage types</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]515
516</div>
517
518<!-- ======================================================================= -->
519<div class="doc_subsection">
520 <a name="linkage">Linkage Types</a>
521</div>
522
523<div class="doc_text">
524
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]525<p>All Global Variables and Functions have one of the following types of
526 linkage:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]527
528<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]529 <dt><tt><b><a name="linkage_private">private</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]530 <dd>Global values with private linkage are only directly accessible by objects
531 in the current module. In particular, linking code into a module with an
532 private global value may cause the private to be renamed as necessary to
533 avoid collisions. Because the symbol is private to the module, all
534 references can be updated. This doesn't show up in any symbol table in the
535 object file.</dd>
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]536
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]537 <dt><tt><b><a name="linkage_linker_private">linker_private</a></b></tt></dt>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]538 <dd>Similar to private, but the symbol is passed through the assembler and
Chris Lattner06c1ecc2009-08-24 04:32:16 +0000[diff] [blame]539 removed by the linker after evaluation. Note that (unlike private
540 symbols) linker_private symbols are subject to coalescing by the linker:
541 weak symbols get merged and redefinitions are rejected. However, unlike
542 normal strong symbols, they are removed by the linker from the final
543 linked image (executable or dynamic library).</dd>
Bill Wendling41a07852009-07-20 01:03:30 +0000[diff] [blame]544
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]545 <dt><tt><b><a name="linkage_internal">internal</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]546 <dd>Similar to private, but the value shows as a local symbol
547 (<tt>STB_LOCAL</tt> in the case of ELF) in the object file. This
548 corresponds to the notion of the '<tt>static</tt>' keyword in C.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]549
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]550 <dt><tt><b><a name="linkage_available_externally">available_externally</a></b></tt></dt>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]551 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]552 into the object file corresponding to the LLVM module. They exist to
553 allow inlining and other optimizations to take place given knowledge of
554 the definition of the global, which is known to be somewhere outside the
555 module. Globals with <tt>available_externally</tt> linkage are allowed to
556 be discarded at will, and are otherwise the same as <tt>linkonce_odr</tt>.
557 This linkage type is only allowed on definitions, not declarations.</dd>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]558
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]559 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]560 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
Chris Lattnerd16c5512010-01-09 19:15:14 +0000[diff] [blame]561 the same name when linkage occurs. This can be used to implement
562 some forms of inline functions, templates, or other code which must be
563 generated in each translation unit that uses it, but where the body may
564 be overridden with a more definitive definition later. Unreferenced
565 <tt>linkonce</tt> globals are allowed to be discarded. Note that
566 <tt>linkonce</tt> linkage does not actually allow the optimizer to
567 inline the body of this function into callers because it doesn't know if
568 this definition of the function is the definitive definition within the
569 program or whether it will be overridden by a stronger definition.
570 To enable inlining and other optimizations, use "<tt>linkonce_odr</tt>"
571 linkage.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]572
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]573 <dt><tt><b><a name="linkage_weak">weak</a></b></tt></dt>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]574 <dd>"<tt>weak</tt>" linkage has the same merging semantics as
575 <tt>linkonce</tt> linkage, except that unreferenced globals with
576 <tt>weak</tt> linkage may not be discarded. This is used for globals that
577 are declared "weak" in C source code.</dd>
578
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]579 <dt><tt><b><a name="linkage_common">common</a></b></tt></dt>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]580 <dd>"<tt>common</tt>" linkage is most similar to "<tt>weak</tt>" linkage, but
581 they are used for tentative definitions in C, such as "<tt>int X;</tt>" at
582 global scope.
583 Symbols with "<tt>common</tt>" linkage are merged in the same way as
584 <tt>weak symbols</tt>, and they may not be deleted if unreferenced.
Chris Lattner83c4a712009-08-05 05:41:44 +0000[diff] [blame]585 <tt>common</tt> symbols may not have an explicit section,
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]586 must have a zero initializer, and may not be marked '<a
Chris Lattner83c4a712009-08-05 05:41:44 +0000[diff] [blame]587 href="#globalvars"><tt>constant</tt></a>'. Functions and aliases may not
588 have common linkage.</dd>
Chris Lattner7215c7f2009-08-05 05:21:07 +0000[diff] [blame]589
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]590
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]591 <dt><tt><b><a name="linkage_appending">appending</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]592 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]593 pointer to array type. When two global variables with appending linkage
594 are linked together, the two global arrays are appended together. This is
595 the LLVM, typesafe, equivalent of having the system linker append together
596 "sections" with identical names when .o files are linked.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]597
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]598 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]599 <dd>The semantics of this linkage follow the ELF object file model: the symbol
600 is weak until linked, if not linked, the symbol becomes null instead of
601 being an undefined reference.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]602
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]603 <dt><tt><b><a name="linkage_linkonce_odr">linkonce_odr</a></b></tt></dt>
604 <dt><tt><b><a name="linkage_weak_odr">weak_odr</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]605 <dd>Some languages allow differing globals to be merged, such as two functions
606 with different semantics. Other languages, such as <tt>C++</tt>, ensure
607 that only equivalent globals are ever merged (the "one definition rule" -
608 "ODR"). Such languages can use the <tt>linkonce_odr</tt>
609 and <tt>weak_odr</tt> linkage types to indicate that the global will only
610 be merged with equivalent globals. These linkage types are otherwise the
611 same as their non-<tt>odr</tt> versions.</dd>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]612
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]613 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]614 <dd>If none of the above identifiers are used, the global is externally
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]615 visible, meaning that it participates in linkage and can be used to
616 resolve external symbol references.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]617</dl>
618
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]619<p>The next two types of linkage are targeted for Microsoft Windows platform
620 only. They are designed to support importing (exporting) symbols from (to)
621 DLLs (Dynamic Link Libraries).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]622
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]623<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]624 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]625 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]626 or variable via a global pointer to a pointer that is set up by the DLL
627 exporting the symbol. On Microsoft Windows targets, the pointer name is
628 formed by combining <code>__imp_</code> and the function or variable
629 name.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]630
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]631 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]632 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]633 pointer to a pointer in a DLL, so that it can be referenced with the
634 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
635 name is formed by combining <code>__imp_</code> and the function or
636 variable name.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]637</dl>
638
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]639<p>For example, since the "<tt>.LC0</tt>" variable is defined to be internal, if
640 another module defined a "<tt>.LC0</tt>" variable and was linked with this
641 one, one of the two would be renamed, preventing a collision. Since
642 "<tt>main</tt>" and "<tt>puts</tt>" are external (i.e., lacking any linkage
643 declarations), they are accessible outside of the current module.</p>
644
645<p>It is illegal for a function <i>declaration</i> to have any linkage type
646 other than "externally visible", <tt>dllimport</tt>
647 or <tt>extern_weak</tt>.</p>
648
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]649<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]650 or <tt>weak_odr</tt> linkages.</p>
651
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]652</div>
653
654<!-- ======================================================================= -->
655<div class="doc_subsection">
656 <a name="callingconv">Calling Conventions</a>
657</div>
658
659<div class="doc_text">
660
661<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]662 and <a href="#i_invoke">invokes</a> can all have an optional calling
663 convention specified for the call. The calling convention of any pair of
664 dynamic caller/callee must match, or the behavior of the program is
665 undefined. The following calling conventions are supported by LLVM, and more
666 may be added in the future:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]667
668<dl>
669 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]670 <dd>This calling convention (the default if no other calling convention is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]671 specified) matches the target C calling conventions. This calling
672 convention supports varargs function calls and tolerates some mismatch in
673 the declared prototype and implemented declaration of the function (as
674 does normal C).</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]675
676 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]677 <dd>This calling convention attempts to make calls as fast as possible
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]678 (e.g. by passing things in registers). This calling convention allows the
679 target to use whatever tricks it wants to produce fast code for the
680 target, without having to conform to an externally specified ABI
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]681 (Application Binary Interface).
682 <a href="CodeGenerator.html#tailcallopt">Tail calls can only be optimized
683 when this convention is used.</a> This calling convention does not
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]684 support varargs and requires the prototype of all callees to exactly match
685 the prototype of the function definition.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]686
687 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]688 <dd>This calling convention attempts to make code in the caller as efficient
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]689 as possible under the assumption that the call is not commonly executed.
690 As such, these calls often preserve all registers so that the call does
691 not break any live ranges in the caller side. This calling convention
692 does not support varargs and requires the prototype of all callees to
693 exactly match the prototype of the function definition.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]694
695 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]696 <dd>Any calling convention may be specified by number, allowing
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]697 target-specific calling conventions to be used. Target specific calling
698 conventions start at 64.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]699</dl>
700
701<p>More calling conventions can be added/defined on an as-needed basis, to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]702 support Pascal conventions or any other well-known target-independent
703 convention.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]704
705</div>
706
707<!-- ======================================================================= -->
708<div class="doc_subsection">
709 <a name="visibility">Visibility Styles</a>
710</div>
711
712<div class="doc_text">
713
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]714<p>All Global Variables and Functions have one of the following visibility
715 styles:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]716
717<dl>
718 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]719 <dd>On targets that use the ELF object file format, default visibility means
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]720 that the declaration is visible to other modules and, in shared libraries,
721 means that the declared entity may be overridden. On Darwin, default
722 visibility means that the declaration is visible to other modules. Default
723 visibility corresponds to "external linkage" in the language.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]724
725 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]726 <dd>Two declarations of an object with hidden visibility refer to the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]727 object if they are in the same shared object. Usually, hidden visibility
728 indicates that the symbol will not be placed into the dynamic symbol
729 table, so no other module (executable or shared library) can reference it
730 directly.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]731
732 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]733 <dd>On ELF, protected visibility indicates that the symbol will be placed in
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]734 the dynamic symbol table, but that references within the defining module
735 will bind to the local symbol. That is, the symbol cannot be overridden by
736 another module.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]737</dl>
738
739</div>
740
741<!-- ======================================================================= -->
742<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]743 <a name="namedtypes">Named Types</a>
744</div>
745
746<div class="doc_text">
747
748<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]749 it easier to read the IR and make the IR more condensed (particularly when
750 recursive types are involved). An example of a name specification is:</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]751
752<div class="doc_code">
753<pre>
754%mytype = type { %mytype*, i32 }
755</pre>
756</div>
757
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]758<p>You may give a name to any <a href="#typesystem">type</a> except
759 "<a href="t_void">void</a>". Type name aliases may be used anywhere a type
760 is expected with the syntax "%mytype".</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]761
762<p>Note that type names are aliases for the structural type that they indicate,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]763 and that you can therefore specify multiple names for the same type. This
764 often leads to confusing behavior when dumping out a .ll file. Since LLVM IR
765 uses structural typing, the name is not part of the type. When printing out
766 LLVM IR, the printer will pick <em>one name</em> to render all types of a
767 particular shape. This means that if you have code where two different
768 source types end up having the same LLVM type, that the dumper will sometimes
769 print the "wrong" or unexpected type. This is an important design point and
770 isn't going to change.</p>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]771
772</div>
773
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]774<!-- ======================================================================= -->
775<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]776 <a name="globalvars">Global Variables</a>
777</div>
778
779<div class="doc_text">
780
781<p>Global variables define regions of memory allocated at compilation time
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]782 instead of run-time. Global variables may optionally be initialized, may
783 have an explicit section to be placed in, and may have an optional explicit
784 alignment specified. A variable may be defined as "thread_local", which
785 means that it will not be shared by threads (each thread will have a
786 separated copy of the variable). A variable may be defined as a global
787 "constant," which indicates that the contents of the variable
788 will <b>never</b> be modified (enabling better optimization, allowing the
789 global data to be placed in the read-only section of an executable, etc).
790 Note that variables that need runtime initialization cannot be marked
791 "constant" as there is a store to the variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]792
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]793<p>LLVM explicitly allows <em>declarations</em> of global variables to be marked
794 constant, even if the final definition of the global is not. This capability
795 can be used to enable slightly better optimization of the program, but
796 requires the language definition to guarantee that optimizations based on the
797 'constantness' are valid for the translation units that do not include the
798 definition.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]799
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]800<p>As SSA values, global variables define pointer values that are in scope
801 (i.e. they dominate) all basic blocks in the program. Global variables
802 always define a pointer to their "content" type because they describe a
803 region of memory, and all memory objects in LLVM are accessed through
804 pointers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]805
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]806<p>A global variable may be declared to reside in a target-specific numbered
807 address space. For targets that support them, address spaces may affect how
808 optimizations are performed and/or what target instructions are used to
809 access the variable. The default address space is zero. The address space
810 qualifier must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]811
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]812<p>LLVM allows an explicit section to be specified for globals. If the target
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]813 supports it, it will emit globals to the section specified.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]814
815<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]816 the alignment is set to zero, the alignment of the global is set by the
817 target to whatever it feels convenient. If an explicit alignment is
818 specified, the global is forced to have at least that much alignment. All
819 alignments must be a power of 2.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]820
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]821<p>For example, the following defines a global in a numbered address space with
822 an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]823
824<div class="doc_code">
825<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]826@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]827</pre>
828</div>
829
830</div>
831
832
833<!-- ======================================================================= -->
834<div class="doc_subsection">
835 <a name="functionstructure">Functions</a>
836</div>
837
838<div class="doc_text">
839
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]840<p>LLVM function definitions consist of the "<tt>define</tt>" keyord, an
841 optional <a href="#linkage">linkage type</a>, an optional
842 <a href="#visibility">visibility style</a>, an optional
843 <a href="#callingconv">calling convention</a>, a return type, an optional
844 <a href="#paramattrs">parameter attribute</a> for the return type, a function
845 name, a (possibly empty) argument list (each with optional
846 <a href="#paramattrs">parameter attributes</a>), optional
847 <a href="#fnattrs">function attributes</a>, an optional section, an optional
848 alignment, an optional <a href="#gc">garbage collector name</a>, an opening
849 curly brace, a list of basic blocks, and a closing curly brace.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]850
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]851<p>LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
852 optional <a href="#linkage">linkage type</a>, an optional
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]853 <a href="#visibility">visibility style</a>, an optional
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]854 <a href="#callingconv">calling convention</a>, a return type, an optional
855 <a href="#paramattrs">parameter attribute</a> for the return type, a function
856 name, a possibly empty list of arguments, an optional alignment, and an
857 optional <a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]858
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]859<p>A function definition contains a list of basic blocks, forming the CFG
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]860 (Control Flow Graph) for the function. Each basic block may optionally start
861 with a label (giving the basic block a symbol table entry), contains a list
862 of instructions, and ends with a <a href="#terminators">terminator</a>
863 instruction (such as a branch or function return).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]864
865<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]866 executed on entrance to the function, and it is not allowed to have
867 predecessor basic blocks (i.e. there can not be any branches to the entry
868 block of a function). Because the block can have no predecessors, it also
869 cannot have any <a href="#i_phi">PHI nodes</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]870
871<p>LLVM allows an explicit section to be specified for functions. If the target
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]872 supports it, it will emit functions to the section specified.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]873
874<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]875 the alignment is set to zero, the alignment of the function is set by the
876 target to whatever it feels convenient. If an explicit alignment is
877 specified, the function is forced to have at least that much alignment. All
878 alignments must be a power of 2.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]879
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]880<h5>Syntax:</h5>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]881<div class="doc_code">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]882<pre>
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]883define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]884 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
885 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
886 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
887 [<a href="#gc">gc</a>] { ... }
888</pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]889</div>
890
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]891</div>
892
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]893<!-- ======================================================================= -->
894<div class="doc_subsection">
895 <a name="aliasstructure">Aliases</a>
896</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]897
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]898<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]899
900<p>Aliases act as "second name" for the aliasee value (which can be either
901 function, global variable, another alias or bitcast of global value). Aliases
902 may have an optional <a href="#linkage">linkage type</a>, and an
903 optional <a href="#visibility">visibility style</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]904
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]905<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]906<div class="doc_code">
907<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]908@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]909</pre>
910</div>
911
912</div>
913
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]914<!-- ======================================================================= -->
915<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]916
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]917<div class="doc_text">
918
919<p>The return type and each parameter of a function type may have a set of
920 <i>parameter attributes</i> associated with them. Parameter attributes are
921 used to communicate additional information about the result or parameters of
922 a function. Parameter attributes are considered to be part of the function,
923 not of the function type, so functions with different parameter attributes
924 can have the same function type.</p>
925
926<p>Parameter attributes are simple keywords that follow the type specified. If
927 multiple parameter attributes are needed, they are space separated. For
928 example:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]929
930<div class="doc_code">
931<pre>
Nick Lewycky3022a742009-02-15 23:06:14 +0000[diff] [blame]932declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]933declare i32 @atoi(i8 zeroext)
934declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]935</pre>
936</div>
937
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]938<p>Note that any attributes for the function result (<tt>nounwind</tt>,
939 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]940
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]941<p>Currently, only the following parameter attributes are defined:</p>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]942
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]943<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]944 <dt><tt><b>zeroext</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]945 <dd>This indicates to the code generator that the parameter or return value
946 should be zero-extended to a 32-bit value by the caller (for a parameter)
947 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]948
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]949 <dt><tt><b>signext</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]950 <dd>This indicates to the code generator that the parameter or return value
951 should be sign-extended to a 32-bit value by the caller (for a parameter)
952 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]953
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]954 <dt><tt><b>inreg</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]955 <dd>This indicates that this parameter or return value should be treated in a
956 special target-dependent fashion during while emitting code for a function
957 call or return (usually, by putting it in a register as opposed to memory,
958 though some targets use it to distinguish between two different kinds of
959 registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]960
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]961 <dt><tt><b><a name="byval">byval</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]962 <dd>This indicates that the pointer parameter should really be passed by value
963 to the function. The attribute implies that a hidden copy of the pointee
964 is made between the caller and the callee, so the callee is unable to
965 modify the value in the callee. This attribute is only valid on LLVM
966 pointer arguments. It is generally used to pass structs and arrays by
967 value, but is also valid on pointers to scalars. The copy is considered
968 to belong to the caller not the callee (for example,
969 <tt><a href="#readonly">readonly</a></tt> functions should not write to
970 <tt>byval</tt> parameters). This is not a valid attribute for return
971 values. The byval attribute also supports specifying an alignment with
972 the align attribute. This has a target-specific effect on the code
973 generator that usually indicates a desired alignment for the synthesized
974 stack slot.</dd>
975
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]976 <dt><tt><b>sret</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]977 <dd>This indicates that the pointer parameter specifies the address of a
978 structure that is the return value of the function in the source program.
979 This pointer must be guaranteed by the caller to be valid: loads and
980 stores to the structure may be assumed by the callee to not to trap. This
981 may only be applied to the first parameter. This is not a valid attribute
982 for return values. </dd>
983
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]984 <dt><tt><b>noalias</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]985 <dd>This indicates that the pointer does not alias any global or any other
986 parameter. The caller is responsible for ensuring that this is the
987 case. On a function return value, <tt>noalias</tt> additionally indicates
988 that the pointer does not alias any other pointers visible to the
989 caller. For further details, please see the discussion of the NoAlias
990 response in
991 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
992 analysis</a>.</dd>
993
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]994 <dt><tt><b>nocapture</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]995 <dd>This indicates that the callee does not make any copies of the pointer
996 that outlive the callee itself. This is not a valid attribute for return
997 values.</dd>
998
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]999 <dt><tt><b>nest</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1000 <dd>This indicates that the pointer parameter can be excised using the
1001 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
1002 attribute for return values.</dd>
1003</dl>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1004
1005</div>
1006
1007<!-- ======================================================================= -->
1008<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1009 <a name="gc">Garbage Collector Names</a>
1010</div>
1011
1012<div class="doc_text">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1013
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1014<p>Each function may specify a garbage collector name, which is simply a
1015 string:</p>
1016
1017<div class="doc_code">
1018<pre>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1019define void @f() gc "name" { ... }
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1020</pre>
1021</div>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1022
1023<p>The compiler declares the supported values of <i>name</i>. Specifying a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1024 collector which will cause the compiler to alter its output in order to
1025 support the named garbage collection algorithm.</p>
1026
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]1027</div>
1028
1029<!-- ======================================================================= -->
1030<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1031 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1032</div>
1033
1034<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1035
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1036<p>Function attributes are set to communicate additional information about a
1037 function. Function attributes are considered to be part of the function, not
1038 of the function type, so functions with different parameter attributes can
1039 have the same function type.</p>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1040
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1041<p>Function attributes are simple keywords that follow the type specified. If
1042 multiple attributes are needed, they are space separated. For example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1043
1044<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1045<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1046define void @f() noinline { ... }
1047define void @f() alwaysinline { ... }
1048define void @f() alwaysinline optsize { ... }
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1049define void @f() optsize { ... }
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1050</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1051</div>
1052
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1053<dl>
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1054 <dt><tt><b>alwaysinline</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1055 <dd>This attribute indicates that the inliner should attempt to inline this
1056 function into callers whenever possible, ignoring any active inlining size
1057 threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1058
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1059 <dt><tt><b>inlinehint</b></tt></dt>
Dale Johannesen7fd6b0f2009-08-26 01:08:21 +0000[diff] [blame]1060 <dd>This attribute indicates that the source code contained a hint that inlining
1061 this function is desirable (such as the "inline" keyword in C/C++). It
1062 is just a hint; it imposes no requirements on the inliner.</dd>
1063
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1064 <dt><tt><b>noinline</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1065 <dd>This attribute indicates that the inliner should never inline this
1066 function in any situation. This attribute may not be used together with
1067 the <tt>alwaysinline</tt> attribute.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1068
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1069 <dt><tt><b>optsize</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1070 <dd>This attribute suggests that optimization passes and code generator passes
1071 make choices that keep the code size of this function low, and otherwise
1072 do optimizations specifically to reduce code size.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1073
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1074 <dt><tt><b>noreturn</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1075 <dd>This function attribute indicates that the function never returns
1076 normally. This produces undefined behavior at runtime if the function
1077 ever does dynamically return.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1078
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1079 <dt><tt><b>nounwind</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1080 <dd>This function attribute indicates that the function never returns with an
1081 unwind or exceptional control flow. If the function does unwind, its
1082 runtime behavior is undefined.</dd>
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1083
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1084 <dt><tt><b>readnone</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1085 <dd>This attribute indicates that the function computes its result (or decides
1086 to unwind an exception) based strictly on its arguments, without
1087 dereferencing any pointer arguments or otherwise accessing any mutable
1088 state (e.g. memory, control registers, etc) visible to caller functions.
1089 It does not write through any pointer arguments
1090 (including <tt><a href="#byval">byval</a></tt> arguments) and never
1091 changes any state visible to callers. This means that it cannot unwind
1092 exceptions by calling the <tt>C++</tt> exception throwing methods, but
1093 could use the <tt>unwind</tt> instruction.</dd>
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1094
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1095 <dt><tt><b><a name="readonly">readonly</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1096 <dd>This attribute indicates that the function does not write through any
1097 pointer arguments (including <tt><a href="#byval">byval</a></tt>
1098 arguments) or otherwise modify any state (e.g. memory, control registers,
1099 etc) visible to caller functions. It may dereference pointer arguments
1100 and read state that may be set in the caller. A readonly function always
1101 returns the same value (or unwinds an exception identically) when called
1102 with the same set of arguments and global state. It cannot unwind an
1103 exception by calling the <tt>C++</tt> exception throwing methods, but may
1104 use the <tt>unwind</tt> instruction.</dd>
Anton Korobeynikovedd7d112009-07-17 18:07:26 +0000[diff] [blame]1105
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1106 <dt><tt><b><a name="ssp">ssp</a></b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1107 <dd>This attribute indicates that the function should emit a stack smashing
1108 protector. It is in the form of a "canary"&mdash;a random value placed on
1109 the stack before the local variables that's checked upon return from the
1110 function to see if it has been overwritten. A heuristic is used to
1111 determine if a function needs stack protectors or not.<br>
1112<br>
1113 If a function that has an <tt>ssp</tt> attribute is inlined into a
1114 function that doesn't have an <tt>ssp</tt> attribute, then the resulting
1115 function will have an <tt>ssp</tt> attribute.</dd>
1116
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1117 <dt><tt><b>sspreq</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1118 <dd>This attribute indicates that the function should <em>always</em> emit a
1119 stack smashing protector. This overrides
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1120 the <tt><a href="#ssp">ssp</a></tt> function attribute.<br>
1121<br>
1122 If a function that has an <tt>sspreq</tt> attribute is inlined into a
1123 function that doesn't have an <tt>sspreq</tt> attribute or which has
1124 an <tt>ssp</tt> attribute, then the resulting function will have
1125 an <tt>sspreq</tt> attribute.</dd>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1126
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1127 <dt><tt><b>noredzone</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1128 <dd>This attribute indicates that the code generator should not use a red
1129 zone, even if the target-specific ABI normally permits it.</dd>
1130
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1131 <dt><tt><b>noimplicitfloat</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1132 <dd>This attributes disables implicit floating point instructions.</dd>
1133
Bill Wendling614b32b2009-11-02 00:24:16 +0000[diff] [blame]1134 <dt><tt><b>naked</b></tt></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1135 <dd>This attribute disables prologue / epilogue emission for the function.
1136 This can have very system-specific consequences.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1137</dl>
1138
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1139</div>
1140
1141<!-- ======================================================================= -->
1142<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1143 <a name="moduleasm">Module-Level Inline Assembly</a>
1144</div>
1145
1146<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1147
1148<p>Modules may contain "module-level inline asm" blocks, which corresponds to
1149 the GCC "file scope inline asm" blocks. These blocks are internally
1150 concatenated by LLVM and treated as a single unit, but may be separated in
1151 the <tt>.ll</tt> file if desired. The syntax is very simple:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1152
1153<div class="doc_code">
1154<pre>
1155module asm "inline asm code goes here"
1156module asm "more can go here"
1157</pre>
1158</div>
1159
1160<p>The strings can contain any character by escaping non-printable characters.
1161 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]1162 for the number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1163
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1164<p>The inline asm code is simply printed to the machine code .s file when
1165 assembly code is generated.</p>
1166
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1167</div>
1168
1169<!-- ======================================================================= -->
1170<div class="doc_subsection">
1171 <a name="datalayout">Data Layout</a>
1172</div>
1173
1174<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1175
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1176<p>A module may specify a target specific data layout string that specifies how
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1177 data is to be laid out in memory. The syntax for the data layout is
1178 simply:</p>
1179
1180<div class="doc_code">
1181<pre>
1182target datalayout = "<i>layout specification</i>"
1183</pre>
1184</div>
1185
1186<p>The <i>layout specification</i> consists of a list of specifications
1187 separated by the minus sign character ('-'). Each specification starts with
1188 a letter and may include other information after the letter to define some
1189 aspect of the data layout. The specifications accepted are as follows:</p>
1190
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1191<dl>
1192 <dt><tt>E</tt></dt>
1193 <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]1194 bits with the most significance have the lowest address location.</dd>
1195
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1196 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1197 <dd>Specifies that the target lays out data in little-endian form. That is,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1198 the bits with the least significance have the lowest address
1199 location.</dd>
1200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1201 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1202 <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]1203 <i>preferred</i> alignments. All sizes are in bits. Specifying
1204 the <i>pref</i> alignment is optional. If omitted, the
1205 preceding <tt>:</tt> should be omitted too.</dd>
1206
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1207 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1208 <dd>This specifies the alignment for an integer type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1209 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1210
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1211 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1212 <dd>This specifies the alignment for a vector type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1213 <i>size</i>.</dd>
1214
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1215 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1216 <dd>This specifies the alignment for a floating point type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1217 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1218 (double).</dd>
1219
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1220 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1221 <dd>This specifies the alignment for an aggregate type of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1222 <i>size</i>.</dd>
1223
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1224 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1225 <dd>This specifies the alignment for a stack object of a given bit
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1226 <i>size</i>.</dd>
Chris Lattnerece41802009-11-07 09:35:34 +0000[diff] [blame]1227
1228 <dt><tt>n<i>size1</i>:<i>size2</i>:<i>size3</i>...</tt></dt>
1229 <dd>This specifies a set of native integer widths for the target CPU
1230 in bits. For example, it might contain "n32" for 32-bit PowerPC,
1231 "n32:64" for PowerPC 64, or "n8:16:32:64" for X86-64. Elements of
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1232 this set are considered to support most general arithmetic
Chris Lattnerece41802009-11-07 09:35:34 +0000[diff] [blame]1233 operations efficiently.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1234</dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1235
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1236<p>When constructing the data layout for a given target, LLVM starts with a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1237 default set of specifications which are then (possibly) overriden by the
1238 specifications in the <tt>datalayout</tt> keyword. The default specifications
1239 are given in this list:</p>
1240
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1241<ul>
1242 <li><tt>E</tt> - big endian</li>
1243 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1244 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1245 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1246 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1247 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1248 <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]1249 alignment of 64-bits</li>
1250 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1251 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1252 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1253 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1254 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1255 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1256</ul>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1257
1258<p>When LLVM is determining the alignment for a given type, it uses the
1259 following rules:</p>
1260
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1261<ol>
1262 <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]1263 specification is used.</li>
1264
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1265 <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]1266 smallest integer type that is larger than the bitwidth of the sought type
1267 is used. If none of the specifications are larger than the bitwidth then
1268 the the largest integer type is used. For example, given the default
1269 specifications above, the i7 type will use the alignment of i8 (next
1270 largest) while both i65 and i256 will use the alignment of i64 (largest
1271 specified).</li>
1272
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1273 <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]1274 largest vector type that is smaller than the sought vector type will be
1275 used as a fall back. This happens because &lt;128 x double&gt; can be
1276 implemented in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1277</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1278
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1279</div>
1280
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1281<!-- ======================================================================= -->
1282<div class="doc_subsection">
1283 <a name="pointeraliasing">Pointer Aliasing Rules</a>
1284</div>
1285
1286<div class="doc_text">
1287
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1288<p>Any memory access must be done through a pointer value associated
Andreas Bolka23bece42009-07-27 20:37:10 +0000[diff] [blame]1289with an address range of the memory access, otherwise the behavior
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1290is undefined. Pointer values are associated with address ranges
1291according to the following rules:</p>
1292
1293<ul>
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1294 <li>A pointer value formed from a
1295 <tt><a href="#i_getelementptr">getelementptr</a></tt> instruction
1296 is associated with the addresses associated with the first operand
1297 of the <tt>getelementptr</tt>.</li>
1298 <li>An address of a global variable is associated with the address
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1299 range of the variable's storage.</li>
1300 <li>The result value of an allocation instruction is associated with
1301 the address range of the allocated storage.</li>
1302 <li>A null pointer in the default address-space is associated with
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1303 no address.</li>
1304 <li>A pointer value formed by an
1305 <tt><a href="#i_inttoptr">inttoptr</a></tt> is associated with all
1306 address ranges of all pointer values that contribute (directly or
1307 indirectly) to the computation of the pointer's value.</li>
1308 <li>The result value of a
1309 <tt><a href="#i_bitcast">bitcast</a></tt> is associated with all
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1310 addresses associated with the operand of the <tt>bitcast</tt>.</li>
1311 <li>An integer constant other than zero or a pointer value returned
1312 from a function not defined within LLVM may be associated with address
1313 ranges allocated through mechanisms other than those provided by
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1314 LLVM. Such ranges shall not overlap with any ranges of addresses
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1315 allocated by mechanisms provided by LLVM.</li>
1316 </ul>
1317
1318<p>LLVM IR does not associate types with memory. The result type of a
Andreas Bolka11fbf432009-07-29 00:02:05 +0000[diff] [blame]1319<tt><a href="#i_load">load</a></tt> merely indicates the size and
1320alignment of the memory from which to load, as well as the
1321interpretation of the value. The first operand of a
1322<tt><a href="#i_store">store</a></tt> similarly only indicates the size
1323and alignment of the store.</p>
Dan Gohman27b47012009-07-27 18:07:55 +0000[diff] [blame]1324
1325<p>Consequently, type-based alias analysis, aka TBAA, aka
1326<tt>-fstrict-aliasing</tt>, is not applicable to general unadorned
1327LLVM IR. <a href="#metadata">Metadata</a> may be used to encode
1328additional information which specialized optimization passes may use
1329to implement type-based alias analysis.</p>
1330
1331</div>
1332
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1333<!-- *********************************************************************** -->
1334<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1335<!-- *********************************************************************** -->
1336
1337<div class="doc_text">
1338
1339<p>The LLVM type system is one of the most important features of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1340 intermediate representation. Being typed enables a number of optimizations
1341 to be performed on the intermediate representation directly, without having
1342 to do extra analyses on the side before the transformation. A strong type
1343 system makes it easier to read the generated code and enables novel analyses
1344 and transformations that are not feasible to perform on normal three address
1345 code representations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1346
1347</div>
1348
1349<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1350<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1351Classifications</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1352
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1353<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1354
1355<p>The types fall into a few useful classifications:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1356
1357<table border="1" cellspacing="0" cellpadding="4">
1358 <tbody>
1359 <tr><th>Classification</th><th>Types</th></tr>
1360 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1361 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1362 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1363 </tr>
1364 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1365 <td><a href="#t_floating">floating point</a></td>
1366 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1367 </tr>
1368 <tr>
1369 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1370 <td><a href="#t_integer">integer</a>,
1371 <a href="#t_floating">floating point</a>,
1372 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1373 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1374 <a href="#t_struct">structure</a>,
1375 <a href="#t_array">array</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1376 <a href="#t_label">label</a>,
1377 <a href="#t_metadata">metadata</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1378 </td>
1379 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1380 <tr>
1381 <td><a href="#t_primitive">primitive</a></td>
1382 <td><a href="#t_label">label</a>,
1383 <a href="#t_void">void</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1384 <a href="#t_floating">floating point</a>,
1385 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1386 </tr>
1387 <tr>
1388 <td><a href="#t_derived">derived</a></td>
1389 <td><a href="#t_integer">integer</a>,
1390 <a href="#t_array">array</a>,
1391 <a href="#t_function">function</a>,
1392 <a href="#t_pointer">pointer</a>,
1393 <a href="#t_struct">structure</a>,
1394 <a href="#t_pstruct">packed structure</a>,
1395 <a href="#t_vector">vector</a>,
1396 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1397 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1398 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1399 </tbody>
1400</table>
1401
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1402<p>The <a href="#t_firstclass">first class</a> types are perhaps the most
1403 important. Values of these types are the only ones which can be produced by
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1404 instructions.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1405
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1406</div>
1407
1408<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1409<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1410
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1411<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1412
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1413<p>The primitive types are the fundamental building blocks of the LLVM
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1414 system.</p>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1415
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1416</div>
1417
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1418<!-- _______________________________________________________________________ -->
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1419<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1420
1421<div class="doc_text">
1422
1423<h5>Overview:</h5>
1424<p>The integer type is a very simple type that simply specifies an arbitrary
1425 bit width for the integer type desired. Any bit width from 1 bit to
1426 2<sup>23</sup>-1 (about 8 million) can be specified.</p>
1427
1428<h5>Syntax:</h5>
1429<pre>
1430 iN
1431</pre>
1432
1433<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1434 value.</p>
1435
1436<h5>Examples:</h5>
1437<table class="layout">
1438 <tr class="layout">
1439 <td class="left"><tt>i1</tt></td>
1440 <td class="left">a single-bit integer.</td>
1441 </tr>
1442 <tr class="layout">
1443 <td class="left"><tt>i32</tt></td>
1444 <td class="left">a 32-bit integer.</td>
1445 </tr>
1446 <tr class="layout">
1447 <td class="left"><tt>i1942652</tt></td>
1448 <td class="left">a really big integer of over 1 million bits.</td>
1449 </tr>
1450</table>
1451
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1452</div>
1453
1454<!-- _______________________________________________________________________ -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1455<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1456
1457<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1458
1459<table>
1460 <tbody>
1461 <tr><th>Type</th><th>Description</th></tr>
1462 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1463 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1464 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1465 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1466 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1467 </tbody>
1468</table>
1469
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1470</div>
1471
1472<!-- _______________________________________________________________________ -->
1473<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1474
1475<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1476
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1477<h5>Overview:</h5>
1478<p>The void type does not represent any value and has no size.</p>
1479
1480<h5>Syntax:</h5>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1481<pre>
1482 void
1483</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1484
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1485</div>
1486
1487<!-- _______________________________________________________________________ -->
1488<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1489
1490<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1491
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1492<h5>Overview:</h5>
1493<p>The label type represents code labels.</p>
1494
1495<h5>Syntax:</h5>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1496<pre>
1497 label
1498</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1499
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1500</div>
1501
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1502<!-- _______________________________________________________________________ -->
1503<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1504
1505<div class="doc_text">
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1506
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1507<h5>Overview:</h5>
Nick Lewyckyeb021882009-09-27 23:27:42 +0000[diff] [blame]1508<p>The metadata type represents embedded metadata. No derived types may be
1509 created from metadata except for <a href="#t_function">function</a>
1510 arguments.
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1511
1512<h5>Syntax:</h5>
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1513<pre>
1514 metadata
1515</pre>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1516
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1517</div>
1518
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1519
1520<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1521<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1522
1523<div class="doc_text">
1524
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1525<p>The real power in LLVM comes from the derived types in the system. This is
1526 what allows a programmer to represent arrays, functions, pointers, and other
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1527 useful types. Each of these types contain one or more element types which
1528 may be a primitive type, or another derived type. For example, it is
1529 possible to have a two dimensional array, using an array as the element type
1530 of another array.</p>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1531
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1532</div>
1533
1534<!-- _______________________________________________________________________ -->
1535<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1536
1537<div class="doc_text">
1538
1539<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1540<p>The array type is a very simple derived type that arranges elements
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1541 sequentially in memory. The array type requires a size (number of elements)
1542 and an underlying data type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1543
1544<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1545<pre>
1546 [&lt;# elements&gt; x &lt;elementtype&gt;]
1547</pre>
1548
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1549<p>The number of elements is a constant integer value; <tt>elementtype</tt> may
1550 be any type with a size.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1551
1552<h5>Examples:</h5>
1553<table class="layout">
1554 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1555 <td class="left"><tt>[40 x i32]</tt></td>
1556 <td class="left">Array of 40 32-bit integer values.</td>
1557 </tr>
1558 <tr class="layout">
1559 <td class="left"><tt>[41 x i32]</tt></td>
1560 <td class="left">Array of 41 32-bit integer values.</td>
1561 </tr>
1562 <tr class="layout">
1563 <td class="left"><tt>[4 x i8]</tt></td>
1564 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1565 </tr>
1566</table>
1567<p>Here are some examples of multidimensional arrays:</p>
1568<table class="layout">
1569 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1570 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1571 <td class="left">3x4 array of 32-bit integer values.</td>
1572 </tr>
1573 <tr class="layout">
1574 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1575 <td class="left">12x10 array of single precision floating point values.</td>
1576 </tr>
1577 <tr class="layout">
1578 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1579 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1580 </tr>
1581</table>
1582
Dan Gohman87ec5ad2009-11-09 19:01:53 +0000[diff] [blame]1583<p>There is no restriction on indexing beyond the end of the array implied by
1584 a static type (though there are restrictions on indexing beyond the bounds
1585 of an allocated object in some cases). This means that single-dimension
1586 'variable sized array' addressing can be implemented in LLVM with a zero
1587 length array type. An implementation of 'pascal style arrays' in LLVM could
1588 use the type "<tt>{ i32, [0 x float]}</tt>", for example.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1589
1590</div>
1591
1592<!-- _______________________________________________________________________ -->
1593<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1594
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1595<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1596
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1597<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1598<p>The function type can be thought of as a function signature. It consists of
1599 a return type and a list of formal parameter types. The return type of a
1600 function type is a scalar type, a void type, or a struct type. If the return
1601 type is a struct type then all struct elements must be of first class types,
1602 and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1603
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1604<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1605<pre>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1606 &lt;returntype&gt; (&lt;parameter list&gt;)
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1607</pre>
1608
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1609<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]1610 specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1611 which indicates that the function takes a variable number of arguments.
1612 Variable argument functions can access their arguments with
1613 the <a href="#int_varargs">variable argument handling intrinsic</a>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1614 functions. '<tt>&lt;returntype&gt;</tt>' is a any type except
Nick Lewyckyeb021882009-09-27 23:27:42 +0000[diff] [blame]1615 <a href="#t_label">label</a>.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1616
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1617<h5>Examples:</h5>
1618<table class="layout">
1619 <tr class="layout">
1620 <td class="left"><tt>i32 (i32)</tt></td>
1621 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1622 </td>
1623 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1624 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1625 </tt></td>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1626 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1627 an <tt>i16</tt> that should be sign extended and a
1628 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1629 <tt>float</tt>.
1630 </td>
1631 </tr><tr class="layout">
1632 <td class="left"><tt>i32 (i8*, ...)</tt></td>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]1633 <td class="left">A vararg function that takes at least one
1634 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1635 which returns an integer. This is the signature for <tt>printf</tt> in
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1636 LLVM.
1637 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1638 </tr><tr class="layout">
1639 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Nick Lewycky3b3b54d2009-09-27 07:55:32 +0000[diff] [blame]1640 <td class="left">A function taking an <tt>i32</tt>, returning a
1641 <a href="#t_struct">structure</a> containing two <tt>i32</tt> values
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1642 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1643 </tr>
1644</table>
1645
1646</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1647
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1648<!-- _______________________________________________________________________ -->
1649<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1650
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1651<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1652
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1653<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1654<p>The structure type is used to represent a collection of data members together
1655 in memory. The packing of the field types is defined to match the ABI of the
1656 underlying processor. The elements of a structure may be any type that has a
1657 size.</p>
1658
Jeffrey Yasskine05d7732010-01-11 19:19:26 +0000[diff] [blame^]1659<p>Structures in memory are accessed using '<tt><a href="#i_load">load</a></tt>'
1660 and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a field
1661 with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
1662 Structures in registers are accessed using the
1663 '<tt><a href="#i_extractvalue">extractvalue</a></tt>' and
1664 '<tt><a href="#i_insertvalue">insertvalue</a></tt>' instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1665<h5>Syntax:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]1666<pre>
1667 { &lt;type list&gt; }
1668</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1669
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1670<h5>Examples:</h5>
1671<table class="layout">
1672 <tr class="layout">
1673 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1674 <td class="left">A triple of three <tt>i32</tt> values</td>
1675 </tr><tr class="layout">
1676 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1677 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1678 second element is a <a href="#t_pointer">pointer</a> to a
1679 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1680 an <tt>i32</tt>.</td>
1681 </tr>
1682</table>
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
Duncan Sands87598b02009-11-27 13:38:03 +0000[diff] [blame]1774 elements) and an underlying primitive data type. Vector types are considered
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1775 <a href="#t_firstclass">first class</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1776
1777<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1778<pre>
1779 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1780</pre>
1781
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1782<p>The number of elements is a constant integer value; elementtype may be any
1783 integer or floating point type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1784
1785<h5>Examples:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1786<table class="layout">
1787 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1788 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1789 <td class="left">Vector of 4 32-bit integer values.</td>
1790 </tr>
1791 <tr class="layout">
1792 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1793 <td class="left">Vector of 8 32-bit floating-point values.</td>
1794 </tr>
1795 <tr class="layout">
1796 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1797 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1798 </tr>
1799</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1800
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1801</div>
1802
1803<!-- _______________________________________________________________________ -->
1804<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1805<div class="doc_text">
1806
1807<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1808<p>Opaque types are used to represent unknown types in the system. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1809 corresponds (for example) to the C notion of a forward declared structure
1810 type. In LLVM, opaque types can eventually be resolved to any type (not just
1811 a structure type).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1812
1813<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1814<pre>
1815 opaque
1816</pre>
1817
1818<h5>Examples:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1819<table class="layout">
1820 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1821 <td class="left"><tt>opaque</tt></td>
1822 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1823 </tr>
1824</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1825
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1826</div>
1827
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1828<!-- ======================================================================= -->
1829<div class="doc_subsection">
1830 <a name="t_uprefs">Type Up-references</a>
1831</div>
1832
1833<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1834
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1835<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1836<p>An "up reference" allows you to refer to a lexically enclosing type without
1837 requiring it to have a name. For instance, a structure declaration may
1838 contain a pointer to any of the types it is lexically a member of. Example
1839 of up references (with their equivalent as named type declarations)
1840 include:</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1841
1842<pre>
Chris Lattner5ad632d2009-02-09 10:00:56 +0000[diff] [blame]1843 { \2 * } %x = type { %x* }
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1844 { \2 }* %y = type { %y }*
1845 \1* %z = type %z*
1846</pre>
1847
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1848<p>An up reference is needed by the asmprinter for printing out cyclic types
1849 when there is no declared name for a type in the cycle. Because the
1850 asmprinter does not want to print out an infinite type string, it needs a
1851 syntax to handle recursive types that have no names (all names are optional
1852 in llvm IR).</p>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1853
1854<h5>Syntax:</h5>
1855<pre>
1856 \&lt;level&gt;
1857</pre>
1858
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1859<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]1860
1861<h5>Examples:</h5>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1862<table class="layout">
1863 <tr class="layout">
1864 <td class="left"><tt>\1*</tt></td>
1865 <td class="left">Self-referential pointer.</td>
1866 </tr>
1867 <tr class="layout">
1868 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1869 <td class="left">Recursive structure where the upref refers to the out-most
1870 structure.</td>
1871 </tr>
1872</table>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1873
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1874</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1875
1876<!-- *********************************************************************** -->
1877<div class="doc_section"> <a name="constants">Constants</a> </div>
1878<!-- *********************************************************************** -->
1879
1880<div class="doc_text">
1881
1882<p>LLVM has several different basic types of constants. This section describes
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1883 them all and their syntax.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1884
1885</div>
1886
1887<!-- ======================================================================= -->
1888<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1889
1890<div class="doc_text">
1891
1892<dl>
1893 <dt><b>Boolean constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1894 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]1895 constants of the <tt><a href="#t_integer">i1</a></tt> type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1896
1897 <dt><b>Integer constants</b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1898 <dd>Standard integers (such as '4') are constants of
1899 the <a href="#t_integer">integer</a> type. Negative numbers may be used
1900 with integer types.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1901
1902 <dt><b>Floating point constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1903 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1904 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
1905 notation (see below). The assembler requires the exact decimal value of a
1906 floating-point constant. For example, the assembler accepts 1.25 but
1907 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1908 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1909
1910 <dt><b>Null pointer constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1911 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1912 and must be of <a href="#t_pointer">pointer type</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1913</dl>
1914
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1915<p>The one non-intuitive notation for constants is the hexadecimal form of
1916 floating point constants. For example, the form '<tt>double
1917 0x432ff973cafa8000</tt>' is equivalent to (but harder to read than)
1918 '<tt>double 4.5e+15</tt>'. The only time hexadecimal floating point
1919 constants are required (and the only time that they are generated by the
1920 disassembler) is when a floating point constant must be emitted but it cannot
1921 be represented as a decimal floating point number in a reasonable number of
1922 digits. For example, NaN's, infinities, and other special values are
1923 represented in their IEEE hexadecimal format so that assembly and disassembly
1924 do not cause any bits to change in the constants.</p>
1925
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1926<p>When using the hexadecimal form, constants of types float and double are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1927 represented using the 16-digit form shown above (which matches the IEEE754
1928 representation for double); float values must, however, be exactly
1929 representable as IEE754 single precision. Hexadecimal format is always used
1930 for long double, and there are three forms of long double. The 80-bit format
1931 used by x86 is represented as <tt>0xK</tt> followed by 20 hexadecimal digits.
1932 The 128-bit format used by PowerPC (two adjacent doubles) is represented
1933 by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit format
1934 is represented by <tt>0xL</tt> followed by 32 hexadecimal digits; no
1935 currently supported target uses this format. Long doubles will only work if
1936 they match the long double format on your target. All hexadecimal formats
1937 are big-endian (sign bit at the left).</p>
1938
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1939</div>
1940
1941<!-- ======================================================================= -->
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1942<div class="doc_subsection">
Bill Wendling1a2630a2009-07-20 02:32:41 +0000[diff] [blame]1943<a name="aggregateconstants"></a> <!-- old anchor -->
1944<a name="complexconstants">Complex Constants</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1945</div>
1946
1947<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1948
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1949<p>Complex constants are a (potentially recursive) combination of simple
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1950 constants and smaller complex constants.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1951
1952<dl>
1953 <dt><b>Structure constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1954 <dd>Structure constants are represented with notation similar to structure
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1955 type definitions (a comma separated list of elements, surrounded by braces
1956 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1957 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>".
1958 Structure constants must have <a href="#t_struct">structure type</a>, and
1959 the number and types of elements must match those specified by the
1960 type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1961
1962 <dt><b>Array constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1963 <dd>Array constants are represented with notation similar to array type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1964 definitions (a comma separated list of elements, surrounded by square
1965 brackets (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74
1966 ]</tt>". Array constants must have <a href="#t_array">array type</a>, and
1967 the number and types of elements must match those specified by the
1968 type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1969
1970 <dt><b>Vector constants</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1971 <dd>Vector constants are represented with notation similar to vector type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1972 definitions (a comma separated list of elements, surrounded by
1973 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32
1974 42, i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must
1975 have <a href="#t_vector">vector type</a>, and the number and types of
1976 elements must match those specified by the type.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1977
1978 <dt><b>Zero initialization</b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1979 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1980 value to zero of <em>any</em> type, including scalar and aggregate types.
1981 This is often used to avoid having to print large zero initializers
1982 (e.g. for large arrays) and is always exactly equivalent to using explicit
1983 zero initializers.</dd>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]1984
1985 <dt><b>Metadata node</b></dt>
Nick Lewyckyf122c7e2009-05-30 16:08:30 +0000[diff] [blame]1986 <dd>A metadata node is a structure-like constant with
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]1987 <a href="#t_metadata">metadata type</a>. For example: "<tt>metadata !{
1988 i32 0, metadata !"test" }</tt>". Unlike other constants that are meant to
1989 be interpreted as part of the instruction stream, metadata is a place to
1990 attach additional information such as debug info.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1991</dl>
1992
1993</div>
1994
1995<!-- ======================================================================= -->
1996<div class="doc_subsection">
1997 <a name="globalconstants">Global Variable and Function Addresses</a>
1998</div>
1999
2000<div class="doc_text">
2001
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2002<p>The addresses of <a href="#globalvars">global variables</a>
2003 and <a href="#functionstructure">functions</a> are always implicitly valid
2004 (link-time) constants. These constants are explicitly referenced when
2005 the <a href="#identifiers">identifier for the global</a> is used and always
2006 have <a href="#t_pointer">pointer</a> type. For example, the following is a
2007 legal LLVM file:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2008
2009<div class="doc_code">
2010<pre>
2011@X = global i32 17
2012@Y = global i32 42
2013@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
2014</pre>
2015</div>
2016
2017</div>
2018
2019<!-- ======================================================================= -->
2020<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
2021<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2022
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2023<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]2024 indicates that the user of the value may receive an unspecified bit-pattern.
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2025 Undefined values may be of any type (other than label or void) and be used
2026 anywhere a constant is permitted.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2027
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2028<p>Undefined values are useful because they indicate to the compiler that the
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2029 program is well defined no matter what value is used. This gives the
2030 compiler more freedom to optimize. Here are some examples of (potentially
2031 surprising) transformations that are valid (in pseudo IR):</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2032
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2033
2034<div class="doc_code">
2035<pre>
2036 %A = add %X, undef
2037 %B = sub %X, undef
2038 %C = xor %X, undef
2039Safe:
2040 %A = undef
2041 %B = undef
2042 %C = undef
2043</pre>
2044</div>
2045
2046<p>This is safe because all of the output bits are affected by the undef bits.
2047Any output bit can have a zero or one depending on the input bits.</p>
2048
2049<div class="doc_code">
2050<pre>
2051 %A = or %X, undef
2052 %B = and %X, undef
2053Safe:
2054 %A = -1
2055 %B = 0
2056Unsafe:
2057 %A = undef
2058 %B = undef
2059</pre>
2060</div>
2061
2062<p>These logical operations have bits that are not always affected by the input.
2063For example, if "%X" has a zero bit, then the output of the 'and' operation will
2064always be a zero, no matter what the corresponding bit from the undef is. As
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2065such, it is unsafe to optimize or assume that the result of the and is undef.
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2066However, it is safe to assume that all bits of the undef could be 0, and
2067optimize the and to 0. Likewise, it is safe to assume that all the bits of
2068the undef operand to the or could be set, allowing the or to be folded to
Chris Lattner9d1e4fa2009-09-11 01:49:31 +0000[diff] [blame]2069-1.</p>
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2070
2071<div class="doc_code">
2072<pre>
2073 %A = select undef, %X, %Y
2074 %B = select undef, 42, %Y
2075 %C = select %X, %Y, undef
2076Safe:
2077 %A = %X (or %Y)
2078 %B = 42 (or %Y)
2079 %C = %Y
2080Unsafe:
2081 %A = undef
2082 %B = undef
2083 %C = undef
2084</pre>
2085</div>
2086
2087<p>This set of examples show that undefined select (and conditional branch)
2088conditions can go "either way" but they have to come from one of the two
2089operands. In the %A example, if %X and %Y were both known to have a clear low
2090bit, then %A would have to have a cleared low bit. However, in the %C example,
2091the optimizer is allowed to assume that the undef operand could be the same as
2092%Y, allowing the whole select to be eliminated.</p>
2093
2094
2095<div class="doc_code">
2096<pre>
2097 %A = xor undef, undef
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2098
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2099 %B = undef
2100 %C = xor %B, %B
2101
2102 %D = undef
2103 %E = icmp lt %D, 4
2104 %F = icmp gte %D, 4
2105
2106Safe:
2107 %A = undef
2108 %B = undef
2109 %C = undef
2110 %D = undef
2111 %E = undef
2112 %F = undef
2113</pre>
2114</div>
2115
2116<p>This example points out that two undef operands are not necessarily the same.
2117This can be surprising to people (and also matches C semantics) where they
2118assume that "X^X" is always zero, even if X is undef. This isn't true for a
2119number of reasons, but the short answer is that an undef "variable" can
2120arbitrarily change its value over its "live range". This is true because the
2121"variable" doesn't actually <em>have a live range</em>. Instead, the value is
2122logically read from arbitrary registers that happen to be around when needed,
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2123so the value is not necessarily consistent over time. In fact, %A and %C need
Chris Lattnerc802e482009-09-08 15:13:16 +0000[diff] [blame]2124to have the same semantics or the core LLVM "replace all uses with" concept
Chris Lattner3d72cd82009-09-07 22:52:39 +0000[diff] [blame]2125would not hold.</p>
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2126
2127<div class="doc_code">
2128<pre>
2129 %A = fdiv undef, %X
2130 %B = fdiv %X, undef
2131Safe:
2132 %A = undef
2133b: unreachable
2134</pre>
2135</div>
2136
2137<p>These examples show the crucial difference between an <em>undefined
2138value</em> and <em>undefined behavior</em>. An undefined value (like undef) is
2139allowed to have an arbitrary bit-pattern. This means that the %A operation
2140can be constant folded to undef because the undef could be an SNaN, and fdiv is
2141not (currently) defined on SNaN's. However, in the second example, we can make
2142a more aggressive assumption: because the undef is allowed to be an arbitrary
2143value, 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]2144has <em>undefined behavior</em>, we are allowed to assume that the operation
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2145does not execute at all. This allows us to delete the divide and all code after
2146it: since the undefined operation "can't happen", the optimizer can assume that
2147it occurs in dead code.
2148</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2149
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2150<div class="doc_code">
2151<pre>
2152a: store undef -> %X
2153b: store %X -> undef
2154Safe:
2155a: &lt;deleted&gt;
2156b: unreachable
2157</pre>
2158</div>
2159
2160<p>These examples reiterate the fdiv example: a store "of" an undefined value
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2161can be assumed to not have any effect: we can assume that the value is
Chris Lattner466291f2009-09-07 23:33:52 +0000[diff] [blame]2162overwritten with bits that happen to match what was already there. However, a
2163store "to" an undefined location could clobber arbitrary memory, therefore, it
2164has undefined behavior.</p>
2165
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2166</div>
2167
2168<!-- ======================================================================= -->
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]2169<div class="doc_subsection"><a name="blockaddress">Addresses of Basic
2170 Blocks</a></div>
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2171<div class="doc_text">
2172
Chris Lattner620cead2009-11-01 01:27:45 +0000[diff] [blame]2173<p><b><tt>blockaddress(@function, %block)</tt></b></p>
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2174
2175<p>The '<tt>blockaddress</tt>' constant computes the address of the specified
Chris Lattner24f2e852009-10-27 21:49:40 +0000[diff] [blame]2176 basic block in the specified function, and always has an i8* type. Taking
Chris Lattner620cead2009-11-01 01:27:45 +0000[diff] [blame]2177 the address of the entry block is illegal.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2178
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2179<p>This value only has defined behavior when used as an operand to the
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2180 '<a href="#i_indirectbr"><tt>indirectbr</tt></a>' instruction or for comparisons
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2181 against null. Pointer equality tests between labels addresses is undefined
2182 behavior - though, again, comparison against null is ok, and no label is
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]2183 equal to the null pointer. This may also be passed around as an opaque
2184 pointer sized value as long as the bits are not inspected. This allows
Chris Lattner0bae7b32009-10-27 21:44:20 +0000[diff] [blame]2185 <tt>ptrtoint</tt> and arithmetic to be performed on these values so long as
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2186 the original value is reconstituted before the <tt>indirectbr</tt>.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2187
Chris Lattner29246b52009-10-27 21:19:13 +0000[diff] [blame]2188<p>Finally, some targets may provide defined semantics when
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2189 using the value as the operand to an inline assembly, but that is target
2190 specific.
2191 </p>
2192
2193</div>
2194
2195
2196<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2197<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
2198</div>
2199
2200<div class="doc_text">
2201
2202<p>Constant expressions are used to allow expressions involving other constants
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2203 to be used as constants. Constant expressions may be of
2204 any <a href="#t_firstclass">first class</a> type and may involve any LLVM
2205 operation that does not have side effects (e.g. load and call are not
2206 supported). The following is the syntax for constant expressions:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2207
2208<dl>
2209 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2210 <dd>Truncate a constant to another type. The bit size of CST must be larger
2211 than the bit size of TYPE. Both types must be integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2212
2213 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2214 <dd>Zero extend a constant to another type. The bit size of CST must be
2215 smaller or equal to the bit size of TYPE. Both types must be
2216 integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2217
2218 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2219 <dd>Sign extend a constant to another type. The bit size of CST must be
2220 smaller or equal to the bit size of TYPE. Both types must be
2221 integers.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2222
2223 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2224 <dd>Truncate a floating point constant to another floating point type. The
2225 size of CST must be larger than the size of TYPE. Both types must be
2226 floating point.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2227
2228 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2229 <dd>Floating point extend a constant to another type. The size of CST must be
2230 smaller or equal to the size of TYPE. Both types must be floating
2231 point.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2232
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]2233 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2234 <dd>Convert a floating point constant to the corresponding unsigned integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2235 constant. TYPE must be a scalar or vector integer type. CST must be of
2236 scalar or vector floating point type. Both CST and TYPE must be scalars,
2237 or vectors of the same number of elements. If the value won't fit in the
2238 integer type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2239
2240 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
2241 <dd>Convert a floating point constant to the corresponding signed integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2242 constant. TYPE must be a scalar or vector integer type. CST must be of
2243 scalar or vector floating point type. Both CST and TYPE must be scalars,
2244 or vectors of the same number of elements. If the value won't fit in the
2245 integer type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2246
2247 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
2248 <dd>Convert an unsigned integer constant to the corresponding floating point
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2249 constant. TYPE must be a scalar or vector floating point type. CST must be
2250 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2251 vectors of the same number of elements. If the value won't fit in the
2252 floating point type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2253
2254 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2255 <dd>Convert a signed integer constant to the corresponding floating point
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2256 constant. TYPE must be a scalar or vector floating point type. CST must be
2257 of scalar or vector integer type. Both CST and TYPE must be scalars, or
2258 vectors of the same number of elements. If the value won't fit in the
2259 floating point type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2260
2261 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2262 <dd>Convert a pointer typed constant to the corresponding integer constant
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2263 <tt>TYPE</tt> must be an integer type. <tt>CST</tt> must be of pointer
2264 type. The <tt>CST</tt> value is zero extended, truncated, or unchanged to
2265 make it fit in <tt>TYPE</tt>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2266
2267 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2268 <dd>Convert a integer constant to a pointer constant. TYPE must be a pointer
2269 type. CST must be of integer type. The CST value is zero extended,
2270 truncated, or unchanged to make it fit in a pointer size. This one is
2271 <i>really</i> dangerous!</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2272
2273 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Chris Lattner557bc5d2009-02-28 18:27:03 +0000[diff] [blame]2274 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2275 are the same as those for the <a href="#i_bitcast">bitcast
2276 instruction</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2277
2278 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]2279 <dt><b><tt>getelementptr inbounds ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2280 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2281 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2282 instruction, the index list may have zero or more indexes, which are
2283 required to make sense for the type of "CSTPTR".</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2284
2285 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2286 <dd>Perform the <a href="#i_select">select operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2287
2288 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2289 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2290
2291 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2292 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2293
2294 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2295 <dd>Perform the <a href="#i_extractelement">extractelement operation</a> on
2296 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2297
2298 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2299 <dd>Perform the <a href="#i_insertelement">insertelement operation</a> on
2300 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2301
2302 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2303 <dd>Perform the <a href="#i_shufflevector">shufflevector operation</a> on
2304 constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2305
2306 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2307 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2308 be any of the <a href="#binaryops">binary</a>
2309 or <a href="#bitwiseops">bitwise binary</a> operations. The constraints
2310 on operands are the same as those for the corresponding instruction
2311 (e.g. no bitwise operations on floating point values are allowed).</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2312</dl>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2313
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2314</div>
2315
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2316<!-- ======================================================================= -->
2317<div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
2318</div>
2319
2320<div class="doc_text">
2321
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2322<p>Embedded metadata provides a way to attach arbitrary data to the instruction
2323 stream without affecting the behaviour of the program. There are two
2324 metadata primitives, strings and nodes. All metadata has the
2325 <tt>metadata</tt> type and is identified in syntax by a preceding exclamation
2326 point ('<tt>!</tt>').</p>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2327
2328<p>A metadata string is a string surrounded by double quotes. It can contain
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2329 any character by escaping non-printable characters with "\xx" where "xx" is
2330 the two digit hex code. For example: "<tt>!"test\00"</tt>".</p>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2331
2332<p>Metadata nodes are represented with notation similar to structure constants
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2333 (a comma separated list of elements, surrounded by braces and preceded by an
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2334 exclamation point). For example: "<tt>!{ metadata !"test\00", i32
2335 10}</tt>".</p>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2336
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2337<p>A metadata node will attempt to track changes to the values it holds. In the
2338 event that a value is deleted, it will be replaced with a typeless
2339 "<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
Nick Lewycky117f4382009-05-10 20:57:05 +0000[diff] [blame]2340
Devang Patelc5bb7252010-01-05 20:41:31 +0000[diff] [blame]2341<p>A named metadata is a collection of metadata nodes. For example: "<tt>!foo =
2342 metadata !{!4, !3}</tt>".
2343
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2344<p>Optimizations may rely on metadata to provide additional information about
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2345 the program that isn't available in the instructions, or that isn't easily
2346 computable. Similarly, the code generator may expect a certain metadata
2347 format to be used to express debugging information.</p>
2348
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2349</div>
2350
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2351<!-- *********************************************************************** -->
2352<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2353<!-- *********************************************************************** -->
2354
2355<!-- ======================================================================= -->
2356<div class="doc_subsection">
2357<a name="inlineasm">Inline Assembler Expressions</a>
2358</div>
2359
2360<div class="doc_text">
2361
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2362<p>LLVM supports inline assembler expressions (as opposed
2363 to <a href="#moduleasm"> Module-Level Inline Assembly</a>) through the use of
2364 a special value. This value represents the inline assembler as a string
2365 (containing the instructions to emit), a list of operand constraints (stored
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2366 as a string), a flag that indicates whether or not the inline asm
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2367 expression has side effects, and a flag indicating whether the function
2368 containing the asm needs to align its stack conservatively. An example
2369 inline assembler expression is:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2370
2371<div class="doc_code">
2372<pre>
2373i32 (i32) asm "bswap $0", "=r,r"
2374</pre>
2375</div>
2376
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2377<p>Inline assembler expressions may <b>only</b> be used as the callee operand of
2378 a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we
2379 have:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2380
2381<div class="doc_code">
2382<pre>
2383%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2384</pre>
2385</div>
2386
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2387<p>Inline asms with side effects not visible in the constraint list must be
2388 marked as having side effects. This is done through the use of the
2389 '<tt>sideeffect</tt>' keyword, like so:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2390
2391<div class="doc_code">
2392<pre>
2393call void asm sideeffect "eieio", ""()
2394</pre>
2395</div>
2396
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2397<p>In some cases inline asms will contain code that will not work unless the
2398 stack is aligned in some way, such as calls or SSE instructions on x86,
2399 yet will not contain code that does that alignment within the asm.
2400 The compiler should make conservative assumptions about what the asm might
2401 contain and should generate its usual stack alignment code in the prologue
2402 if the '<tt>alignstack</tt>' keyword is present:</p>
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2403
2404<div class="doc_code">
2405<pre>
Dale Johannesen5ee3e4b2009-10-21 23:28:00 +0000[diff] [blame]2406call void asm alignstack "eieio", ""()
Dale Johannesen648950f2009-10-13 21:56:55 +0000[diff] [blame]2407</pre>
2408</div>
2409
2410<p>If both keywords appear the '<tt>sideeffect</tt>' keyword must come
2411 first.</p>
2412
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2413<p>TODO: The format of the asm and constraints string still need to be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2414 documented here. Constraints on what can be done (e.g. duplication, moving,
2415 etc need to be documented). This is probably best done by reference to
2416 another document that covers inline asm from a holistic perspective.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2417
2418</div>
2419
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2420
2421<!-- *********************************************************************** -->
2422<div class="doc_section">
2423 <a name="intrinsic_globals">Intrinsic Global Variables</a>
2424</div>
2425<!-- *********************************************************************** -->
2426
2427<p>LLVM has a number of "magic" global variables that contain data that affect
2428code generation or other IR semantics. These are documented here. All globals
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]2429of this sort should have a section specified as "<tt>llvm.metadata</tt>". This
2430section and all globals that start with "<tt>llvm.</tt>" are reserved for use
2431by LLVM.</p>
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2432
2433<!-- ======================================================================= -->
2434<div class="doc_subsection">
2435<a name="intg_used">The '<tt>llvm.used</tt>' Global Variable</a>
2436</div>
2437
2438<div class="doc_text">
2439
2440<p>The <tt>@llvm.used</tt> global is an array with i8* element type which has <a
2441href="#linkage_appending">appending linkage</a>. This array contains a list of
2442pointers to global variables and functions which may optionally have a pointer
2443cast formed of bitcast or getelementptr. For example, a legal use of it is:</p>
2444
2445<pre>
2446 @X = global i8 4
2447 @Y = global i32 123
2448
2449 @llvm.used = appending global [2 x i8*] [
2450 i8* @X,
2451 i8* bitcast (i32* @Y to i8*)
2452 ], section "llvm.metadata"
2453</pre>
2454
2455<p>If a global variable appears in the <tt>@llvm.used</tt> list, then the
2456compiler, assembler, and linker are required to treat the symbol as if there is
2457a reference to the global that it cannot see. For example, if a variable has
2458internal linkage and no references other than that from the <tt>@llvm.used</tt>
2459list, it cannot be deleted. This is commonly used to represent references from
2460inline asms and other things the compiler cannot "see", and corresponds to
2461"attribute((used))" in GNU C.</p>
2462
2463<p>On some targets, the code generator must emit a directive to the assembler or
2464object file to prevent the assembler and linker from molesting the symbol.</p>
2465
2466</div>
2467
2468<!-- ======================================================================= -->
2469<div class="doc_subsection">
Chris Lattner1e0e0d12009-07-20 06:14:25 +0000[diff] [blame]2470<a name="intg_compiler_used">The '<tt>llvm.compiler.used</tt>' Global Variable</a>
2471</div>
2472
2473<div class="doc_text">
2474
2475<p>The <tt>@llvm.compiler.used</tt> directive is the same as the
2476<tt>@llvm.used</tt> directive, except that it only prevents the compiler from
2477touching the symbol. On targets that support it, this allows an intelligent
2478linker to optimize references to the symbol without being impeded as it would be
2479by <tt>@llvm.used</tt>.</p>
2480
2481<p>This is a rare construct that should only be used in rare circumstances, and
2482should not be exposed to source languages.</p>
2483
2484</div>
2485
2486<!-- ======================================================================= -->
2487<div class="doc_subsection">
Chris Lattner75c24e02009-07-20 05:55:19 +0000[diff] [blame]2488<a name="intg_global_ctors">The '<tt>llvm.global_ctors</tt>' Global Variable</a>
2489</div>
2490
2491<div class="doc_text">
2492
2493<p>TODO: Describe this.</p>
2494
2495</div>
2496
2497<!-- ======================================================================= -->
2498<div class="doc_subsection">
2499<a name="intg_global_dtors">The '<tt>llvm.global_dtors</tt>' Global Variable</a>
2500</div>
2501
2502<div class="doc_text">
2503
2504<p>TODO: Describe this.</p>
2505
2506</div>
2507
2508
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2509<!-- *********************************************************************** -->
2510<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2511<!-- *********************************************************************** -->
2512
2513<div class="doc_text">
2514
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2515<p>The LLVM instruction set consists of several different classifications of
2516 instructions: <a href="#terminators">terminator
2517 instructions</a>, <a href="#binaryops">binary instructions</a>,
2518 <a href="#bitwiseops">bitwise binary instructions</a>,
2519 <a href="#memoryops">memory instructions</a>, and
2520 <a href="#otherops">other instructions</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2521
2522</div>
2523
2524<!-- ======================================================================= -->
2525<div class="doc_subsection"> <a name="terminators">Terminator
2526Instructions</a> </div>
2527
2528<div class="doc_text">
2529
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2530<p>As mentioned <a href="#functionstructure">previously</a>, every basic block
2531 in a program ends with a "Terminator" instruction, which indicates which
2532 block should be executed after the current block is finished. These
2533 terminator instructions typically yield a '<tt>void</tt>' value: they produce
2534 control flow, not values (the one exception being the
2535 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2536
2537<p>There are six different terminator instructions: the
2538 '<a href="#i_ret"><tt>ret</tt></a>' instruction, the
2539 '<a href="#i_br"><tt>br</tt></a>' instruction, the
2540 '<a href="#i_switch"><tt>switch</tt></a>' instruction, the
Bill Wendlingb4ae2292009-11-02 00:25:26 +0000[diff] [blame]2541 '<a href="#i_indirectbr">'<tt>indirectbr</tt></a>' Instruction, the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2542 '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the
2543 '<a href="#i_unwind"><tt>unwind</tt></a>' instruction, and the
2544 '<a href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2545
2546</div>
2547
2548<!-- _______________________________________________________________________ -->
2549<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2550Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2551
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2552<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2553
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2554<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2555<pre>
2556 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]2557 ret void <i>; Return from void function</i>
2558</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2559
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2560<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2561<p>The '<tt>ret</tt>' instruction is used to return control flow (and optionally
2562 a value) from a function back to the caller.</p>
2563
2564<p>There are two forms of the '<tt>ret</tt>' instruction: one that returns a
2565 value and then causes control flow, and one that just causes control flow to
2566 occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2567
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2568<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2569<p>The '<tt>ret</tt>' instruction optionally accepts a single argument, the
2570 return value. The type of the return value must be a
2571 '<a href="#t_firstclass">first class</a>' type.</p>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2572
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2573<p>A function is not <a href="#wellformed">well formed</a> if it it has a
2574 non-void return type and contains a '<tt>ret</tt>' instruction with no return
2575 value or a return value with a type that does not match its type, or if it
2576 has a void return type and contains a '<tt>ret</tt>' instruction with a
2577 return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2578
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2579<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2580<p>When the '<tt>ret</tt>' instruction is executed, control flow returns back to
2581 the calling function's context. If the caller is a
2582 "<a href="#i_call"><tt>call</tt></a>" instruction, execution continues at the
2583 instruction after the call. If the caller was an
2584 "<a href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues at
2585 the beginning of the "normal" destination block. If the instruction returns
2586 a value, that value shall set the call or invoke instruction's return
2587 value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2588
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2589<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2590<pre>
2591 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2592 ret void <i>; Return from a void function</i>
Bill Wendlingd163e2d2009-02-28 22:12:54 +0000[diff] [blame]2593 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]2594</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2595
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2596</div>
2597<!-- _______________________________________________________________________ -->
2598<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2599
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2600<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2601
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2602<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2603<pre>
2604 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]2605</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2606
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2607<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2608<p>The '<tt>br</tt>' instruction is used to cause control flow to transfer to a
2609 different basic block in the current function. There are two forms of this
2610 instruction, corresponding to a conditional branch and an unconditional
2611 branch.</p>
2612
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2613<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2614<p>The conditional branch form of the '<tt>br</tt>' instruction takes a single
2615 '<tt>i1</tt>' value and two '<tt>label</tt>' values. The unconditional form
2616 of the '<tt>br</tt>' instruction takes a single '<tt>label</tt>' value as a
2617 target.</p>
2618
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2619<h5>Semantics:</h5>
2620<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2621 argument is evaluated. If the value is <tt>true</tt>, control flows to the
2622 '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2623 control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2624
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2625<h5>Example:</h5>
Bill Wendling6ec40612009-07-20 02:39:26 +0000[diff] [blame]2626<pre>
2627Test:
2628 %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b
2629 br i1 %cond, label %IfEqual, label %IfUnequal
2630IfEqual:
2631 <a href="#i_ret">ret</a> i32 1
2632IfUnequal:
2633 <a href="#i_ret">ret</a> i32 0
2634</pre>
2635
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2636</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2637
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2638<!-- _______________________________________________________________________ -->
2639<div class="doc_subsubsection">
2640 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2641</div>
2642
2643<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2644
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2645<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2646<pre>
2647 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2648</pre>
2649
2650<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2651<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]2652 several different places. It is a generalization of the '<tt>br</tt>'
2653 instruction, allowing a branch to occur to one of many possible
2654 destinations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2655
2656<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2657<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2658 comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination,
2659 and an array of pairs of comparison value constants and '<tt>label</tt>'s.
2660 The table is not allowed to contain duplicate constant entries.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2661
2662<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2663<p>The <tt>switch</tt> instruction specifies a table of values and
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2664 destinations. When the '<tt>switch</tt>' instruction is executed, this table
2665 is searched for the given value. If the value is found, control flow is
Benjamin Kramer5fb9d7e2009-10-12 14:46:08 +0000[diff] [blame]2666 transferred to the corresponding destination; otherwise, control flow is
2667 transferred to the default destination.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2668
2669<h5>Implementation:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2670<p>Depending on properties of the target machine and the particular
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2671 <tt>switch</tt> instruction, this instruction may be code generated in
2672 different ways. For example, it could be generated as a series of chained
2673 conditional branches or with a lookup table.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2674
2675<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2676<pre>
2677 <i>; Emulate a conditional br instruction</i>
2678 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2679 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2680
2681 <i>; Emulate an unconditional br instruction</i>
2682 switch i32 0, label %dest [ ]
2683
2684 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2685 switch i32 %val, label %otherwise [ i32 0, label %onzero
2686 i32 1, label %onone
2687 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2688</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2689
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2690</div>
2691
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2692
2693<!-- _______________________________________________________________________ -->
2694<div class="doc_subsubsection">
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2695 <a name="i_indirectbr">'<tt>indirectbr</tt>' Instruction</a>
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2696</div>
2697
2698<div class="doc_text">
2699
2700<h5>Syntax:</h5>
2701<pre>
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2702 indirectbr &lt;somety&gt;* &lt;address&gt;, [ label &lt;dest1&gt;, label &lt;dest2&gt;, ... ]
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2703</pre>
2704
2705<h5>Overview:</h5>
2706
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2707<p>The '<tt>indirectbr</tt>' instruction implements an indirect branch to a label
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2708 within the current function, whose address is specified by
Chris Lattnerd07c8372009-10-27 21:01:34 +0000[diff] [blame]2709 "<tt>address</tt>". Address must be derived from a <a
2710 href="#blockaddress">blockaddress</a> constant.</p>
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2711
2712<h5>Arguments:</h5>
2713
2714<p>The '<tt>address</tt>' argument is the address of the label to jump to. The
2715 rest of the arguments indicate the full set of possible destinations that the
2716 address may point to. Blocks are allowed to occur multiple times in the
2717 destination list, though this isn't particularly useful.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]2718
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2719<p>This destination list is required so that dataflow analysis has an accurate
2720 understanding of the CFG.</p>
2721
2722<h5>Semantics:</h5>
2723
2724<p>Control transfers to the block specified in the address argument. All
2725 possible destination blocks must be listed in the label list, otherwise this
2726 instruction has undefined behavior. This implies that jumps to labels
2727 defined in other functions have undefined behavior as well.</p>
2728
2729<h5>Implementation:</h5>
2730
2731<p>This is typically implemented with a jump through a register.</p>
2732
2733<h5>Example:</h5>
2734<pre>
Chris Lattner4c3800f2009-10-28 00:19:10 +0000[diff] [blame]2735 indirectbr i8* %Addr, [ label %bb1, label %bb2, label %bb3 ]
Chris Lattnere0787282009-10-27 19:13:16 +0000[diff] [blame]2736</pre>
2737
2738</div>
2739
2740
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2741<!-- _______________________________________________________________________ -->
2742<div class="doc_subsubsection">
2743 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2744</div>
2745
2746<div class="doc_text">
2747
2748<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2749<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2750 &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]2751 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2752</pre>
2753
2754<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2755<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2756 function, with the possibility of control flow transfer to either the
2757 '<tt>normal</tt>' label or the '<tt>exception</tt>' label. If the callee
2758 function returns with the "<tt><a href="#i_ret">ret</a></tt>" instruction,
2759 control flow will return to the "normal" label. If the callee (or any
2760 indirect callees) returns with the "<a href="#i_unwind"><tt>unwind</tt></a>"
2761 instruction, control is interrupted and continued at the dynamically nearest
2762 "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2763
2764<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2765<p>This instruction requires several arguments:</p>
2766
2767<ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2768 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
2769 convention</a> the call should use. If none is specified, the call
2770 defaults to using C calling conventions.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2771
2772 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2773 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
2774 '<tt>inreg</tt>' attributes are valid here.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2775
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2776 <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]2777 function value being invoked. In most cases, this is a direct function
2778 invocation, but indirect <tt>invoke</tt>s are just as possible, branching
2779 off an arbitrary pointer to function value.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2780
2781 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2782 function to be invoked. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2783
2784 <li>'<tt>function args</tt>': argument list whose types match the function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2785 signature argument types. If the function signature indicates the
2786 function accepts a variable number of arguments, the extra arguments can
2787 be specified.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2788
2789 <li>'<tt>normal label</tt>': the label reached when the called function
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2790 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2791
2792 <li>'<tt>exception label</tt>': the label reached when a callee returns with
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2793 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2794
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2795 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2796 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2797 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2798</ol>
2799
2800<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2801<p>This instruction is designed to operate as a standard
2802 '<tt><a href="#i_call">call</a></tt>' instruction in most regards. The
2803 primary difference is that it establishes an association with a label, which
2804 is used by the runtime library to unwind the stack.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2805
2806<p>This instruction is used in languages with destructors to ensure that proper
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2807 cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2808 exception. Additionally, this is important for implementation of
2809 '<tt>catch</tt>' clauses in high-level languages that support them.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2810
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2811<p>For the purposes of the SSA form, the definition of the value returned by the
2812 '<tt>invoke</tt>' instruction is deemed to occur on the edge from the current
2813 block to the "normal" label. If the callee unwinds then no return value is
2814 available.</p>
Dan Gohman140ba5d2009-05-22 21:47:08 +0000[diff] [blame]2815
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2816<h5>Example:</h5>
2817<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2818 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2819 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2820 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2821 unwind label %TestCleanup <i>; {i32}:retval set</i>
2822</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2823
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2824</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2825
2826<!-- _______________________________________________________________________ -->
2827
2828<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2829Instruction</a> </div>
2830
2831<div class="doc_text">
2832
2833<h5>Syntax:</h5>
2834<pre>
2835 unwind
2836</pre>
2837
2838<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2839<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2840 at the first callee in the dynamic call stack which used
2841 an <a href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call.
2842 This is primarily used to implement exception handling.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2843
2844<h5>Semantics:</h5>
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2845<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2846 immediately halt. The dynamic call stack is then searched for the
2847 first <a href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack.
2848 Once found, execution continues at the "exceptional" destination block
2849 specified by the <tt>invoke</tt> instruction. If there is no <tt>invoke</tt>
2850 instruction in the dynamic call chain, undefined behavior results.</p>
2851
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2852</div>
2853
2854<!-- _______________________________________________________________________ -->
2855
2856<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2857Instruction</a> </div>
2858
2859<div class="doc_text">
2860
2861<h5>Syntax:</h5>
2862<pre>
2863 unreachable
2864</pre>
2865
2866<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2867<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2868 instruction is used to inform the optimizer that a particular portion of the
2869 code is not reachable. This can be used to indicate that the code after a
2870 no-return function cannot be reached, and other facts.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2871
2872<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2873<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2874
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2875</div>
2876
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2877<!-- ======================================================================= -->
2878<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2879
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2880<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2881
2882<p>Binary operators are used to do most of the computation in a program. They
2883 require two operands of the same type, execute an operation on them, and
2884 produce a single value. The operands might represent multiple data, as is
2885 the case with the <a href="#t_vector">vector</a> data type. The result value
2886 has the same type as its operands.</p>
2887
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2888<p>There are several different binary operators:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2889
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2890</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2891
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2892<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2893<div class="doc_subsubsection">
2894 <a name="i_add">'<tt>add</tt>' Instruction</a>
2895</div>
2896
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2897<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2898
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2899<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2900<pre>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]2901 &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]2902 &lt;result&gt; = add nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2903 &lt;result&gt; = add nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2904 &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]2905</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2906
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2907<h5>Overview:</h5>
2908<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2909
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2910<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2911<p>The two arguments to the '<tt>add</tt>' instruction must
2912 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2913 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2914
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2915<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2916<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2917
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2918<p>If the sum has unsigned overflow, the result returned is the mathematical
2919 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]2920
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2921<p>Because LLVM integers use a two's complement representation, this instruction
2922 is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2923
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]2924<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
2925 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
2926 <tt>nsw</tt> keywords are present, the result value of the <tt>add</tt>
2927 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]2928
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2929<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2930<pre>
2931 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2932</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2933
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2934</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2935
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2936<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2937<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2938 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2939</div>
2940
2941<div class="doc_text">
2942
2943<h5>Syntax:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2944<pre>
2945 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2946</pre>
2947
2948<h5>Overview:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2949<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2950
2951<h5>Arguments:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2952<p>The two arguments to the '<tt>fadd</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2953 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2954 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2955
2956<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2957<p>The value produced is the floating point sum of the two operands.</p>
2958
2959<h5>Example:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2960<pre>
2961 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2962</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2963
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2964</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2965
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2966<!-- _______________________________________________________________________ -->
2967<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2968 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2969</div>
2970
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2971<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2972
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2973<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2974<pre>
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]2975 &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]2976 &lt;result&gt; = sub nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2977 &lt;result&gt; = sub nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2978 &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]2979</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2980
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2981<h5>Overview:</h5>
2982<p>The '<tt>sub</tt>' instruction returns the difference of its two
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2983 operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2984
2985<p>Note that the '<tt>sub</tt>' instruction is used to represent the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2986 '<tt>neg</tt>' instruction present in most other intermediate
2987 representations.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2988
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2989<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2990<p>The two arguments to the '<tt>sub</tt>' instruction must
2991 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2992 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2993
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2994<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2995<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2996
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2997<p>If the difference has unsigned overflow, the result returned is the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]2998 mathematical result modulo 2<sup>n</sup>, where n is the bit width of the
2999 result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3000
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3001<p>Because LLVM integers use a two's complement representation, this instruction
3002 is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3003
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3004<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3005 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3006 <tt>nsw</tt> keywords are present, the result value of the <tt>sub</tt>
3007 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3008
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3009<h5>Example:</h5>
3010<pre>
3011 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
3012 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
3013</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3014
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3015</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3016
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3017<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3018<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3019 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
3020</div>
3021
3022<div class="doc_text">
3023
3024<h5>Syntax:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3025<pre>
3026 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3027</pre>
3028
3029<h5>Overview:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3030<p>The '<tt>fsub</tt>' instruction returns the difference of its two
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3031 operands.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3032
3033<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3034 '<tt>fneg</tt>' instruction present in most other intermediate
3035 representations.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3036
3037<h5>Arguments:</h5>
Bill Wendling1a2630a2009-07-20 02:32:41 +0000[diff] [blame]3038<p>The two arguments to the '<tt>fsub</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3039 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3040 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3041
3042<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3043<p>The value produced is the floating point difference of the two operands.</p>
3044
3045<h5>Example:</h5>
3046<pre>
3047 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
3048 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
3049</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3050
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3051</div>
3052
3053<!-- _______________________________________________________________________ -->
3054<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3055 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
3056</div>
3057
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3058<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3059
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3060<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3061<pre>
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3062 &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]3063 &lt;result&gt; = mul nuw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3064 &lt;result&gt; = mul nsw &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3065 &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]3066</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3067
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3068<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3069<p>The '<tt>mul</tt>' instruction returns the product of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3070
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3071<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3072<p>The two arguments to the '<tt>mul</tt>' instruction must
3073 be <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3074 integer values. Both arguments must have identical types.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3075
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3076<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3077<p>The value produced is the integer product of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3078
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3079<p>If the result of the multiplication has unsigned overflow, the result
3080 returned is the mathematical result modulo 2<sup>n</sup>, where n is the bit
3081 width of the result.</p>
3082
3083<p>Because LLVM integers use a two's complement representation, and the result
3084 is the same width as the operands, this instruction returns the correct
3085 result for both signed and unsigned integers. If a full product
3086 (e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands should
3087 be sign-extended or zero-extended as appropriate to the width of the full
3088 product.</p>
3089
Dan Gohman46e96012009-07-22 22:44:56 +0000[diff] [blame]3090<p><tt>nuw</tt> and <tt>nsw</tt> stand for &quot;No Unsigned Wrap&quot;
3091 and &quot;No Signed Wrap&quot;, respectively. If the <tt>nuw</tt> and/or
3092 <tt>nsw</tt> keywords are present, the result value of the <tt>mul</tt>
3093 is undefined if unsigned and/or signed overflow, respectively, occurs.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3094
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3095<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3096<pre>
3097 &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3098</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3099
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3100</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3101
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3102<!-- _______________________________________________________________________ -->
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3103<div class="doc_subsubsection">
3104 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
3105</div>
3106
3107<div class="doc_text">
3108
3109<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3110<pre>
3111 &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]3112</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3113
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3114<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3115<p>The '<tt>fmul</tt>' instruction returns the product of its two operands.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3116
3117<h5>Arguments:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3118<p>The two arguments to the '<tt>fmul</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3119 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3120 floating point values. Both arguments must have identical types.</p>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3121
3122<h5>Semantics:</h5>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3123<p>The value produced is the floating point product of the two operands.</p>
3124
3125<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3126<pre>
3127 &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]3128</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3129
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]3130</div>
3131
3132<!-- _______________________________________________________________________ -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3133<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
3134</a></div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3135
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3136<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3137
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3138<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3139<pre>
3140 &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]3141</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3142
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3143<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3144<p>The '<tt>udiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3145
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3146<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3147<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3148 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3149 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3150
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3151<h5>Semantics:</h5>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3152<p>The value produced is the unsigned integer quotient of the two operands.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3153
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3154<p>Note that unsigned integer division and signed integer division are distinct
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3155 operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
3156
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3157<p>Division by zero leads to undefined behavior.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3158
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3159<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3160<pre>
3161 &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3162</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3163
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3164</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3165
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3166<!-- _______________________________________________________________________ -->
3167<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
3168</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3169
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3170<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3171
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3172<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3173<pre>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3174 &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]3175 &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]3176</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3177
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3178<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3179<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3180
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3181<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3182<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3183 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3184 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3185
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3186<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3187<p>The value produced is the signed integer quotient of the two operands rounded
3188 towards zero.</p>
3189
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3190<p>Note that signed integer division and unsigned integer division are distinct
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3191 operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
3192
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3193<p>Division by zero leads to undefined behavior. Overflow also leads to
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3194 undefined behavior; this is a rare case, but can occur, for example, by doing
3195 a 32-bit division of -2147483648 by -1.</p>
3196
Dan Gohman67fa48e2009-07-22 00:04:19 +0000[diff] [blame]3197<p>If the <tt>exact</tt> keyword is present, the result value of the
3198 <tt>sdiv</tt> is undefined if the result would be rounded or if overflow
3199 would occur.</p>
Dan Gohmaned5fcb22009-07-20 22:41:19 +0000[diff] [blame]3200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3201<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3202<pre>
3203 &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3204</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3205
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3206</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3207
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3208<!-- _______________________________________________________________________ -->
3209<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
3210Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3211
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3212<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3213
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3214<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3215<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3216 &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]3217</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3218
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3219<h5>Overview:</h5>
3220<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3221
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3222<h5>Arguments:</h5>
3223<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3224 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3225 floating point values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3226
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3227<h5>Semantics:</h5>
3228<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3229
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3230<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3231<pre>
3232 &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]3233</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3234
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3235</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3236
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3237<!-- _______________________________________________________________________ -->
3238<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
3239</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3240
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3241<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3242
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3243<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3244<pre>
3245 &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]3246</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3247
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3248<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3249<p>The '<tt>urem</tt>' instruction returns the remainder from the unsigned
3250 division of its two arguments.</p>
3251
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3252<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3253<p>The two arguments to the '<tt>urem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3254 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3255 values. Both arguments must have identical types.</p>
3256
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3257<h5>Semantics:</h5>
3258<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3259 This instruction always performs an unsigned division to get the
3260 remainder.</p>
3261
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3262<p>Note that unsigned integer remainder and signed integer remainder are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3263 distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
3264
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3265<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3266
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3267<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3268<pre>
3269 &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3270</pre>
3271
3272</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3273
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3274<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3275<div class="doc_subsubsection">
3276 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
3277</div>
3278
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3279<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3280
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3281<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3282<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3283 &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]3284</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3285
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3286<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3287<p>The '<tt>srem</tt>' instruction returns the remainder from the signed
3288 division of its two operands. This instruction can also take
3289 <a href="#t_vector">vector</a> versions of the values in which case the
3290 elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]3291
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3292<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3293<p>The two arguments to the '<tt>srem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3294 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3295 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3296
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3297<h5>Semantics:</h5>
3298<p>This instruction returns the <i>remainder</i> of a division (where the result
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3299 has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
3300 operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
3301 a value. For more information about the difference,
3302 see <a href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
3303 Math Forum</a>. For a table of how this is implemented in various languages,
3304 please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
3305 Wikipedia: modulo operation</a>.</p>
3306
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3307<p>Note that signed integer remainder and unsigned integer remainder are
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3308 distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
3309
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]3310<p>Taking the remainder of a division by zero leads to undefined behavior.
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3311 Overflow also leads to undefined behavior; this is a rare case, but can
3312 occur, for example, by taking the remainder of a 32-bit division of
3313 -2147483648 by -1. (The remainder doesn't actually overflow, but this rule
3314 lets srem be implemented using instructions that return both the result of
3315 the division and the remainder.)</p>
3316
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3317<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3318<pre>
3319 &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3320</pre>
3321
3322</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3323
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3324<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3325<div class="doc_subsubsection">
3326 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
3327
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3328<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3329
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3330<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3331<pre>
3332 &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]3333</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3334
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3335<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3336<p>The '<tt>frem</tt>' instruction returns the remainder from the division of
3337 its two operands.</p>
3338
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3339<h5>Arguments:</h5>
3340<p>The two arguments to the '<tt>frem</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3341 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
3342 floating point values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3343
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3344<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3345<p>This instruction returns the <i>remainder</i> of a division. The remainder
3346 has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3347
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3348<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3349<pre>
3350 &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]3351</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3352
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3353</div>
3354
3355<!-- ======================================================================= -->
3356<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
3357Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3358
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3359<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3360
3361<p>Bitwise binary operators are used to do various forms of bit-twiddling in a
3362 program. They are generally very efficient instructions and can commonly be
3363 strength reduced from other instructions. They require two operands of the
3364 same type, execute an operation on them, and produce a single value. The
3365 resulting value is the same type as its operands.</p>
3366
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3367</div>
3368
3369<!-- _______________________________________________________________________ -->
3370<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
3371Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3372
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3373<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3374
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3375<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3376<pre>
3377 &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]3378</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3379
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3380<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3381<p>The '<tt>shl</tt>' instruction returns the first operand shifted to the left
3382 a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3383
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3384<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3385<p>Both arguments to the '<tt>shl</tt>' instruction must be the
3386 same <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
3387 integer type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3388
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3389<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3390<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod
3391 2<sup>n</sup>, where <tt>n</tt> is the width of the result. If <tt>op2</tt>
3392 is (statically or dynamically) negative or equal to or larger than the number
3393 of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3394 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3395 shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3396
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3397<h5>Example:</h5>
3398<pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3399 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
3400 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
3401 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3402 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3403 &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]3404</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3405
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3406</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3407
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3408<!-- _______________________________________________________________________ -->
3409<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
3410Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3411
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3412<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3413
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3414<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3415<pre>
3416 &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]3417</pre>
3418
3419<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3420<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
3421 operand shifted to the right a specified number of bits with zero fill.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3422
3423<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3424<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3425 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3426 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3427
3428<h5>Semantics:</h5>
3429<p>This instruction always performs a logical shift right operation. The most
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3430 significant bits of the result will be filled with zero bits after the shift.
3431 If <tt>op2</tt> is (statically or dynamically) equal to or larger than the
3432 number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
3433 vectors, each vector element of <tt>op1</tt> is shifted by the corresponding
3434 shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3435
3436<h5>Example:</h5>
3437<pre>
3438 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
3439 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
3440 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
3441 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3442 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3443 &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]3444</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3445
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3446</div>
3447
3448<!-- _______________________________________________________________________ -->
3449<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
3450Instruction</a> </div>
3451<div class="doc_text">
3452
3453<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3454<pre>
3455 &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]3456</pre>
3457
3458<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3459<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3460 operand shifted to the right a specified number of bits with sign
3461 extension.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3462
3463<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3464<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3465 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3466 type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3467
3468<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3469<p>This instruction always performs an arithmetic shift right operation, The
3470 most significant bits of the result will be filled with the sign bit
3471 of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
3472 larger than the number of bits in <tt>op1</tt>, the result is undefined. If
3473 the arguments are vectors, each vector element of <tt>op1</tt> is shifted by
3474 the corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3475
3476<h5>Example:</h5>
3477<pre>
3478 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3479 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3480 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3481 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3482 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3483 &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]3484</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3485
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3486</div>
3487
3488<!-- _______________________________________________________________________ -->
3489<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3490Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3491
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3492<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3493
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3494<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3495<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3496 &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]3497</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3498
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3499<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3500<p>The '<tt>and</tt>' instruction returns the bitwise logical and of its two
3501 operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3502
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3503<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3504<p>The two arguments to the '<tt>and</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3505 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3506 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3508<h5>Semantics:</h5>
3509<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3510
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3511<table border="1" cellspacing="0" cellpadding="4">
3512 <tbody>
3513 <tr>
3514 <td>In0</td>
3515 <td>In1</td>
3516 <td>Out</td>
3517 </tr>
3518 <tr>
3519 <td>0</td>
3520 <td>0</td>
3521 <td>0</td>
3522 </tr>
3523 <tr>
3524 <td>0</td>
3525 <td>1</td>
3526 <td>0</td>
3527 </tr>
3528 <tr>
3529 <td>1</td>
3530 <td>0</td>
3531 <td>0</td>
3532 </tr>
3533 <tr>
3534 <td>1</td>
3535 <td>1</td>
3536 <td>1</td>
3537 </tr>
3538 </tbody>
3539</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3540
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3541<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3542<pre>
3543 &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]3544 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3545 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3546</pre>
3547</div>
3548<!-- _______________________________________________________________________ -->
3549<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3550
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3551<div class="doc_text">
3552
3553<h5>Syntax:</h5>
3554<pre>
3555 &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
3556</pre>
3557
3558<h5>Overview:</h5>
3559<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive or of its
3560 two operands.</p>
3561
3562<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3563<p>The two arguments to the '<tt>or</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3564 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3565 values. Both arguments must have identical types.</p>
3566
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3567<h5>Semantics:</h5>
3568<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3569
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3570<table border="1" cellspacing="0" cellpadding="4">
3571 <tbody>
3572 <tr>
3573 <td>In0</td>
3574 <td>In1</td>
3575 <td>Out</td>
3576 </tr>
3577 <tr>
3578 <td>0</td>
3579 <td>0</td>
3580 <td>0</td>
3581 </tr>
3582 <tr>
3583 <td>0</td>
3584 <td>1</td>
3585 <td>1</td>
3586 </tr>
3587 <tr>
3588 <td>1</td>
3589 <td>0</td>
3590 <td>1</td>
3591 </tr>
3592 <tr>
3593 <td>1</td>
3594 <td>1</td>
3595 <td>1</td>
3596 </tr>
3597 </tbody>
3598</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3599
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3600<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3601<pre>
3602 &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3603 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3604 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3605</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3606
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3607</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3608
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3609<!-- _______________________________________________________________________ -->
3610<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3611Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3612
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3613<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3614
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3615<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3616<pre>
3617 &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]3618</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3619
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3620<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3621<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive or of
3622 its two operands. The <tt>xor</tt> is used to implement the "one's
3623 complement" operation, which is the "~" operator in C.</p>
3624
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3625<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3626<p>The two arguments to the '<tt>xor</tt>' instruction must be
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3627 <a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3628 values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3629
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3630<h5>Semantics:</h5>
3631<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3632
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3633<table border="1" cellspacing="0" cellpadding="4">
3634 <tbody>
3635 <tr>
3636 <td>In0</td>
3637 <td>In1</td>
3638 <td>Out</td>
3639 </tr>
3640 <tr>
3641 <td>0</td>
3642 <td>0</td>
3643 <td>0</td>
3644 </tr>
3645 <tr>
3646 <td>0</td>
3647 <td>1</td>
3648 <td>1</td>
3649 </tr>
3650 <tr>
3651 <td>1</td>
3652 <td>0</td>
3653 <td>1</td>
3654 </tr>
3655 <tr>
3656 <td>1</td>
3657 <td>1</td>
3658 <td>0</td>
3659 </tr>
3660 </tbody>
3661</table>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3662
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3663<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3664<pre>
3665 &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3666 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3667 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3668 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3669</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3670
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3671</div>
3672
3673<!-- ======================================================================= -->
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3674<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3675 <a name="vectorops">Vector Operations</a>
3676</div>
3677
3678<div class="doc_text">
3679
3680<p>LLVM supports several instructions to represent vector operations in a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3681 target-independent manner. These instructions cover the element-access and
3682 vector-specific operations needed to process vectors effectively. While LLVM
3683 does directly support these vector operations, many sophisticated algorithms
3684 will want to use target-specific intrinsics to take full advantage of a
3685 specific target.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3686
3687</div>
3688
3689<!-- _______________________________________________________________________ -->
3690<div class="doc_subsubsection">
3691 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3692</div>
3693
3694<div class="doc_text">
3695
3696<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3697<pre>
3698 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
3699</pre>
3700
3701<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3702<p>The '<tt>extractelement</tt>' instruction extracts a single scalar element
3703 from a vector at a specified index.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3704
3705
3706<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3707<p>The first operand of an '<tt>extractelement</tt>' instruction is a value
3708 of <a href="#t_vector">vector</a> type. The second operand is an index
3709 indicating the position from which to extract the element. The index may be
3710 a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3711
3712<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3713<p>The result is a scalar of the same type as the element type of
3714 <tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3715 <tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3716 results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3717
3718<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3719<pre>
Gabor Greifb2c73142009-10-28 13:14:50 +0000[diff] [blame]3720 &lt;result&gt; = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3721</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3722
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3723</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3724
3725<!-- _______________________________________________________________________ -->
3726<div class="doc_subsubsection">
3727 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3728</div>
3729
3730<div class="doc_text">
3731
3732<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3733<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]3734 &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]3735</pre>
3736
3737<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3738<p>The '<tt>insertelement</tt>' instruction inserts a scalar element into a
3739 vector at a specified index.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3740
3741<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3742<p>The first operand of an '<tt>insertelement</tt>' instruction is a value
3743 of <a href="#t_vector">vector</a> type. The second operand is a scalar value
3744 whose type must equal the element type of the first operand. The third
3745 operand is an index indicating the position at which to insert the value.
3746 The index may be a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3747
3748<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3749<p>The result is a vector of the same type as <tt>val</tt>. Its element values
3750 are those of <tt>val</tt> except at position <tt>idx</tt>, where it gets the
3751 value <tt>elt</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3752 results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3753
3754<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3755<pre>
Gabor Greifb2c73142009-10-28 13:14:50 +0000[diff] [blame]3756 &lt;result&gt; = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3757</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3758
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3759</div>
3760
3761<!-- _______________________________________________________________________ -->
3762<div class="doc_subsubsection">
3763 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3764</div>
3765
3766<div class="doc_text">
3767
3768<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3769<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3770 &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]3771</pre>
3772
3773<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3774<p>The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
3775 from two input vectors, returning a vector with the same element type as the
3776 input and length that is the same as the shuffle mask.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3777
3778<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3779<p>The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3780 with types that match each other. The third argument is a shuffle mask whose
3781 element type is always 'i32'. The result of the instruction is a vector
3782 whose length is the same as the shuffle mask and whose element type is the
3783 same as the element type of the first two operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3784
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3785<p>The shuffle mask operand is required to be a constant vector with either
3786 constant integer or undef values.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3787
3788<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3789<p>The elements of the two input vectors are numbered from left to right across
3790 both of the vectors. The shuffle mask operand specifies, for each element of
3791 the result vector, which element of the two input vectors the result element
3792 gets. The element selector may be undef (meaning "don't care") and the
3793 second operand may be undef if performing a shuffle from only one vector.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3794
3795<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3796<pre>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3797 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3798 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3799 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3800 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3801 &lt;result&gt; = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3802 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3803 &lt;result&gt; = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3804 &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]3805</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3806
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3807</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3808
3809<!-- ======================================================================= -->
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3810<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3811 <a name="aggregateops">Aggregate Operations</a>
3812</div>
3813
3814<div class="doc_text">
3815
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3816<p>LLVM supports several instructions for working with aggregate values.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3817
3818</div>
3819
3820<!-- _______________________________________________________________________ -->
3821<div class="doc_subsubsection">
3822 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3823</div>
3824
3825<div class="doc_text">
3826
3827<h5>Syntax:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3828<pre>
3829 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3830</pre>
3831
3832<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3833<p>The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3834 or array element from an aggregate value.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3835
3836<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3837<p>The first operand of an '<tt>extractvalue</tt>' instruction is a value
3838 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3839 operands are constant indices to specify which value to extract in a similar
3840 manner as indices in a
3841 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3842
3843<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3844<p>The result is the value at the position in the aggregate specified by the
3845 index operands.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3846
3847<h5>Example:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3848<pre>
Gabor Greifb2c73142009-10-28 13:14:50 +0000[diff] [blame]3849 &lt;result&gt; = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3850</pre>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3851
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3852</div>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3853
3854<!-- _______________________________________________________________________ -->
3855<div class="doc_subsubsection">
3856 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3857</div>
3858
3859<div class="doc_text">
3860
3861<h5>Syntax:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3862<pre>
Jeffrey Yasskine05d7732010-01-11 19:19:26 +0000[diff] [blame^]3863 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;elt&gt;, &lt;idx&gt; <i>; yields &lt;aggregate type&gt;</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3864</pre>
3865
3866<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3867<p>The '<tt>insertvalue</tt>' instruction inserts a value into a struct field or
3868 array element in an aggregate.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3869
3870
3871<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3872<p>The first operand of an '<tt>insertvalue</tt>' instruction is a value
3873 of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type. The
3874 second operand is a first-class value to insert. The following operands are
3875 constant indices indicating the position at which to insert the value in a
3876 similar manner as indices in a
3877 '<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction. The
3878 value to insert must have the same type as the value identified by the
3879 indices.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3880
3881<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3882<p>The result is an aggregate of the same type as <tt>val</tt>. Its value is
3883 that of <tt>val</tt> except that the value at the position specified by the
3884 indices is that of <tt>elt</tt>.</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3885
3886<h5>Example:</h5>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3887<pre>
Jeffrey Yasskine05d7732010-01-11 19:19:26 +0000[diff] [blame^]3888 %agg1 = insertvalue {i32, float} undef, i32 1, 0 <i>; yields {i32 1, float undef}</i>
3889 %agg2 = insertvalue {i32, float} %agg1, float %val, 1 <i>; yields {i32 1, float %val}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3890</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3891
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3892</div>
3893
3894
3895<!-- ======================================================================= -->
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]3896<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3897 <a name="memoryops">Memory Access and Addressing Operations</a>
3898</div>
3899
3900<div class="doc_text">
3901
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3902<p>A key design point of an SSA-based representation is how it represents
3903 memory. In LLVM, no memory locations are in SSA form, which makes things
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]3904 very simple. This section describes how to read, write, and allocate
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3905 memory in LLVM.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3906
3907</div>
3908
3909<!-- _______________________________________________________________________ -->
3910<div class="doc_subsubsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3911 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3912</div>
3913
3914<div class="doc_text">
3915
3916<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3917<pre>
3918 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3919</pre>
3920
3921<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3922<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3923 currently executing function, to be automatically released when this function
3924 returns to its caller. The object is always allocated in the generic address
3925 space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3926
3927<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3928<p>The '<tt>alloca</tt>' instruction
3929 allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt> bytes of memory on the
3930 runtime stack, returning a pointer of the appropriate type to the program.
3931 If "NumElements" is specified, it is the number of elements allocated,
3932 otherwise "NumElements" is defaulted to be one. If a constant alignment is
3933 specified, the value result of the allocation is guaranteed to be aligned to
3934 at least that boundary. If not specified, or if zero, the target can choose
3935 to align the allocation on any convenient boundary compatible with the
3936 type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3937
3938<p>'<tt>type</tt>' may be any sized type.</p>
3939
3940<h5>Semantics:</h5>
Bill Wendling2a454572009-05-08 20:49:29 +0000[diff] [blame]3941<p>Memory is allocated; a pointer is returned. The operation is undefined if
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3942 there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
3943 memory is automatically released when the function returns. The
3944 '<tt>alloca</tt>' instruction is commonly used to represent automatic
3945 variables that must have an address available. When the function returns
3946 (either with the <tt><a href="#i_ret">ret</a></tt>
3947 or <tt><a href="#i_unwind">unwind</a></tt> instructions), the memory is
3948 reclaimed. Allocating zero bytes is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3949
3950<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3951<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3952 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3953 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3954 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3955 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3956</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3957
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3958</div>
3959
3960<!-- _______________________________________________________________________ -->
3961<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3962Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3963
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3964<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3965
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3966<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3967<pre>
3968 &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]
3969 &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]
3970</pre>
3971
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3972<h5>Overview:</h5>
3973<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3974
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3975<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3976<p>The argument to the '<tt>load</tt>' instruction specifies the memory address
3977 from which to load. The pointer must point to
3978 a <a href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3979 marked as <tt>volatile</tt>, then the optimizer is not allowed to modify the
3980 number or order of execution of this <tt>load</tt> with other
3981 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3982 instructions. </p>
3983
3984<p>The optional constant "align" argument specifies the alignment of the
3985 operation (that is, the alignment of the memory address). A value of 0 or an
3986 omitted "align" argument means that the operation has the preferential
3987 alignment for the target. It is the responsibility of the code emitter to
3988 ensure that the alignment information is correct. Overestimating the
3989 alignment results in an undefined behavior. Underestimating the alignment may
3990 produce less efficient code. An alignment of 1 is always safe.</p>
3991
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3992<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]3993<p>The location of memory pointed to is loaded. If the value being loaded is of
3994 scalar type then the number of bytes read does not exceed the minimum number
3995 of bytes needed to hold all bits of the type. For example, loading an
3996 <tt>i24</tt> reads at most three bytes. When loading a value of a type like
3997 <tt>i20</tt> with a size that is not an integral number of bytes, the result
3998 is undefined if the value was not originally written using a store of the
3999 same type.</p>
4000
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4001<h5>Examples:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4002<pre>
4003 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
4004 <a href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4005 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
4006</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4007
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4008</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4009
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4010<!-- _______________________________________________________________________ -->
4011<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
4012Instruction</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4013
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4014<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4015
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4016<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4017<pre>
4018 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]4019 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
4020</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4021
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4022<h5>Overview:</h5>
4023<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4024
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4025<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4026<p>There are two arguments to the '<tt>store</tt>' instruction: a value to store
4027 and an address at which to store it. The type of the
4028 '<tt>&lt;pointer&gt;</tt>' operand must be a pointer to
4029 the <a href="#t_firstclass">first class</a> type of the
4030 '<tt>&lt;value&gt;</tt>' operand. If the <tt>store</tt> is marked
4031 as <tt>volatile</tt>, then the optimizer is not allowed to modify the number
4032 or order of execution of this <tt>store</tt> with other
4033 volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
4034 instructions.</p>
4035
4036<p>The optional constant "align" argument specifies the alignment of the
4037 operation (that is, the alignment of the memory address). A value of 0 or an
4038 omitted "align" argument means that the operation has the preferential
4039 alignment for the target. It is the responsibility of the code emitter to
4040 ensure that the alignment information is correct. Overestimating the
4041 alignment results in an undefined behavior. Underestimating the alignment may
4042 produce less efficient code. An alignment of 1 is always safe.</p>
4043
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4044<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4045<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>' at the
4046 location specified by the '<tt>&lt;pointer&gt;</tt>' operand. If
4047 '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes written
4048 does not exceed the minimum number of bytes needed to hold all bits of the
4049 type. For example, storing an <tt>i24</tt> writes at most three bytes. When
4050 writing a value of a type like <tt>i20</tt> with a size that is not an
4051 integral number of bytes, it is unspecified what happens to the extra bits
4052 that do not belong to the type, but they will typically be overwritten.</p>
4053
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4054<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4055<pre>
4056 %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]4057 store i32 3, i32* %ptr <i>; yields {void}</i>
4058 %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]4059</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4060
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4061</div>
4062
4063<!-- _______________________________________________________________________ -->
4064<div class="doc_subsubsection">
4065 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
4066</div>
4067
4068<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4069
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4070<h5>Syntax:</h5>
4071<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4072 &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]4073 &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]4074</pre>
4075
4076<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4077<p>The '<tt>getelementptr</tt>' instruction is used to get the address of a
4078 subelement of an aggregate data structure. It performs address calculation
4079 only and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4080
4081<h5>Arguments:</h5>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4082<p>The first argument is always a pointer, and forms the basis of the
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4083 calculation. The remaining arguments are indices that indicate which of the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4084 elements of the aggregate object are indexed. The interpretation of each
4085 index is dependent on the type being indexed into. The first index always
4086 indexes the pointer value given as the first argument, the second index
4087 indexes a value of the type pointed to (not necessarily the value directly
4088 pointed to, since the first index can be non-zero), etc. The first type
4089 indexed into must be a pointer value, subsequent types can be arrays, vectors
4090 and structs. Note that subsequent types being indexed into can never be
4091 pointers, since that would require loading the pointer before continuing
4092 calculation.</p>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4093
4094<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]4095 When indexing into a (optionally packed) structure, only <tt>i32</tt> integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4096 <b>constants</b> are allowed. When indexing into an array, pointer or
Chris Lattnere92fc832009-07-29 06:44:13 +0000[diff] [blame]4097 vector, integers of any width are allowed, and they are not required to be
4098 constant.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4099
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4100<p>For example, let's consider a C code fragment and how it gets compiled to
4101 LLVM:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4102
4103<div class="doc_code">
4104<pre>
4105struct RT {
4106 char A;
4107 int B[10][20];
4108 char C;
4109};
4110struct ST {
4111 int X;
4112 double Y;
4113 struct RT Z;
4114};
4115
4116int *foo(struct ST *s) {
4117 return &amp;s[1].Z.B[5][13];
4118}
4119</pre>
4120</div>
4121
4122<p>The LLVM code generated by the GCC frontend is:</p>
4123
4124<div class="doc_code">
4125<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]4126%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
4127%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4128
Dan Gohman47360842009-07-25 02:23:48 +0000[diff] [blame]4129define i32* @foo(%ST* %s) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4130entry:
4131 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
4132 ret i32* %reg
4133}
4134</pre>
4135</div>
4136
4137<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4138<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]4139 type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
4140 }</tt>' type, a structure. The second index indexes into the third element
4141 of the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
4142 i8 }</tt>' type, another structure. The third index indexes into the second
4143 element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
4144 array. The two dimensions of the array are subscripted into, yielding an
4145 '<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a
4146 pointer to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4147
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4148<p>Note that it is perfectly legal to index partially through a structure,
4149 returning a pointer to an inner element. Because of this, the LLVM code for
4150 the given testcase is equivalent to:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4151
4152<pre>
Dan Gohman47360842009-07-25 02:23:48 +0000[diff] [blame]4153 define i32* @foo(%ST* %s) {
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4154 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
4155 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
4156 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
4157 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
4158 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
4159 ret i32* %t5
4160 }
4161</pre>
4162
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4163<p>If the <tt>inbounds</tt> keyword is present, the result value of the
Dan Gohman4cba1562009-07-29 16:00:30 +0000[diff] [blame]4164 <tt>getelementptr</tt> is undefined if the base pointer is not an
4165 <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]4166 that would be formed by successive addition of the offsets implied by the
4167 indices to the base address with infinitely precise arithmetic are not an
4168 <i>in bounds</i> address of that allocated object.
Dan Gohman4cba1562009-07-29 16:00:30 +0000[diff] [blame]4169 The <i>in bounds</i> addresses for an allocated object are all the addresses
Dan Gohmanbc4c87f2009-08-20 17:08:17 +0000[diff] [blame]4170 that point into the object, plus the address one byte past the end.</p>
Dan Gohman106b2ae2009-07-27 21:53:46 +0000[diff] [blame]4171
4172<p>If the <tt>inbounds</tt> keyword is not present, the offsets are added to
4173 the base address with silently-wrapping two's complement arithmetic, and
4174 the result value of the <tt>getelementptr</tt> may be outside the object
4175 pointed to by the base pointer. The result value may not necessarily be
4176 used to access memory though, even if it happens to point into allocated
4177 storage. See the <a href="#pointeraliasing">Pointer Aliasing Rules</a>
4178 section for more information.</p>
4179
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4180<p>The getelementptr instruction is often confusing. For some more insight into
4181 how it works, see <a href="GetElementPtr.html">the getelementptr FAQ</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4182
4183<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4184<pre>
4185 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]4186 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
4187 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4188 %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]4189 <i>; yields i8*:eptr</i>
4190 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta4f9a0dc2009-04-25 07:27:44 +0000[diff] [blame]4191 <i>; yields i32*:iptr</i>
Sanjiv Gupta1e46c582009-04-24 16:38:13 +0000[diff] [blame]4192 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4193</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4194
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4195</div>
4196
4197<!-- ======================================================================= -->
4198<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
4199</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4200
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4201<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4202
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4203<p>The instructions in this category are the conversion instructions (casting)
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4204 which all take a single operand and a type. They perform various bit
4205 conversions on the operand.</p>
4206
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4207</div>
4208
4209<!-- _______________________________________________________________________ -->
4210<div class="doc_subsubsection">
4211 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
4212</div>
4213<div class="doc_text">
4214
4215<h5>Syntax:</h5>
4216<pre>
4217 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4218</pre>
4219
4220<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4221<p>The '<tt>trunc</tt>' instruction truncates its operand to the
4222 type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4223
4224<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4225<p>The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
4226 be an <a href="#t_integer">integer</a> type, and a type that specifies the
4227 size and type of the result, which must be
4228 an <a href="#t_integer">integer</a> type. The bit size of <tt>value</tt> must
4229 be larger than the bit size of <tt>ty2</tt>. Equal sized types are not
4230 allowed.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4231
4232<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4233<p>The '<tt>trunc</tt>' instruction truncates the high order bits
4234 in <tt>value</tt> and converts the remaining bits to <tt>ty2</tt>. Since the
4235 source size must be larger than the destination size, <tt>trunc</tt> cannot
4236 be a <i>no-op cast</i>. It will always truncate bits.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4237
4238<h5>Example:</h5>
4239<pre>
4240 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
4241 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4242 %Z = trunc i32 122 to i1 <i>; yields i1:false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4243</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4244
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4245</div>
4246
4247<!-- _______________________________________________________________________ -->
4248<div class="doc_subsubsection">
4249 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
4250</div>
4251<div class="doc_text">
4252
4253<h5>Syntax:</h5>
4254<pre>
4255 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4256</pre>
4257
4258<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4259<p>The '<tt>zext</tt>' instruction zero extends its operand to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4260 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4261
4262
4263<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4264<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]4265 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4266 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4267 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4268 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4269
4270<h5>Semantics:</h5>
4271<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]4272 bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4273
4274<p>When zero extending from i1, the result will always be either 0 or 1.</p>
4275
4276<h5>Example:</h5>
4277<pre>
4278 %X = zext i32 257 to i64 <i>; yields i64:257</i>
4279 %Y = zext i1 true to i32 <i>; yields i32:1</i>
4280</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4281
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4282</div>
4283
4284<!-- _______________________________________________________________________ -->
4285<div class="doc_subsubsection">
4286 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
4287</div>
4288<div class="doc_text">
4289
4290<h5>Syntax:</h5>
4291<pre>
4292 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4293</pre>
4294
4295<h5>Overview:</h5>
4296<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
4297
4298<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4299<p>The '<tt>sext</tt>' instruction takes a value to cast, which must be of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4300 <a href="#t_integer">integer</a> type, and a type to cast it to, which must
4301 also be of <a href="#t_integer">integer</a> type. The bit size of the
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4302 <tt>value</tt> must be smaller than the bit size of the destination type,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4303 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4304
4305<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4306<p>The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
4307 bit (highest order bit) of the <tt>value</tt> until it reaches the bit size
4308 of the type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4309
4310<p>When sign extending from i1, the extension always results in -1 or 0.</p>
4311
4312<h5>Example:</h5>
4313<pre>
4314 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
4315 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
4316</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4317
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4318</div>
4319
4320<!-- _______________________________________________________________________ -->
4321<div class="doc_subsubsection">
4322 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
4323</div>
4324
4325<div class="doc_text">
4326
4327<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4328<pre>
4329 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4330</pre>
4331
4332<h5>Overview:</h5>
4333<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4334 <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4335
4336<h5>Arguments:</h5>
4337<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4338 point</a> value to cast and a <a href="#t_floating">floating point</a> type
4339 to cast it to. The size of <tt>value</tt> must be larger than the size of
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4340 <tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4341 <i>no-op cast</i>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4342
4343<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4344<p>The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4345 <a href="#t_floating">floating point</a> type to a smaller
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4346 <a href="#t_floating">floating point</a> type. If the value cannot fit
4347 within the destination type, <tt>ty2</tt>, then the results are
4348 undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4349
4350<h5>Example:</h5>
4351<pre>
4352 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4353 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4354</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4355
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4356</div>
4357
4358<!-- _______________________________________________________________________ -->
4359<div class="doc_subsubsection">
4360 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4361</div>
4362<div class="doc_text">
4363
4364<h5>Syntax:</h5>
4365<pre>
4366 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4367</pre>
4368
4369<h5>Overview:</h5>
4370<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]4371 floating point value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4372
4373<h5>Arguments:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4374<p>The '<tt>fpext</tt>' instruction takes a
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4375 <a href="#t_floating">floating point</a> <tt>value</tt> to cast, and
4376 a <a href="#t_floating">floating point</a> type to cast it to. The source
4377 type must be smaller than the destination type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4378
4379<h5>Semantics:</h5>
4380<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4381 <a href="#t_floating">floating point</a> type to a larger
4382 <a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4383 used to make a <i>no-op cast</i> because it always changes bits. Use
4384 <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]4385
4386<h5>Example:</h5>
4387<pre>
4388 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4389 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4390</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4391
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4392</div>
4393
4394<!-- _______________________________________________________________________ -->
4395<div class="doc_subsubsection">
4396 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4397</div>
4398<div class="doc_text">
4399
4400<h5>Syntax:</h5>
4401<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4402 &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]4403</pre>
4404
4405<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4406<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4407 unsigned integer equivalent of type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4408
4409<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4410<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
4411 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4412 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4413 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4414 vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4415
4416<h5>Semantics:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4417<p>The '<tt>fptoui</tt>' instruction converts its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4418 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4419 towards zero) unsigned integer value. If the value cannot fit
4420 in <tt>ty2</tt>, the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4421
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4422<h5>Example:</h5>
4423<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4424 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4425 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4426 %Z = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4427</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4428
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4429</div>
4430
4431<!-- _______________________________________________________________________ -->
4432<div class="doc_subsubsection">
4433 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4434</div>
4435<div class="doc_text">
4436
4437<h5>Syntax:</h5>
4438<pre>
4439 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4440</pre>
4441
4442<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4443<p>The '<tt>fptosi</tt>' instruction converts
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4444 <a href="#t_floating">floating point</a> <tt>value</tt> to
4445 type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4446
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4447<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4448<p>The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
4449 scalar or vector <a href="#t_floating">floating point</a> value, and a type
4450 to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4451 type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4452 vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4453
4454<h5>Semantics:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4455<p>The '<tt>fptosi</tt>' instruction converts its
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4456 <a href="#t_floating">floating point</a> operand into the nearest (rounding
4457 towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4458 the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4459
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4460<h5>Example:</h5>
4461<pre>
4462 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4463 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4464 %Z = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4465</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4466
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4467</div>
4468
4469<!-- _______________________________________________________________________ -->
4470<div class="doc_subsubsection">
4471 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4472</div>
4473<div class="doc_text">
4474
4475<h5>Syntax:</h5>
4476<pre>
4477 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4478</pre>
4479
4480<h5>Overview:</h5>
4481<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4482 integer and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4483
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4484<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4485<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]4486 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4487 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4488 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4489 floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4490
4491<h5>Semantics:</h5>
4492<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4493 integer quantity and converts it to the corresponding floating point
4494 value. If the value cannot fit in the floating point value, the results are
4495 undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4496
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4497<h5>Example:</h5>
4498<pre>
4499 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4500 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4501</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4502
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4503</div>
4504
4505<!-- _______________________________________________________________________ -->
4506<div class="doc_subsubsection">
4507 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4508</div>
4509<div class="doc_text">
4510
4511<h5>Syntax:</h5>
4512<pre>
4513 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4514</pre>
4515
4516<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4517<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed integer
4518 and converts that value to the <tt>ty2</tt> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4519
4520<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4521<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]4522 scalar or vector <a href="#t_integer">integer</a> value, and a type to cast
4523 it to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4524 type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4525 floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4526
4527<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4528<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed integer
4529 quantity and converts it to the corresponding floating point value. If the
4530 value cannot fit in the floating point value, the results are undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4531
4532<h5>Example:</h5>
4533<pre>
4534 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4535 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4536</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4537
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4538</div>
4539
4540<!-- _______________________________________________________________________ -->
4541<div class="doc_subsubsection">
4542 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4543</div>
4544<div class="doc_text">
4545
4546<h5>Syntax:</h5>
4547<pre>
4548 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4549</pre>
4550
4551<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4552<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4553 the integer type <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4554
4555<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4556<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4557 must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
4558 <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]4559
4560<h5>Semantics:</h5>
4561<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4562 <tt>ty2</tt> by interpreting the pointer value as an integer and either
4563 truncating or zero extending that value to the size of the integer type. If
4564 <tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4565 <tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4566 are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4567 change.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4568
4569<h5>Example:</h5>
4570<pre>
4571 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4572 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4573</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4574
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4575</div>
4576
4577<!-- _______________________________________________________________________ -->
4578<div class="doc_subsubsection">
4579 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4580</div>
4581<div class="doc_text">
4582
4583<h5>Syntax:</h5>
4584<pre>
4585 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4586</pre>
4587
4588<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4589<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to a
4590 pointer type, <tt>ty2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4591
4592<h5>Arguments:</h5>
4593<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4594 value to cast, and a type to cast it to, which must be a
4595 <a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4596
4597<h5>Semantics:</h5>
4598<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4599 <tt>ty2</tt> by applying either a zero extension or a truncation depending on
4600 the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4601 size of a pointer then a truncation is done. If <tt>value</tt> is smaller
4602 than the size of a pointer then a zero extension is done. If they are the
4603 same size, nothing is done (<i>no-op cast</i>).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4604
4605<h5>Example:</h5>
4606<pre>
4607 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
Gabor Greif463c9342009-10-28 09:21:30 +0000[diff] [blame]4608 %Y = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4609 %Z = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4610</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4611
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4612</div>
4613
4614<!-- _______________________________________________________________________ -->
4615<div class="doc_subsubsection">
4616 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4617</div>
4618<div class="doc_text">
4619
4620<h5>Syntax:</h5>
4621<pre>
4622 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4623</pre>
4624
4625<h5>Overview:</h5>
4626<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4627 <tt>ty2</tt> without changing any bits.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4628
4629<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4630<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be a
4631 non-aggregate first class value, and a type to cast it to, which must also be
4632 a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes
4633 of <tt>value</tt> and the destination type, <tt>ty2</tt>, must be
4634 identical. If the source type is a pointer, the destination type must also be
4635 a pointer. This instruction supports bitwise conversion of vectors to
4636 integers and to vectors of other types (as long as they have the same
4637 size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4638
4639<h5>Semantics:</h5>
4640<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4641 <tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4642 this conversion. The conversion is done as if the <tt>value</tt> had been
4643 stored to memory and read back as type <tt>ty2</tt>. Pointer types may only
4644 be converted to other pointer types with this instruction. To convert
4645 pointers to other types, use the <a href="#i_inttoptr">inttoptr</a> or
4646 <a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4647
4648<h5>Example:</h5>
4649<pre>
4650 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4651 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4652 %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]4653</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4654
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4655</div>
4656
4657<!-- ======================================================================= -->
4658<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4659
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4660<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4661
4662<p>The instructions in this category are the "miscellaneous" instructions, which
4663 defy better classification.</p>
4664
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4665</div>
4666
4667<!-- _______________________________________________________________________ -->
4668<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4669</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4670
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4671<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4672
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4673<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4674<pre>
4675 &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]4676</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4677
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4678<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4679<p>The '<tt>icmp</tt>' instruction returns a boolean value or a vector of
4680 boolean values based on comparison of its two integer, integer vector, or
4681 pointer operands.</p>
4682
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4683<h5>Arguments:</h5>
4684<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4685 the condition code indicating the kind of comparison to perform. It is not a
4686 value, just a keyword. The possible condition code are:</p>
4687
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4688<ol>
4689 <li><tt>eq</tt>: equal</li>
4690 <li><tt>ne</tt>: not equal </li>
4691 <li><tt>ugt</tt>: unsigned greater than</li>
4692 <li><tt>uge</tt>: unsigned greater or equal</li>
4693 <li><tt>ult</tt>: unsigned less than</li>
4694 <li><tt>ule</tt>: unsigned less or equal</li>
4695 <li><tt>sgt</tt>: signed greater than</li>
4696 <li><tt>sge</tt>: signed greater or equal</li>
4697 <li><tt>slt</tt>: signed less than</li>
4698 <li><tt>sle</tt>: signed less or equal</li>
4699</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4700
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4701<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4702 <a href="#t_pointer">pointer</a> or integer <a href="#t_vector">vector</a>
4703 typed. They must also be identical types.</p>
4704
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4705<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4706<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to the
4707 condition code given as <tt>cond</tt>. The comparison performed always yields
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]4708 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]4709 result, as follows:</p>
4710
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4711<ol>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4712 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4713 <tt>false</tt> otherwise. No sign interpretation is necessary or
4714 performed.</li>
4715
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4716 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4717 <tt>false</tt> otherwise. No sign interpretation is necessary or
4718 performed.</li>
4719
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4720 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4721 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4722
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4723 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4724 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4725 to <tt>op2</tt>.</li>
4726
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4727 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4728 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4729
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4730 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4731 <tt>true</tt> if <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4732
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4733 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4734 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4735
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4736 <li><tt>sge</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4737 <tt>true</tt> if <tt>op1</tt> is greater than or equal
4738 to <tt>op2</tt>.</li>
4739
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4740 <li><tt>slt</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4741 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
4742
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4743 <li><tt>sle</tt>: interprets the operands as signed values and yields
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4744 <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]4745</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4746
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4747<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4748 values are compared as if they were integers.</p>
4749
4750<p>If the operands are integer vectors, then they are compared element by
4751 element. The result is an <tt>i1</tt> vector with the same number of elements
4752 as the values being compared. Otherwise, the result is an <tt>i1</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4753
4754<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4755<pre>
4756 &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4757 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4758 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4759 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4760 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4761 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4762</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4763
4764<p>Note that the code generator does not yet support vector types with
4765 the <tt>icmp</tt> instruction.</p>
4766
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4767</div>
4768
4769<!-- _______________________________________________________________________ -->
4770<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4771</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4772
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4773<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4774
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4775<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4776<pre>
4777 &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]4778</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4779
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4780<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4781<p>The '<tt>fcmp</tt>' instruction returns a boolean value or vector of boolean
4782 values based on comparison of its operands.</p>
4783
4784<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]4785(<a href="#t_integer"><tt>i1</tt></a>).</p>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4786
4787<p>If the operands are floating point vectors, then the result type is a vector
4788 of boolean with the same number of elements as the operands being
4789 compared.</p>
4790
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4791<h5>Arguments:</h5>
4792<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4793 the condition code indicating the kind of comparison to perform. It is not a
4794 value, just a keyword. The possible condition code are:</p>
4795
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4796<ol>
4797 <li><tt>false</tt>: no comparison, always returns false</li>
4798 <li><tt>oeq</tt>: ordered and equal</li>
4799 <li><tt>ogt</tt>: ordered and greater than </li>
4800 <li><tt>oge</tt>: ordered and greater than or equal</li>
4801 <li><tt>olt</tt>: ordered and less than </li>
4802 <li><tt>ole</tt>: ordered and less than or equal</li>
4803 <li><tt>one</tt>: ordered and not equal</li>
4804 <li><tt>ord</tt>: ordered (no nans)</li>
4805 <li><tt>ueq</tt>: unordered or equal</li>
4806 <li><tt>ugt</tt>: unordered or greater than </li>
4807 <li><tt>uge</tt>: unordered or greater than or equal</li>
4808 <li><tt>ult</tt>: unordered or less than </li>
4809 <li><tt>ule</tt>: unordered or less than or equal</li>
4810 <li><tt>une</tt>: unordered or not equal</li>
4811 <li><tt>uno</tt>: unordered (either nans)</li>
4812 <li><tt>true</tt>: no comparison, always returns true</li>
4813</ol>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4814
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4815<p><i>Ordered</i> means that neither operand is a QNAN while
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4816 <i>unordered</i> means that either operand may be a QNAN.</p>
4817
4818<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be either
4819 a <a href="#t_floating">floating point</a> type or
4820 a <a href="#t_vector">vector</a> of floating point type. They must have
4821 identical types.</p>
4822
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4823<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4824<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4825 according to the condition code given as <tt>cond</tt>. If the operands are
4826 vectors, then the vectors are compared element by element. Each comparison
Nick Lewycky244cf482009-09-27 00:45:11 +0000[diff] [blame]4827 performed always yields an <a href="#t_integer">i1</a> result, as
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4828 follows:</p>
4829
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4830<ol>
4831 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4832
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4833 <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]4834 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4835
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4836 <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]4837 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
4838
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4839 <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]4840 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4841
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4842 <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]4843 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4844
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4845 <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]4846 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4847
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4848 <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]4849 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4850
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4851 <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]4852
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4853 <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]4854 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
4855
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4856 <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]4857 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
4858
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4859 <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]4860 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
4861
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4862 <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]4863 <tt>op1</tt> is less than <tt>op2</tt>.</li>
4864
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4865 <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]4866 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
4867
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]4868 <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]4869 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
4870
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4871 <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]4872
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4873 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4874</ol>
4875
4876<h5>Example:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4877<pre>
4878 &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]4879 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4880 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4881 &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]4882</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4883
4884<p>Note that the code generator does not yet support vector types with
4885 the <tt>fcmp</tt> instruction.</p>
4886
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4887</div>
4888
4889<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4890<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4891 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4892</div>
4893
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4894<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4895
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4896<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4897<pre>
4898 &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...
4899</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4900
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4901<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4902<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in the
4903 SSA graph representing the function.</p>
4904
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4905<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4906<p>The type of the incoming values is specified with the first type field. After
4907 this, the '<tt>phi</tt>' instruction takes a list of pairs as arguments, with
4908 one pair for each predecessor basic block of the current block. Only values
4909 of <a href="#t_firstclass">first class</a> type may be used as the value
4910 arguments to the PHI node. Only labels may be used as the label
4911 arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4912
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4913<p>There must be no non-phi instructions between the start of a basic block and
4914 the PHI instructions: i.e. PHI instructions must be first in a basic
4915 block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4916
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4917<p>For the purposes of the SSA form, the use of each incoming value is deemed to
4918 occur on the edge from the corresponding predecessor block to the current
4919 block (but after any definition of an '<tt>invoke</tt>' instruction's return
4920 value on the same edge).</p>
Jay Foad8e2fd2c2009-06-03 10:20:10 +0000[diff] [blame]4921
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4922<h5>Semantics:</h5>
4923<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4924 specified by the pair corresponding to the predecessor basic block that
4925 executed just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4926
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4927<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4928<pre>
4929Loop: ; Infinite loop that counts from 0 on up...
4930 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4931 %nextindvar = add i32 %indvar, 1
4932 br label %Loop
4933</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4934
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4935</div>
4936
4937<!-- _______________________________________________________________________ -->
4938<div class="doc_subsubsection">
4939 <a name="i_select">'<tt>select</tt>' Instruction</a>
4940</div>
4941
4942<div class="doc_text">
4943
4944<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4945<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4946 &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>
4947
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4948 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4949</pre>
4950
4951<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4952<p>The '<tt>select</tt>' instruction is used to choose one value based on a
4953 condition, without branching.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4954
4955
4956<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4957<p>The '<tt>select</tt>' instruction requires an 'i1' value or a vector of 'i1'
4958 values indicating the condition, and two values of the
4959 same <a href="#t_firstclass">first class</a> type. If the val1/val2 are
4960 vectors and the condition is a scalar, then entire vectors are selected, not
4961 individual elements.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4962
4963<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4964<p>If the condition is an i1 and it evaluates to 1, the instruction returns the
4965 first value argument; otherwise, it returns the second value argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4966
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]4967<p>If the condition is a vector of i1, then the value arguments must be vectors
4968 of the same size, and the selection is done element by element.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4969
4970<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4971<pre>
4972 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4973</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4974
4975<p>Note that the code generator does not yet support conditions
4976 with vector type.</p>
4977
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4978</div>
4979
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4980<!-- _______________________________________________________________________ -->
4981<div class="doc_subsubsection">
4982 <a name="i_call">'<tt>call</tt>' Instruction</a>
4983</div>
4984
4985<div class="doc_text">
4986
4987<h5>Syntax:</h5>
4988<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4989 &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]4990</pre>
4991
4992<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4993<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4994
4995<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4996<p>This instruction requires several arguments:</p>
4997
4998<ol>
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]4999 <li>The optional "tail" marker indicates that the callee function does not
5000 access any allocas or varargs in the caller. Note that calls may be
5001 marked "tail" even if they do not occur before
5002 a <a href="#i_ret"><tt>ret</tt></a> instruction. If the "tail" marker is
5003 present, the function call is eligible for tail call optimization,
5004 but <a href="CodeGenerator.html#tailcallopt">might not in fact be
5005 optimized into a jump</a>. As of this writing, the extra requirements for
5006 a call to actually be optimized are:
5007 <ul>
5008 <li>Caller and callee both have the calling
5009 convention <tt>fastcc</tt>.</li>
5010 <li>The call is in tail position (ret immediately follows call and ret
5011 uses value of call or is void).</li>
5012 <li>Option <tt>-tailcallopt</tt> is enabled,
5013 or <code>llvm::PerformTailCallOpt</code> is <code>true</code>.</li>
5014 <li><a href="CodeGenerator.html#tailcallopt">Platform specific
5015 constraints are met.</a></li>
5016 </ul>
5017 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5018
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5019 <li>The optional "cconv" marker indicates which <a href="#callingconv">calling
5020 convention</a> the call should use. If none is specified, the call
Jeffrey Yasskin3e1a3a02010-01-09 19:44:16 +0000[diff] [blame]5021 defaults to using C calling conventions. The calling convention of the
5022 call must match the calling convention of the target function, or else the
5023 behavior is undefined.</li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5024
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5025 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
5026 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>', and
5027 '<tt>inreg</tt>' attributes are valid here.</li>
5028
5029 <li>'<tt>ty</tt>': the type of the call instruction itself which is also the
5030 type of the return value. Functions that return no value are marked
5031 <tt><a href="#t_void">void</a></tt>.</li>
5032
5033 <li>'<tt>fnty</tt>': shall be the signature of the pointer to function value
5034 being invoked. The argument types must match the types implied by this
5035 signature. This type can be omitted if the function is not varargs and if
5036 the function type does not return a pointer to a function.</li>
5037
5038 <li>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
5039 be invoked. In most cases, this is a direct function invocation, but
5040 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
5041 to function value.</li>
5042
5043 <li>'<tt>function args</tt>': argument list whose types match the function
5044 signature argument types. All arguments must be of
5045 <a href="#t_firstclass">first class</a> type. If the function signature
5046 indicates the function accepts a variable number of arguments, the extra
5047 arguments can be specified.</li>
5048
5049 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
5050 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
5051 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5052</ol>
5053
5054<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5055<p>The '<tt>call</tt>' instruction is used to cause control flow to transfer to
5056 a specified function, with its incoming arguments bound to the specified
5057 values. Upon a '<tt><a href="#i_ret">ret</a></tt>' instruction in the called
5058 function, control flow continues with the instruction after the function
5059 call, and the return value of the function is bound to the result
5060 argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5061
5062<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5063<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]5064 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]5065 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
5066 %X = tail call i32 @foo() <i>; yields i32</i>
5067 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
5068 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]5069
5070 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]5071 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]5072 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
5073 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]5074 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]5075 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5076</pre>
5077
Dale Johannesen9fcf4a92009-09-24 18:38:21 +0000[diff] [blame]5078<p>llvm treats calls to some functions with names and arguments that match the
Dale Johannesen4730cd32009-09-25 17:04:42 +0000[diff] [blame]5079standard C99 library as being the C99 library functions, and may perform
5080optimizations or generate code for them under that assumption. This is
5081something we'd like to change in the future to provide better support for
5082freestanding environments and non-C-based langauges.</p>
Dale Johannesen9fcf4a92009-09-24 18:38:21 +0000[diff] [blame]5083
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5084</div>
5085
5086<!-- _______________________________________________________________________ -->
5087<div class="doc_subsubsection">
5088 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
5089</div>
5090
5091<div class="doc_text">
5092
5093<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5094<pre>
5095 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
5096</pre>
5097
5098<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5099<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5100 the "variable argument" area of a function call. It is used to implement the
5101 <tt>va_arg</tt> macro in C.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5102
5103<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5104<p>This instruction takes a <tt>va_list*</tt> value and the type of the
5105 argument. It returns a value of the specified argument type and increments
5106 the <tt>va_list</tt> to point to the next argument. The actual type
5107 of <tt>va_list</tt> is target specific.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5108
5109<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5110<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified type
5111 from the specified <tt>va_list</tt> and causes the <tt>va_list</tt> to point
5112 to the next argument. For more information, see the variable argument
5113 handling <a href="#int_varargs">Intrinsic Functions</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5114
5115<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]5116 take a variable number of arguments, for example, the <tt>vfprintf</tt>
5117 function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5118
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5119<p><tt>va_arg</tt> is an LLVM instruction instead of
5120 an <a href="#intrinsics">intrinsic function</a> because it takes a type as an
5121 argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5122
5123<h5>Example:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5124<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
5125
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5126<p>Note that the code generator does not yet fully support va_arg on many
5127 targets. Also, it does not currently support va_arg with aggregate types on
5128 any target.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]5129
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5130</div>
5131
5132<!-- *********************************************************************** -->
5133<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
5134<!-- *********************************************************************** -->
5135
5136<div class="doc_text">
5137
5138<p>LLVM supports the notion of an "intrinsic function". These functions have
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5139 well known names and semantics and are required to follow certain
5140 restrictions. Overall, these intrinsics represent an extension mechanism for
5141 the LLVM language that does not require changing all of the transformations
5142 in LLVM when adding to the language (or the bitcode reader/writer, the
5143 parser, etc...).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5144
5145<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5146 prefix is reserved in LLVM for intrinsic names; thus, function names may not
5147 begin with this prefix. Intrinsic functions must always be external
5148 functions: you cannot define the body of intrinsic functions. Intrinsic
5149 functions may only be used in call or invoke instructions: it is illegal to
5150 take the address of an intrinsic function. Additionally, because intrinsic
5151 functions are part of the LLVM language, it is required if any are added that
5152 they be documented here.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5153
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5154<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents a
5155 family of functions that perform the same operation but on different data
5156 types. Because LLVM can represent over 8 million different integer types,
5157 overloading is used commonly to allow an intrinsic function to operate on any
5158 integer type. One or more of the argument types or the result type can be
5159 overloaded to accept any integer type. Argument types may also be defined as
5160 exactly matching a previous argument's type or the result type. This allows
5161 an intrinsic function which accepts multiple arguments, but needs all of them
5162 to be of the same type, to only be overloaded with respect to a single
5163 argument or the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5164
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5165<p>Overloaded intrinsics will have the names of its overloaded argument types
5166 encoded into its function name, each preceded by a period. Only those types
5167 which are overloaded result in a name suffix. Arguments whose type is matched
5168 against another type do not. For example, the <tt>llvm.ctpop</tt> function
5169 can take an integer of any width and returns an integer of exactly the same
5170 integer width. This leads to a family of functions such as
5171 <tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29
5172 %val)</tt>. Only one type, the return type, is overloaded, and only one type
5173 suffix is required. Because the argument's type is matched against the return
5174 type, it does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5175
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]5176<p>To learn how to add an intrinsic function, please see the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5177 <a href="ExtendingLLVM.html">Extending LLVM Guide</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5178
5179</div>
5180
5181<!-- ======================================================================= -->
5182<div class="doc_subsection">
5183 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
5184</div>
5185
5186<div class="doc_text">
5187
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5188<p>Variable argument support is defined in LLVM with
5189 the <a href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
5190 intrinsic functions. These functions are related to the similarly named
5191 macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5192
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5193<p>All of these functions operate on arguments that use a target-specific value
5194 type "<tt>va_list</tt>". The LLVM assembly language reference manual does
5195 not define what this type is, so all transformations should be prepared to
5196 handle these functions regardless of the type used.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5197
5198<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]5199 instruction and the variable argument handling intrinsic functions are
5200 used.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5201
5202<div class="doc_code">
5203<pre>
5204define i32 @test(i32 %X, ...) {
5205 ; Initialize variable argument processing
5206 %ap = alloca i8*
5207 %ap2 = bitcast i8** %ap to i8*
5208 call void @llvm.va_start(i8* %ap2)
5209
5210 ; Read a single integer argument
5211 %tmp = va_arg i8** %ap, i32
5212
5213 ; Demonstrate usage of llvm.va_copy and llvm.va_end
5214 %aq = alloca i8*
5215 %aq2 = bitcast i8** %aq to i8*
5216 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
5217 call void @llvm.va_end(i8* %aq2)
5218
5219 ; Stop processing of arguments.
5220 call void @llvm.va_end(i8* %ap2)
5221 ret i32 %tmp
5222}
5223
5224declare void @llvm.va_start(i8*)
5225declare void @llvm.va_copy(i8*, i8*)
5226declare void @llvm.va_end(i8*)
5227</pre>
5228</div>
5229
5230</div>
5231
5232<!-- _______________________________________________________________________ -->
5233<div class="doc_subsubsection">
5234 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
5235</div>
5236
5237
5238<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5239
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5240<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5241<pre>
5242 declare void %llvm.va_start(i8* &lt;arglist&gt;)
5243</pre>
5244
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5245<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5246<p>The '<tt>llvm.va_start</tt>' intrinsic initializes <tt>*&lt;arglist&gt;</tt>
5247 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]5248
5249<h5>Arguments:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5250<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]5251
5252<h5>Semantics:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5253<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]5254 macro available in C. In a target-dependent way, it initializes
5255 the <tt>va_list</tt> element to which the argument points, so that the next
5256 call to <tt>va_arg</tt> will produce the first variable argument passed to
5257 the function. Unlike the C <tt>va_start</tt> macro, this intrinsic does not
5258 need to know the last argument of the function as the compiler can figure
5259 that out.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5260
5261</div>
5262
5263<!-- _______________________________________________________________________ -->
5264<div class="doc_subsubsection">
5265 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
5266</div>
5267
5268<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5269
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5270<h5>Syntax:</h5>
5271<pre>
5272 declare void @llvm.va_end(i8* &lt;arglist&gt;)
5273</pre>
5274
5275<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5276<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5277 which has been initialized previously
5278 with <tt><a href="#int_va_start">llvm.va_start</a></tt>
5279 or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5280
5281<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5282<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
5283
5284<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5285<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]5286 macro available in C. In a target-dependent way, it destroys
5287 the <tt>va_list</tt> element to which the argument points. Calls
5288 to <a href="#int_va_start"><tt>llvm.va_start</tt></a>
5289 and <a href="#int_va_copy"> <tt>llvm.va_copy</tt></a> must be matched exactly
5290 with calls to <tt>llvm.va_end</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5291
5292</div>
5293
5294<!-- _______________________________________________________________________ -->
5295<div class="doc_subsubsection">
5296 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
5297</div>
5298
5299<div class="doc_text">
5300
5301<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5302<pre>
5303 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
5304</pre>
5305
5306<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5307<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5308 from the source argument list to the destination argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5309
5310<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5311<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]5312 The second argument is a pointer to a <tt>va_list</tt> element to copy
5313 from.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5314
5315<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5316<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]5317 macro available in C. In a target-dependent way, it copies the
5318 source <tt>va_list</tt> element into the destination <tt>va_list</tt>
5319 element. This intrinsic is necessary because
5320 the <tt><a href="#int_va_start"> llvm.va_start</a></tt> intrinsic may be
5321 arbitrarily complex and require, for example, memory allocation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5322
5323</div>
5324
5325<!-- ======================================================================= -->
5326<div class="doc_subsection">
5327 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
5328</div>
5329
5330<div class="doc_text">
5331
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5332<p>LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5333Collection</a> (GC) requires the implementation and generation of these
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5334intrinsics. These intrinsics allow identification of <a href="#int_gcroot">GC
5335roots on the stack</a>, as well as garbage collector implementations that
5336require <a href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a>
5337barriers. Front-ends for type-safe garbage collected languages should generate
5338these intrinsics to make use of the LLVM garbage collectors. For more details,
5339see <a href="GarbageCollection.html">Accurate Garbage Collection with
5340LLVM</a>.</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]5341
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5342<p>The garbage collection intrinsics only operate on objects in the generic
5343 address space (address space zero).</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]5344
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5345</div>
5346
5347<!-- _______________________________________________________________________ -->
5348<div class="doc_subsubsection">
5349 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
5350</div>
5351
5352<div class="doc_text">
5353
5354<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5355<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5356 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5357</pre>
5358
5359<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5360<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]5361 the code generator, and allows some metadata to be associated with it.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5362
5363<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5364<p>The first argument specifies the address of a stack object that contains the
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5365 root pointer. The second pointer (which must be either a constant or a
5366 global value address) contains the meta-data to be associated with the
5367 root.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5368
5369<h5>Semantics:</h5>
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]5370<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]5371 location. At compile-time, the code generator generates information to allow
5372 the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5373 intrinsic may only be used in a function which <a href="#gc">specifies a GC
5374 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5375
5376</div>
5377
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5378<!-- _______________________________________________________________________ -->
5379<div class="doc_subsubsection">
5380 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5381</div>
5382
5383<div class="doc_text">
5384
5385<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5386<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5387 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5388</pre>
5389
5390<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5391<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5392 locations, allowing garbage collector implementations that require read
5393 barriers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5394
5395<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5396<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]5397 allocated from the garbage collector. The first object is a pointer to the
5398 start of the referenced object, if needed by the language runtime (otherwise
5399 null).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5400
5401<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5402<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5403 instruction, but may be replaced with substantially more complex code by the
5404 garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5405 may only be used in a function which <a href="#gc">specifies a GC
5406 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5407
5408</div>
5409
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5410<!-- _______________________________________________________________________ -->
5411<div class="doc_subsubsection">
5412 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5413</div>
5414
5415<div class="doc_text">
5416
5417<h5>Syntax:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5418<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5419 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5420</pre>
5421
5422<h5>Overview:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5423<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5424 locations, allowing garbage collector implementations that require write
5425 barriers (such as generational or reference counting collectors).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5426
5427<h5>Arguments:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5428<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]5429 object to store it to, and the third is the address of the field of Obj to
5430 store to. If the runtime does not require a pointer to the object, Obj may
5431 be null.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5432
5433<h5>Semantics:</h5>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5434<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5435 instruction, but may be replaced with substantially more complex code by the
5436 garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5437 may only be used in a function which <a href="#gc">specifies a GC
5438 algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5439
5440</div>
5441
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5442<!-- ======================================================================= -->
5443<div class="doc_subsection">
5444 <a name="int_codegen">Code Generator Intrinsics</a>
5445</div>
5446
5447<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5448
5449<p>These intrinsics are provided by LLVM to expose special features that may
5450 only be implemented with code generator support.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5451
5452</div>
5453
5454<!-- _______________________________________________________________________ -->
5455<div class="doc_subsubsection">
5456 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5457</div>
5458
5459<div class="doc_text">
5460
5461<h5>Syntax:</h5>
5462<pre>
5463 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
5464</pre>
5465
5466<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5467<p>The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5468 target-specific value indicating the return address of the current function
5469 or one of its callers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5470
5471<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5472<p>The argument to this intrinsic indicates which function to return the address
5473 for. Zero indicates the calling function, one indicates its caller, etc.
5474 The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5475
5476<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5477<p>The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer
5478 indicating the return address of the specified call frame, or zero if it
5479 cannot be identified. The value returned by this intrinsic is likely to be
5480 incorrect or 0 for arguments other than zero, so it should only be used for
5481 debugging purposes.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5482
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5483<p>Note that calling this intrinsic does not prevent function inlining or other
5484 aggressive transformations, so the value returned may not be that of the
5485 obvious source-language caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5486
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5487</div>
5488
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5489<!-- _______________________________________________________________________ -->
5490<div class="doc_subsubsection">
5491 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5492</div>
5493
5494<div class="doc_text">
5495
5496<h5>Syntax:</h5>
5497<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5498 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5499</pre>
5500
5501<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5502<p>The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5503 target-specific frame pointer value for the specified stack frame.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5504
5505<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5506<p>The argument to this intrinsic indicates which function to return the frame
5507 pointer for. Zero indicates the calling function, one indicates its caller,
5508 etc. The argument is <b>required</b> to be a constant integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5509
5510<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5511<p>The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer
5512 indicating the frame address of the specified call frame, or zero if it
5513 cannot be identified. The value returned by this intrinsic is likely to be
5514 incorrect or 0 for arguments other than zero, so it should only be used for
5515 debugging purposes.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5516
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5517<p>Note that calling this intrinsic does not prevent function inlining or other
5518 aggressive transformations, so the value returned may not be that of the
5519 obvious source-language caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5520
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5521</div>
5522
5523<!-- _______________________________________________________________________ -->
5524<div class="doc_subsubsection">
5525 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5526</div>
5527
5528<div class="doc_text">
5529
5530<h5>Syntax:</h5>
5531<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5532 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5533</pre>
5534
5535<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5536<p>The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state
5537 of the function stack, for use
5538 with <a href="#int_stackrestore"> <tt>llvm.stackrestore</tt></a>. This is
5539 useful for implementing language features like scoped automatic variable
5540 sized arrays in C99.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5541
5542<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5543<p>This intrinsic returns a opaque pointer value that can be passed
5544 to <a href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When
5545 an <tt>llvm.stackrestore</tt> intrinsic is executed with a value saved
5546 from <tt>llvm.stacksave</tt>, it effectively restores the state of the stack
5547 to the state it was in when the <tt>llvm.stacksave</tt> intrinsic executed.
5548 In practice, this pops any <a href="#i_alloca">alloca</a> blocks from the
5549 stack that were allocated after the <tt>llvm.stacksave</tt> was executed.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5550
5551</div>
5552
5553<!-- _______________________________________________________________________ -->
5554<div class="doc_subsubsection">
5555 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5556</div>
5557
5558<div class="doc_text">
5559
5560<h5>Syntax:</h5>
5561<pre>
5562 declare void @llvm.stackrestore(i8 * %ptr)
5563</pre>
5564
5565<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5566<p>The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5567 the function stack to the state it was in when the
5568 corresponding <a href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic
5569 executed. This is useful for implementing language features like scoped
5570 automatic variable sized arrays in C99.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5571
5572<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5573<p>See the description
5574 for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5575
5576</div>
5577
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5578<!-- _______________________________________________________________________ -->
5579<div class="doc_subsubsection">
5580 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5581</div>
5582
5583<div class="doc_text">
5584
5585<h5>Syntax:</h5>
5586<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5587 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]5588</pre>
5589
5590<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5591<p>The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to
5592 insert a prefetch instruction if supported; otherwise, it is a noop.
5593 Prefetches have no effect on the behavior of the program but can change its
5594 performance characteristics.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5595
5596<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5597<p><tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the
5598 specifier determining if the fetch should be for a read (0) or write (1),
5599 and <tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5600 locality, to (3) - extremely local keep in cache. The <tt>rw</tt>
5601 and <tt>locality</tt> arguments must be constant integers.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5602
5603<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5604<p>This intrinsic does not modify the behavior of the program. In particular,
5605 prefetches cannot trap and do not produce a value. On targets that support
5606 this intrinsic, the prefetch can provide hints to the processor cache for
5607 better performance.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5608
5609</div>
5610
5611<!-- _______________________________________________________________________ -->
5612<div class="doc_subsubsection">
5613 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5614</div>
5615
5616<div class="doc_text">
5617
5618<h5>Syntax:</h5>
5619<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5620 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5621</pre>
5622
5623<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5624<p>The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program
5625 Counter (PC) in a region of code to simulators and other tools. The method
5626 is target specific, but it is expected that the marker will use exported
5627 symbols to transmit the PC of the marker. The marker makes no guarantees
5628 that it will remain with any specific instruction after optimizations. It is
5629 possible that the presence of a marker will inhibit optimizations. The
5630 intended use is to be inserted after optimizations to allow correlations of
5631 simulation runs.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5632
5633<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5634<p><tt>id</tt> is a numerical id identifying the marker.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5635
5636<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5637<p>This intrinsic does not modify the behavior of the program. Backends that do
5638 not support this intrinisic may ignore it.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5639
5640</div>
5641
5642<!-- _______________________________________________________________________ -->
5643<div class="doc_subsubsection">
5644 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5645</div>
5646
5647<div class="doc_text">
5648
5649<h5>Syntax:</h5>
5650<pre>
5651 declare i64 @llvm.readcyclecounter( )
5652</pre>
5653
5654<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5655<p>The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5656 counter register (or similar low latency, high accuracy clocks) on those
5657 targets that support it. On X86, it should map to RDTSC. On Alpha, it
5658 should map to RPCC. As the backing counters overflow quickly (on the order
5659 of 9 seconds on alpha), this should only be used for small timings.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5660
5661<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5662<p>When directly supported, reading the cycle counter should not modify any
5663 memory. Implementations are allowed to either return a application specific
5664 value or a system wide value. On backends without support, this is lowered
5665 to a constant 0.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5666
5667</div>
5668
5669<!-- ======================================================================= -->
5670<div class="doc_subsection">
5671 <a name="int_libc">Standard C Library Intrinsics</a>
5672</div>
5673
5674<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5675
5676<p>LLVM provides intrinsics for a few important standard C library functions.
5677 These intrinsics allow source-language front-ends to pass information about
5678 the alignment of the pointer arguments to the code generator, providing
5679 opportunity for more efficient code generation.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5680
5681</div>
5682
5683<!-- _______________________________________________________________________ -->
5684<div class="doc_subsubsection">
5685 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5686</div>
5687
5688<div class="doc_text">
5689
5690<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5691<p>This is an overloaded intrinsic. You can use <tt>llvm.memcpy</tt> on any
5692 integer bit width. Not all targets support all bit widths however.</p>
5693
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5694<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5695 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5696 i8 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5697 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5698 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5699 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5700 i32 &lt;len&gt;, i32 &lt;align&gt;)
5701 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5702 i64 &lt;len&gt;, i32 &lt;align&gt;)
5703</pre>
5704
5705<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5706<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5707 source location to the destination location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5708
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5709<p>Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5710 intrinsics do not return a value, and takes an extra alignment argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5711
5712<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5713<p>The first argument is a pointer to the destination, the second is a pointer
5714 to the source. The third argument is an integer argument specifying the
5715 number of bytes to copy, and the fourth argument is the alignment of the
5716 source and destination locations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5717
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5718<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5719 then the caller guarantees that both the source and destination pointers are
5720 aligned to that boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5721
5722<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5723<p>The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the
5724 source location to the destination location, which are not allowed to
5725 overlap. It copies "len" bytes of memory over. If the argument is known to
5726 be aligned to some boundary, this can be specified as the fourth argument,
5727 otherwise it should be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5728
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5729</div>
5730
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5731<!-- _______________________________________________________________________ -->
5732<div class="doc_subsubsection">
5733 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5734</div>
5735
5736<div class="doc_text">
5737
5738<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5739<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]5740 width. Not all targets support all bit widths however.</p>
5741
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5742<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5743 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5744 i8 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5745 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5746 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5747 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5748 i32 &lt;len&gt;, i32 &lt;align&gt;)
5749 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5750 i64 &lt;len&gt;, i32 &lt;align&gt;)
5751</pre>
5752
5753<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5754<p>The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the
5755 source location to the destination location. It is similar to the
5756 '<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to
5757 overlap.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5758
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5759<p>Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5760 intrinsics do not return a value, and takes an extra alignment argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5761
5762<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5763<p>The first argument is a pointer to the destination, the second is a pointer
5764 to the source. The third argument is an integer argument specifying the
5765 number of bytes to copy, and the fourth argument is the alignment of the
5766 source and destination locations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5767
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5768<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5769 then the caller guarantees that the source and destination pointers are
5770 aligned to that boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5771
5772<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5773<p>The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the
5774 source location to the destination location, which may overlap. It copies
5775 "len" bytes of memory over. If the argument is known to be aligned to some
5776 boundary, this can be specified as the fourth argument, otherwise it should
5777 be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5778
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5779</div>
5780
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5781<!-- _______________________________________________________________________ -->
5782<div class="doc_subsubsection">
5783 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5784</div>
5785
5786<div class="doc_text">
5787
5788<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5789<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]5790 width. Not all targets support all bit widths however.</p>
5791
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5792<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5793 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5794 i8 &lt;len&gt;, i32 &lt;align&gt;)
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5795 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5796 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5797 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5798 i32 &lt;len&gt;, i32 &lt;align&gt;)
5799 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5800 i64 &lt;len&gt;, i32 &lt;align&gt;)
5801</pre>
5802
5803<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5804<p>The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a
5805 particular byte value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5806
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5807<p>Note that, unlike the standard libc function, the <tt>llvm.memset</tt>
5808 intrinsic does not return a value, and takes an extra alignment argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5809
5810<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5811<p>The first argument is a pointer to the destination to fill, the second is the
5812 byte value to fill it with, the third argument is an integer argument
5813 specifying the number of bytes to fill, and the fourth argument is the known
5814 alignment of destination location.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5815
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5816<p>If the call to this intrinisic has an alignment value that is not 0 or 1,
5817 then the caller guarantees that the destination pointer is aligned to that
5818 boundary.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5819
5820<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5821<p>The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting
5822 at the destination location. If the argument is known to be aligned to some
5823 boundary, this can be specified as the fourth argument, otherwise it should
5824 be set to 0 or 1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5825
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5826</div>
5827
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5828<!-- _______________________________________________________________________ -->
5829<div class="doc_subsubsection">
5830 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5831</div>
5832
5833<div class="doc_text">
5834
5835<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5836<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
5837 floating point or vector of floating point type. Not all targets support all
5838 types however.</p>
5839
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5840<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5841 declare float @llvm.sqrt.f32(float %Val)
5842 declare double @llvm.sqrt.f64(double %Val)
5843 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5844 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5845 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5846</pre>
5847
5848<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5849<p>The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
5850 returning the same value as the libm '<tt>sqrt</tt>' functions would.
5851 Unlike <tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined
5852 behavior for negative numbers other than -0.0 (which allows for better
5853 optimization, because there is no need to worry about errno being
5854 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]5855
5856<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5857<p>The argument and return value are floating point numbers of the same
5858 type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5859
5860<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5861<p>This function returns the sqrt of the specified operand if it is a
5862 nonnegative floating point number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5863
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5864</div>
5865
5866<!-- _______________________________________________________________________ -->
5867<div class="doc_subsubsection">
5868 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5869</div>
5870
5871<div class="doc_text">
5872
5873<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5874<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
5875 floating point or vector of floating point type. Not all targets support all
5876 types however.</p>
5877
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5878<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5879 declare float @llvm.powi.f32(float %Val, i32 %power)
5880 declare double @llvm.powi.f64(double %Val, i32 %power)
5881 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5882 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5883 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5884</pre>
5885
5886<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5887<p>The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5888 specified (positive or negative) power. The order of evaluation of
5889 multiplications is not defined. When a vector of floating point type is
5890 used, the second argument remains a scalar integer value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5891
5892<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5893<p>The second argument is an integer power, and the first is a value to raise to
5894 that power.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5895
5896<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5897<p>This function returns the first value raised to the second power with an
5898 unspecified sequence of rounding operations.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5899
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5900</div>
5901
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5902<!-- _______________________________________________________________________ -->
5903<div class="doc_subsubsection">
5904 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5905</div>
5906
5907<div class="doc_text">
5908
5909<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5910<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5911 floating point or vector of floating point type. Not all targets support all
5912 types however.</p>
5913
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5914<pre>
5915 declare float @llvm.sin.f32(float %Val)
5916 declare double @llvm.sin.f64(double %Val)
5917 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5918 declare fp128 @llvm.sin.f128(fp128 %Val)
5919 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5920</pre>
5921
5922<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5923<p>The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5924
5925<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5926<p>The argument and return value are floating point numbers of the same
5927 type.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5928
5929<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5930<p>This function returns the sine of the specified operand, returning the same
5931 values as the libm <tt>sin</tt> functions would, and handles error conditions
5932 in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5933
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5934</div>
5935
5936<!-- _______________________________________________________________________ -->
5937<div class="doc_subsubsection">
5938 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5939</div>
5940
5941<div class="doc_text">
5942
5943<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5944<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5945 floating point or vector of floating point type. Not all targets support all
5946 types however.</p>
5947
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5948<pre>
5949 declare float @llvm.cos.f32(float %Val)
5950 declare double @llvm.cos.f64(double %Val)
5951 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5952 declare fp128 @llvm.cos.f128(fp128 %Val)
5953 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5954</pre>
5955
5956<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5957<p>The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5958
5959<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5960<p>The argument and return value are floating point numbers of the same
5961 type.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5962
5963<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5964<p>This function returns the cosine of the specified operand, returning the same
5965 values as the libm <tt>cos</tt> functions would, and handles error conditions
5966 in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5967
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5968</div>
5969
5970<!-- _______________________________________________________________________ -->
5971<div class="doc_subsubsection">
5972 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5973</div>
5974
5975<div class="doc_text">
5976
5977<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5978<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5979 floating point or vector of floating point type. Not all targets support all
5980 types however.</p>
5981
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5982<pre>
5983 declare float @llvm.pow.f32(float %Val, float %Power)
5984 declare double @llvm.pow.f64(double %Val, double %Power)
5985 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5986 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5987 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5988</pre>
5989
5990<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5991<p>The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5992 specified (positive or negative) power.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5993
5994<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5995<p>The second argument is a floating point power, and the first is a value to
5996 raise to that power.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5997
5998<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]5999<p>This function returns the first value raised to the second power, returning
6000 the same values as the libm <tt>pow</tt> functions would, and handles error
6001 conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6002
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]6003</div>
6004
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6005<!-- ======================================================================= -->
6006<div class="doc_subsection">
6007 <a name="int_manip">Bit Manipulation Intrinsics</a>
6008</div>
6009
6010<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6011
6012<p>LLVM provides intrinsics for a few important bit manipulation operations.
6013 These allow efficient code generation for some algorithms.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6014
6015</div>
6016
6017<!-- _______________________________________________________________________ -->
6018<div class="doc_subsubsection">
6019 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
6020</div>
6021
6022<div class="doc_text">
6023
6024<h5>Syntax:</h5>
6025<p>This is an overloaded intrinsic function. You can use bswap on any integer
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6026 type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
6027
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6028<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6029 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
6030 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
6031 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6032</pre>
6033
6034<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6035<p>The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
6036 values with an even number of bytes (positive multiple of 16 bits). These
6037 are useful for performing operations on data that is not in the target's
6038 native byte order.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6039
6040<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6041<p>The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
6042 and low byte of the input i16 swapped. Similarly,
6043 the <tt>llvm.bswap.i32</tt> intrinsic returns an i32 value that has the four
6044 bytes of the input i32 swapped, so that if the input bytes are numbered 0, 1,
6045 2, 3 then the returned i32 will have its bytes in 3, 2, 1, 0 order.
6046 The <tt>llvm.bswap.i48</tt>, <tt>llvm.bswap.i64</tt> and other intrinsics
6047 extend this concept to additional even-byte lengths (6 bytes, 8 bytes and
6048 more, respectively).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6049
6050</div>
6051
6052<!-- _______________________________________________________________________ -->
6053<div class="doc_subsubsection">
6054 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
6055</div>
6056
6057<div class="doc_text">
6058
6059<h5>Syntax:</h5>
6060<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]6061 width. Not all targets support all bit widths however.</p>
6062
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6063<pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6064 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6065 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6066 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6067 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
6068 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6069</pre>
6070
6071<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6072<p>The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set
6073 in a value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6074
6075<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6076<p>The only argument is the value to be counted. The argument may be of any
6077 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6078
6079<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6080<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]6081
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6082</div>
6083
6084<!-- _______________________________________________________________________ -->
6085<div class="doc_subsubsection">
6086 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
6087</div>
6088
6089<div class="doc_text">
6090
6091<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6092<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
6093 integer bit width. Not all targets support all bit widths however.</p>
6094
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6095<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6096 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
6097 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6098 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6099 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
6100 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6101</pre>
6102
6103<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6104<p>The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
6105 leading zeros in a variable.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6106
6107<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6108<p>The only argument is the value to be counted. The argument may be of any
6109 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6110
6111<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6112<p>The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant)
6113 zeros in a variable. If the src == 0 then the result is the size in bits of
6114 the type of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6115
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6116</div>
6117
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6118<!-- _______________________________________________________________________ -->
6119<div class="doc_subsubsection">
6120 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
6121</div>
6122
6123<div class="doc_text">
6124
6125<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6126<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
6127 integer bit width. Not all targets support all bit widths however.</p>
6128
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6129<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6130 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
6131 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6132 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6133 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
6134 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6135</pre>
6136
6137<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6138<p>The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
6139 trailing zeros.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6140
6141<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6142<p>The only argument is the value to be counted. The argument may be of any
6143 integer type. The return type must match the argument type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6144
6145<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6146<p>The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant)
6147 zeros in a variable. If the src == 0 then the result is the size in bits of
6148 the type of src. For example, <tt>llvm.cttz(2) = 1</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6149
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6150</div>
6151
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6152<!-- ======================================================================= -->
6153<div class="doc_subsection">
6154 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6155</div>
6156
6157<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6158
6159<p>LLVM provides intrinsics for some arithmetic with overflow operations.</p>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6160
6161</div>
6162
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6163<!-- _______________________________________________________________________ -->
6164<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6165 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6166</div>
6167
6168<div class="doc_text">
6169
6170<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6171<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]6172 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6173
6174<pre>
6175 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6176 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6177 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6178</pre>
6179
6180<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6181<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6182 a signed addition of the two arguments, and indicate whether an overflow
6183 occurred during the signed summation.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6184
6185<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6186<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]6187 be of integer types of any bit width, but they must have the same bit
6188 width. The second element of the result structure must be of
6189 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6190 undergo signed addition.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6191
6192<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6193<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6194 a signed addition of the two variables. They return a structure &mdash; the
6195 first element of which is the signed summation, and the second element of
6196 which is a bit specifying if the signed summation resulted in an
6197 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6198
6199<h5>Examples:</h5>
6200<pre>
6201 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6202 %sum = extractvalue {i32, i1} %res, 0
6203 %obit = extractvalue {i32, i1} %res, 1
6204 br i1 %obit, label %overflow, label %normal
6205</pre>
6206
6207</div>
6208
6209<!-- _______________________________________________________________________ -->
6210<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6211 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6212</div>
6213
6214<div class="doc_text">
6215
6216<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6217<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]6218 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6219
6220<pre>
6221 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6222 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6223 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6224</pre>
6225
6226<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6227<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6228 an unsigned addition of the two arguments, and indicate whether a carry
6229 occurred during the unsigned summation.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6230
6231<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6232<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]6233 be of integer types of any bit width, but they must have the same bit
6234 width. The second element of the result structure must be of
6235 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6236 undergo unsigned addition.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6237
6238<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6239<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6240 an unsigned addition of the two arguments. They return a structure &mdash;
6241 the first element of which is the sum, and the second element of which is a
6242 bit specifying if the unsigned summation resulted in a carry.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6243
6244<h5>Examples:</h5>
6245<pre>
6246 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6247 %sum = extractvalue {i32, i1} %res, 0
6248 %obit = extractvalue {i32, i1} %res, 1
6249 br i1 %obit, label %carry, label %normal
6250</pre>
6251
6252</div>
6253
6254<!-- _______________________________________________________________________ -->
6255<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6256 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6257</div>
6258
6259<div class="doc_text">
6260
6261<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6262<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]6263 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6264
6265<pre>
6266 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6267 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6268 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6269</pre>
6270
6271<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6272<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6273 a signed subtraction of the two arguments, and indicate whether an overflow
6274 occurred during the signed subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6275
6276<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6277<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]6278 be of integer types of any bit width, but they must have the same bit
6279 width. The second element of the result structure must be of
6280 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6281 undergo signed subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6282
6283<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6284<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6285 a signed subtraction of the two arguments. They return a structure &mdash;
6286 the first element of which is the subtraction, and the second element of
6287 which is a bit specifying if the signed subtraction resulted in an
6288 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6289
6290<h5>Examples:</h5>
6291<pre>
6292 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6293 %sum = extractvalue {i32, i1} %res, 0
6294 %obit = extractvalue {i32, i1} %res, 1
6295 br i1 %obit, label %overflow, label %normal
6296</pre>
6297
6298</div>
6299
6300<!-- _______________________________________________________________________ -->
6301<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6302 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6303</div>
6304
6305<div class="doc_text">
6306
6307<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6308<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]6309 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6310
6311<pre>
6312 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6313 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6314 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6315</pre>
6316
6317<h5>Overview:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6318<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6319 an unsigned subtraction of the two arguments, and indicate whether an
6320 overflow occurred during the unsigned subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6321
6322<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6323<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]6324 be of integer types of any bit width, but they must have the same bit
6325 width. The second element of the result structure must be of
6326 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6327 undergo unsigned subtraction.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6328
6329<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6330<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6331 an unsigned subtraction of the two arguments. They return a structure &mdash;
6332 the first element of which is the subtraction, and the second element of
6333 which is a bit specifying if the unsigned subtraction resulted in an
6334 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6335
6336<h5>Examples:</h5>
6337<pre>
6338 %res = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6339 %sum = extractvalue {i32, i1} %res, 0
6340 %obit = extractvalue {i32, i1} %res, 1
6341 br i1 %obit, label %overflow, label %normal
6342</pre>
6343
6344</div>
6345
6346<!-- _______________________________________________________________________ -->
6347<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6348 <a name="int_smul_overflow">'<tt>llvm.smul.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6349</div>
6350
6351<div class="doc_text">
6352
6353<h5>Syntax:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6354<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]6355 on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6356
6357<pre>
6358 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6359 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6360 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6361</pre>
6362
6363<h5>Overview:</h5>
6364
6365<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6366 a signed multiplication of the two arguments, and indicate whether an
6367 overflow occurred during the signed multiplication.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6368
6369<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6370<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]6371 be of integer types of any bit width, but they must have the same bit
6372 width. The second element of the result structure must be of
6373 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6374 undergo signed multiplication.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6375
6376<h5>Semantics:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6377<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6378 a signed multiplication of the two arguments. They return a structure &mdash;
6379 the first element of which is the multiplication, and the second element of
6380 which is a bit specifying if the signed multiplication resulted in an
6381 overflow.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6382
6383<h5>Examples:</h5>
6384<pre>
6385 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6386 %sum = extractvalue {i32, i1} %res, 0
6387 %obit = extractvalue {i32, i1} %res, 1
6388 br i1 %obit, label %overflow, label %normal
6389</pre>
6390
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6391</div>
6392
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6393<!-- _______________________________________________________________________ -->
6394<div class="doc_subsubsection">
6395 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6396</div>
6397
6398<div class="doc_text">
6399
6400<h5>Syntax:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6401<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]6402 on any integer bit width.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6403
6404<pre>
6405 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6406 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6407 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6408</pre>
6409
6410<h5>Overview:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6411<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6412 a unsigned multiplication of the two arguments, and indicate whether an
6413 overflow occurred during the unsigned multiplication.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6414
6415<h5>Arguments:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6416<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]6417 be of integer types of any bit width, but they must have the same bit
6418 width. The second element of the result structure must be of
6419 type <tt>i1</tt>. <tt>%a</tt> and <tt>%b</tt> are the two values that will
6420 undergo unsigned multiplication.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6421
6422<h5>Semantics:</h5>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6423<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6424 an unsigned multiplication of the two arguments. They return a structure
6425 &mdash; the first element of which is the multiplication, and the second
6426 element of which is a bit specifying if the unsigned multiplication resulted
6427 in an overflow.</p>
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6428
6429<h5>Examples:</h5>
6430<pre>
6431 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6432 %sum = extractvalue {i32, i1} %res, 0
6433 %obit = extractvalue {i32, i1} %res, 1
6434 br i1 %obit, label %overflow, label %normal
6435</pre>
6436
6437</div>
6438
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6439<!-- ======================================================================= -->
6440<div class="doc_subsection">
6441 <a name="int_debugger">Debugger Intrinsics</a>
6442</div>
6443
6444<div class="doc_text">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6445
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6446<p>The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt>
6447 prefix), are described in
6448 the <a href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source
6449 Level Debugging</a> document.</p>
6450
6451</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6452
6453<!-- ======================================================================= -->
6454<div class="doc_subsection">
6455 <a name="int_eh">Exception Handling Intrinsics</a>
6456</div>
6457
6458<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6459
6460<p>The LLVM exception handling intrinsics (which all start with
6461 <tt>llvm.eh.</tt> prefix), are described in
6462 the <a href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6463 Handling</a> document.</p>
6464
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6465</div>
6466
6467<!-- ======================================================================= -->
6468<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6469 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6470</div>
6471
6472<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6473
6474<p>This intrinsic makes it possible to excise one parameter, marked with
6475 the <tt>nest</tt> attribute, from a function. The result is a callable
6476 function pointer lacking the nest parameter - the caller does not need to
6477 provide a value for it. Instead, the value to use is stored in advance in a
6478 "trampoline", a block of memory usually allocated on the stack, which also
6479 contains code to splice the nest value into the argument list. This is used
6480 to implement the GCC nested function address extension.</p>
6481
6482<p>For example, if the function is
6483 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
6484 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as
6485 follows:</p>
6486
6487<div class="doc_code">
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6488<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6489 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6490 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6491 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6492 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6493</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6494</div>
6495
6496<p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6497 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
6498
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6499</div>
6500
6501<!-- _______________________________________________________________________ -->
6502<div class="doc_subsubsection">
6503 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6504</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6505
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6506<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6507
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6508<h5>Syntax:</h5>
6509<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6510 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]6511</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6512
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6513<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6514<p>This fills the memory pointed to by <tt>tramp</tt> with code and returns a
6515 function pointer suitable for executing it.</p>
6516
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6517<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6518<p>The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6519 pointers. The <tt>tramp</tt> argument must point to a sufficiently large and
6520 sufficiently aligned block of memory; this memory is written to by the
6521 intrinsic. Note that the size and the alignment are target-specific - LLVM
6522 currently provides no portable way of determining them, so a front-end that
6523 generates this intrinsic needs to have some target-specific knowledge.
6524 The <tt>func</tt> argument must hold a function bitcast to
6525 an <tt>i8*</tt>.</p>
6526
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6527<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6528<p>The block of memory pointed to by <tt>tramp</tt> is filled with target
6529 dependent code, turning it into a function. A pointer to this function is
6530 returned, but needs to be bitcast to an <a href="#int_trampoline">appropriate
6531 function pointer type</a> before being called. The new function's signature
6532 is the same as that of <tt>func</tt> with any arguments marked with
6533 the <tt>nest</tt> attribute removed. At most one such <tt>nest</tt> argument
6534 is allowed, and it must be of pointer type. Calling the new function is
6535 equivalent to calling <tt>func</tt> with the same argument list, but
6536 with <tt>nval</tt> used for the missing <tt>nest</tt> argument. If, after
6537 calling <tt>llvm.init.trampoline</tt>, the memory pointed to
6538 by <tt>tramp</tt> is modified, then the effect of any later call to the
6539 returned function pointer is undefined.</p>
6540
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6541</div>
6542
6543<!-- ======================================================================= -->
6544<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6545 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6546</div>
6547
6548<div class="doc_text">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6549
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6550<p>These intrinsic functions expand the "universal IR" of LLVM to represent
6551 hardware constructs for atomic operations and memory synchronization. This
6552 provides an interface to the hardware, not an interface to the programmer. It
6553 is aimed at a low enough level to allow any programming models or APIs
6554 (Application Programming Interfaces) which need atomic behaviors to map
6555 cleanly onto it. It is also modeled primarily on hardware behavior. Just as
6556 hardware provides a "universal IR" for source languages, it also provides a
6557 starting point for developing a "universal" atomic operation and
6558 synchronization IR.</p>
6559
6560<p>These do <em>not</em> form an API such as high-level threading libraries,
6561 software transaction memory systems, atomic primitives, and intrinsic
6562 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6563 application libraries. The hardware interface provided by LLVM should allow
6564 a clean implementation of all of these APIs and parallel programming models.
6565 No one model or paradigm should be selected above others unless the hardware
6566 itself ubiquitously does so.</p>
6567
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6568</div>
6569
6570<!-- _______________________________________________________________________ -->
6571<div class="doc_subsubsection">
6572 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6573</div>
6574<div class="doc_text">
6575<h5>Syntax:</h5>
6576<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6577 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]6578</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6579
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6580<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6581<p>The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6582 specific pairs of memory access types.</p>
6583
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6584<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6585<p>The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6586 The first four arguments enables a specific barrier as listed below. The
6587 fith argument specifies that the barrier applies to io or device or uncached
6588 memory.</p>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6589
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6590<ul>
6591 <li><tt>ll</tt>: load-load barrier</li>
6592 <li><tt>ls</tt>: load-store barrier</li>
6593 <li><tt>sl</tt>: store-load barrier</li>
6594 <li><tt>ss</tt>: store-store barrier</li>
6595 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
6596</ul>
6597
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6598<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6599<p>This intrinsic causes the system to enforce some ordering constraints upon
6600 the loads and stores of the program. This barrier does not
6601 indicate <em>when</em> any events will occur, it only enforces
6602 an <em>order</em> in which they occur. For any of the specified pairs of load
6603 and store operations (f.ex. load-load, or store-load), all of the first
6604 operations preceding the barrier will complete before any of the second
6605 operations succeeding the barrier begin. Specifically the semantics for each
6606 pairing is as follows:</p>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6607
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6608<ul>
6609 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6610 after the barrier begins.</li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]6611 <li><tt>ls</tt>: All loads before the barrier must complete before any
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6612 store after the barrier begins.</li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]6613 <li><tt>ss</tt>: All stores before the barrier must complete before any
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6614 store after the barrier begins.</li>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]6615 <li><tt>sl</tt>: All stores before the barrier must complete before any
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6616 load after the barrier begins.</li>
6617</ul>
6618
6619<p>These semantics are applied with a logical "and" behavior when more than one
6620 is enabled in a single memory barrier intrinsic.</p>
6621
6622<p>Backends may implement stronger barriers than those requested when they do
6623 not support as fine grained a barrier as requested. Some architectures do
6624 not need all types of barriers and on such architectures, these become
6625 noops.</p>
6626
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6627<h5>Example:</h5>
6628<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6629%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6630%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6631 store i32 4, %ptr
6632
6633%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6634 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6635 <i>; guarantee the above finishes</i>
6636 store i32 8, %ptr <i>; before this begins</i>
6637</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6638
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6639</div>
6640
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6641<!-- _______________________________________________________________________ -->
6642<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6643 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6644</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6645
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6646<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6647
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6648<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6649<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6650 any integer bit width and for different address spaces. Not all targets
6651 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6652
6653<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6654 declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
6655 declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
6656 declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
6657 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]6658</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6659
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6660<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6661<p>This loads a value in memory and compares it to a given value. If they are
6662 equal, it stores a new value into the memory.</p>
6663
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6664<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6665<p>The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result
6666 as well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6667 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6668 this integer type. While any bit width integer may be used, targets may only
6669 lower representations they support in hardware.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6670
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6671<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6672<p>This entire intrinsic must be executed atomically. It first loads the value
6673 in memory pointed to by <tt>ptr</tt> and compares it with the
6674 value <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the
6675 memory. The loaded value is yielded in all cases. This provides the
6676 equivalent of an atomic compare-and-swap operation within the SSA
6677 framework.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6678
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6679<h5>Examples:</h5>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6680<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6681%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6682%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6683 store i32 4, %ptr
6684
6685%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6686%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6687 <i>; yields {i32}:result1 = 4</i>
6688%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6689%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6690
6691%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6692%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6693 <i>; yields {i32}:result2 = 8</i>
6694%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6695
6696%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6697</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6698
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6699</div>
6700
6701<!-- _______________________________________________________________________ -->
6702<div class="doc_subsubsection">
6703 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6704</div>
6705<div class="doc_text">
6706<h5>Syntax:</h5>
6707
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6708<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6709 integer bit width. Not all targets support all bit widths however.</p>
6710
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6711<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6712 declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6713 declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6714 declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6715 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]6716</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6717
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6718<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6719<p>This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6720 the value from memory. It then stores the value in <tt>val</tt> in the memory
6721 at <tt>ptr</tt>.</p>
6722
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6723<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6724<p>The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both
6725 the <tt>val</tt> argument and the result must be integers of the same bit
6726 width. The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6727 integer type. The targets may only lower integer representations they
6728 support.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6729
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6730<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6731<p>This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6732 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6733 equivalent of an atomic swap operation within the SSA framework.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6734
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6735<h5>Examples:</h5>
6736<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6737%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6738%ptr = bitcast i8* %mallocP to i32*
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6739 store i32 4, %ptr
6740
6741%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6742%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6743 <i>; yields {i32}:result1 = 4</i>
6744%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6745%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6746
6747%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6748%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6749 <i>; yields {i32}:result2 = 8</i>
6750
6751%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6752%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6753</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6754
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6755</div>
6756
6757<!-- _______________________________________________________________________ -->
6758<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6759 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6760
6761</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6762
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6763<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6764
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6765<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6766<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on
6767 any integer bit width. Not all targets support all bit widths however.</p>
6768
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6769<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6770 declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6771 declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6772 declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6773 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]6774</pre>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6775
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6776<h5>Overview:</h5>
6777<p>This intrinsic adds <tt>delta</tt> to the value stored in memory
6778 at <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6779
6780<h5>Arguments:</h5>
6781<p>The intrinsic takes two arguments, the first a pointer to an integer value
6782 and the second an integer value. The result is also an integer value. These
6783 integer types can have any bit width, but they must all have the same bit
6784 width. The targets may only lower integer representations they support.</p>
6785
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6786<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6787<p>This intrinsic does a series of operations atomically. It first loads the
6788 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6789 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]6790
6791<h5>Examples:</h5>
6792<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6793%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6794%ptr = bitcast i8* %mallocP to i32*
6795 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6796%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6797 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6798%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6799 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6800%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6801 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6802%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6803</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6804
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6805</div>
6806
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6807<!-- _______________________________________________________________________ -->
6808<div class="doc_subsubsection">
6809 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6810
6811</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6812
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6813<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6814
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6815<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6816<p>This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
6817 any integer bit width and for different address spaces. Not all targets
6818 support all bit widths however.</p>
6819
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6820<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6821 declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6822 declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6823 declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6824 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]6825</pre>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6826
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6827<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]6828<p>This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6829 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.</p>
6830
6831<h5>Arguments:</h5>
6832<p>The intrinsic takes two arguments, the first a pointer to an integer value
6833 and the second an integer value. The result is also an integer value. These
6834 integer types can have any bit width, but they must all have the same bit
6835 width. The targets may only lower integer representations they support.</p>
6836
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6837<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6838<p>This intrinsic does a series of operations atomically. It first loads the
6839 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6840 result to <tt>ptr</tt>. It yields the original value stored
6841 at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6842
6843<h5>Examples:</h5>
6844<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6845%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6846%ptr = bitcast i8* %mallocP to i32*
6847 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6848%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6849 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6850%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6851 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6852%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6853 <i>; yields {i32}:result3 = 2</i>
6854%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6855</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6856
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6857</div>
6858
6859<!-- _______________________________________________________________________ -->
6860<div class="doc_subsubsection">
6861 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6862 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6863 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6864 <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]6865</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6866
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6867<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6868
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6869<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6870<p>These are overloaded intrinsics. You can
6871 use <tt>llvm.atomic.load_and</tt>, <tt>llvm.atomic.load_nand</tt>,
6872 <tt>llvm.atomic.load_or</tt>, and <tt>llvm.atomic.load_xor</tt> on any integer
6873 bit width and for different address spaces. Not all targets support all bit
6874 widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6875
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6876<pre>
6877 declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6878 declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6879 declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6880 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]6881</pre>
6882
6883<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6884 declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6885 declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6886 declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6887 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]6888</pre>
6889
6890<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6891 declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6892 declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6893 declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6894 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]6895</pre>
6896
6897<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6898 declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6899 declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6900 declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6901 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]6902</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6903
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6904<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6905<p>These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6906 the value stored in memory at <tt>ptr</tt>. It yields the original value
6907 at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6908
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6909<h5>Arguments:</h5>
6910<p>These intrinsics take two arguments, the first a pointer to an integer value
6911 and the second an integer value. The result is also an integer value. These
6912 integer types can have any bit width, but they must all have the same bit
6913 width. The targets may only lower integer representations they support.</p>
6914
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6915<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6916<p>These intrinsics does a series of operations atomically. They first load the
6917 value stored at <tt>ptr</tt>. They then do the bitwise
6918 operation <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the
6919 original value stored at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6920
6921<h5>Examples:</h5>
6922<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]6923%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
6924%ptr = bitcast i8* %mallocP to i32*
6925 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6926%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6927 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6928%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6929 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6930%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6931 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6932%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6933 <i>; yields {i32}:result3 = FF</i>
6934%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6935</pre>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6936
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6937</div>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6938
6939<!-- _______________________________________________________________________ -->
6940<div class="doc_subsubsection">
6941 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6942 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6943 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6944 <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]6945</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6946
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6947<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6948
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6949<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6950<p>These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6951 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
6952 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6953 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6954
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6955<pre>
6956 declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6957 declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6958 declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6959 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]6960</pre>
6961
6962<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6963 declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6964 declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6965 declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6966 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]6967</pre>
6968
6969<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6970 declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6971 declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6972 declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6973 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]6974</pre>
6975
6976<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6977 declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6978 declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6979 declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6980 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]6981</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6982
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6983<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]6984<p>These intrinsics takes the signed or unsigned minimum or maximum of
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6985 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6986 original value at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6987
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6988<h5>Arguments:</h5>
6989<p>These intrinsics take two arguments, the first a pointer to an integer value
6990 and the second an integer value. The result is also an integer value. These
6991 integer types can have any bit width, but they must all have the same bit
6992 width. The targets may only lower integer representations they support.</p>
6993
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6994<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]6995<p>These intrinsics does a series of operations atomically. They first load the
6996 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or
6997 max <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They
6998 yield the original value stored at <tt>ptr</tt>.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6999
7000<h5>Examples:</h5>
7001<pre>
Victor Hernandez6b054092009-10-26 23:44:29 +0000[diff] [blame]7002%mallocP = tail call i8* @malloc(i32 ptrtoint (i32* getelementptr (i32* null, i32 1) to i32))
7003%ptr = bitcast i8* %mallocP to i32*
7004 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]7005%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7006 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]7007%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7008 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]7009%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7010 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]7011%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7012 <i>; yields {i32}:result3 = 8</i>
7013%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
7014</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7015
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]7016</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7017
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7018
7019<!-- ======================================================================= -->
7020<div class="doc_subsection">
7021 <a name="int_memorymarkers">Memory Use Markers</a>
7022</div>
7023
7024<div class="doc_text">
7025
7026<p>This class of intrinsics exists to information about the lifetime of memory
7027 objects and ranges where variables are immutable.</p>
7028
7029</div>
7030
7031<!-- _______________________________________________________________________ -->
7032<div class="doc_subsubsection">
7033 <a name="int_lifetime_start">'<tt>llvm.lifetime.start</tt>' Intrinsic</a>
7034</div>
7035
7036<div class="doc_text">
7037
7038<h5>Syntax:</h5>
7039<pre>
7040 declare void @llvm.lifetime.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7041</pre>
7042
7043<h5>Overview:</h5>
7044<p>The '<tt>llvm.lifetime.start</tt>' intrinsic specifies the start of a memory
7045 object's lifetime.</p>
7046
7047<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7048<p>The first argument is a constant integer representing the size of the
7049 object, or -1 if it is variable sized. The second argument is a pointer to
7050 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7051
7052<h5>Semantics:</h5>
7053<p>This intrinsic indicates that before this point in the code, the value of the
7054 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]7055 never be used and has an undefined value. A load from the pointer that
7056 precedes this intrinsic can be replaced with
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7057 <tt>'<a href="#undefvalues">undef</a>'</tt>.</p>
7058
7059</div>
7060
7061<!-- _______________________________________________________________________ -->
7062<div class="doc_subsubsection">
7063 <a name="int_lifetime_end">'<tt>llvm.lifetime.end</tt>' Intrinsic</a>
7064</div>
7065
7066<div class="doc_text">
7067
7068<h5>Syntax:</h5>
7069<pre>
7070 declare void @llvm.lifetime.end(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7071</pre>
7072
7073<h5>Overview:</h5>
7074<p>The '<tt>llvm.lifetime.end</tt>' intrinsic specifies the end of a memory
7075 object's lifetime.</p>
7076
7077<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7078<p>The first argument is a constant integer representing the size of the
7079 object, or -1 if it is variable sized. The second argument is a pointer to
7080 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7081
7082<h5>Semantics:</h5>
7083<p>This intrinsic indicates that after this point in the code, the value of the
7084 memory pointed to by <tt>ptr</tt> is dead. This means that it is known to
7085 never be used and has an undefined value. Any stores into the memory object
7086 following this intrinsic may be removed as dead.
7087
7088</div>
7089
7090<!-- _______________________________________________________________________ -->
7091<div class="doc_subsubsection">
7092 <a name="int_invariant_start">'<tt>llvm.invariant.start</tt>' Intrinsic</a>
7093</div>
7094
7095<div class="doc_text">
7096
7097<h5>Syntax:</h5>
7098<pre>
7099 declare {}* @llvm.invariant.start(i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;) readonly
7100</pre>
7101
7102<h5>Overview:</h5>
7103<p>The '<tt>llvm.invariant.start</tt>' intrinsic specifies that the contents of
7104 a memory object will not change.</p>
7105
7106<h5>Arguments:</h5>
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7107<p>The first argument is a constant integer representing the size of the
7108 object, or -1 if it is variable sized. The second argument is a pointer to
7109 the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7110
7111<h5>Semantics:</h5>
7112<p>This intrinsic indicates that until an <tt>llvm.invariant.end</tt> that uses
7113 the return value, the referenced memory location is constant and
7114 unchanging.</p>
7115
7116</div>
7117
7118<!-- _______________________________________________________________________ -->
7119<div class="doc_subsubsection">
7120 <a name="int_invariant_end">'<tt>llvm.invariant.end</tt>' Intrinsic</a>
7121</div>
7122
7123<div class="doc_text">
7124
7125<h5>Syntax:</h5>
7126<pre>
7127 declare void @llvm.invariant.end({}* &lt;start&gt;, i64 &lt;size&gt;, i8* nocapture &lt;ptr&gt;)
7128</pre>
7129
7130<h5>Overview:</h5>
7131<p>The '<tt>llvm.invariant.end</tt>' intrinsic specifies that the contents of
7132 a memory object are mutable.</p>
7133
7134<h5>Arguments:</h5>
7135<p>The first argument is the matching <tt>llvm.invariant.start</tt> intrinsic.
Nick Lewycky23f49982009-10-13 07:57:33 +0000[diff] [blame]7136 The second argument is a constant integer representing the size of the
7137 object, or -1 if it is variable sized and the third argument is a pointer
7138 to the object.</p>
Nick Lewyckyc888d352009-10-13 07:03:23 +0000[diff] [blame]7139
7140<h5>Semantics:</h5>
7141<p>This intrinsic indicates that the memory is mutable again.</p>
7142
7143</div>
7144
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]7145<!-- ======================================================================= -->
7146<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7147 <a name="int_general">General Intrinsics</a>
7148</div>
7149
7150<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7151
7152<p>This class of intrinsics is designed to be generic and has no specific
7153 purpose.</p>
7154
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7155</div>
7156
7157<!-- _______________________________________________________________________ -->
7158<div class="doc_subsubsection">
7159 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
7160</div>
7161
7162<div class="doc_text">
7163
7164<h5>Syntax:</h5>
7165<pre>
7166 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7167</pre>
7168
7169<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7170<p>The '<tt>llvm.var.annotation</tt>' intrinsic.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7171
7172<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7173<p>The first argument is a pointer to a value, the second is a pointer to a
7174 global string, the third is a pointer to a global string which is the source
7175 file name, and the last argument is the line number.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7176
7177<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7178<p>This intrinsic allows annotation of local variables with arbitrary strings.
7179 This can be useful for special purpose optimizations that want to look for
7180 these annotations. These have no other defined use, they are ignored by code
7181 generation and optimization.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7182
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7183</div>
7184
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7185<!-- _______________________________________________________________________ -->
7186<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]7187 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7188</div>
7189
7190<div class="doc_text">
7191
7192<h5>Syntax:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7193<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7194 any integer bit width.</p>
7195
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7196<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]7197 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7198 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7199 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7200 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7201 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]7202</pre>
7203
7204<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7205<p>The '<tt>llvm.annotation</tt>' intrinsic.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7206
7207<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7208<p>The first argument is an integer value (result of some expression), the
7209 second is a pointer to a global string, the third is a pointer to a global
7210 string which is the source file name, and the last argument is the line
7211 number. It returns the value of the first argument.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7212
7213<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7214<p>This intrinsic allows annotations to be put on arbitrary expressions with
7215 arbitrary strings. This can be useful for special purpose optimizations that
7216 want to look for these annotations. These have no other defined use, they
7217 are ignored by code generation and optimization.</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7218
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7219</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7220
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7221<!-- _______________________________________________________________________ -->
7222<div class="doc_subsubsection">
7223 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7224</div>
7225
7226<div class="doc_text">
7227
7228<h5>Syntax:</h5>
7229<pre>
7230 declare void @llvm.trap()
7231</pre>
7232
7233<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7234<p>The '<tt>llvm.trap</tt>' intrinsic.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7235
7236<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7237<p>None.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7238
7239<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7240<p>This intrinsics is lowered to the target dependent trap instruction. If the
7241 target does not have a trap instruction, this intrinsic will be lowered to
7242 the call of the <tt>abort()</tt> function.</p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7243
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7244</div>
7245
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7246<!-- _______________________________________________________________________ -->
7247<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]7248 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7249</div>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7250
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7251<div class="doc_text">
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7252
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7253<h5>Syntax:</h5>
7254<pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7255 declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7256</pre>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7257
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7258<h5>Overview:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7259<p>The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and
7260 stores it onto the stack at <tt>slot</tt>. The stack slot is adjusted to
7261 ensure that it is placed on the stack before local variables.</p>
7262
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7263<h5>Arguments:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7264<p>The <tt>llvm.stackprotector</tt> intrinsic requires two pointer
7265 arguments. The first argument is the value loaded from the stack
7266 guard <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt>
7267 that has enough space to hold the value of the guard.</p>
7268
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7269<h5>Semantics:</h5>
Bill Wendlingf85859d2009-07-20 02:29:24 +0000[diff] [blame]7270<p>This intrinsic causes the prologue/epilogue inserter to force the position of
7271 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7272 stack. This is to ensure that if a local variable on the stack is
7273 overwritten, it will destroy the value of the guard. When the function exits,
7274 the guard on the stack is checked against the original guard. If they're
7275 different, then the program aborts by calling the <tt>__stack_chk_fail()</tt>
7276 function.</p>
7277
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7278</div>
7279
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7280<!-- _______________________________________________________________________ -->
7281<div class="doc_subsubsection">
7282 <a name="int_objectsize">'<tt>llvm.objectsize</tt>' Intrinsic</a>
7283</div>
7284
7285<div class="doc_text">
7286
7287<h5>Syntax:</h5>
7288<pre>
Eric Christopher0101f9d2009-12-23 00:29:49 +0000[diff] [blame]7289 declare i32 @llvm.objectsize.i32( i8* &lt;object&gt;, i1 &lt;type&gt; )
7290 declare i64 @llvm.objectsize.i64( i8* &lt;object&gt;, i1 &lt;type&gt; )
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7291</pre>
7292
7293<h5>Overview:</h5>
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7294<p>The <tt>llvm.objectsize</tt> intrinsic is designed to provide information
Eric Christopher184f44d2010-01-08 21:42:39 +0000[diff] [blame]7295 to the optimizers to discover at compile time either a) when an
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7296 operation like memcpy will either overflow a buffer that corresponds to
7297 an object, or b) to determine that a runtime check for overflow isn't
7298 necessary. An object in this context means an allocation of a
Eric Christopher0101f9d2009-12-23 00:29:49 +0000[diff] [blame]7299 specific class, structure, array, or other object.</p>
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7300
7301<h5>Arguments:</h5>
7302<p>The <tt>llvm.objectsize</tt> intrinsic takes two arguments. The first
Eric Christopher0101f9d2009-12-23 00:29:49 +0000[diff] [blame]7303 argument is a pointer to or into the <tt>object</tt>. The second argument
7304 is a boolean 0 or 1. This argument determines whether you want the
7305 maximum (0) or minimum (1) bytes remaining. This needs to be a literal 0 or
7306 1, variables are not allowed.</p>
7307
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7308<h5>Semantics:</h5>
7309<p>The <tt>llvm.objectsize</tt> intrinsic is lowered to either a constant
Eric Christophera1151bf2009-12-05 02:46:03 +0000[diff] [blame]7310 representing the size of the object concerned or <tt>i32/i64 -1 or 0</tt>
7311 (depending on the <tt>type</tt> argument if the size cannot be determined
7312 at compile time.</p>
Eric Christopher767a3722009-11-30 08:03:53 +0000[diff] [blame]7313
7314</div>
7315
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7316<!-- *********************************************************************** -->
7317<hr>
7318<address>
7319 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7320 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7321 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7322 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]7323
7324 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7325 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7326 Last modified: $Date$
7327</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]7328
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7329</body>
7330</html>