Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 57b06bac32bb02f532fb4722a258b71896dfefd1 / . / docs / LangRef.html
blob: 416b105cd7286d436092b25f3099b7205482933e [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>LLVM Assembly Language Reference Manual</title>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <meta name="author" content="Chris Lattner">
8 <meta name="description"
9 content="LLVM Assembly Language Reference Manual.">
10 <link rel="stylesheet" href="llvm.css" type="text/css">
11</head>
12
13<body>
14
15<div class="doc_title"> LLVM Language Reference Manual </div>
16<ol>
17 <li><a href="#abstract">Abstract</a></li>
18 <li><a href="#introduction">Introduction</a></li>
19 <li><a href="#identifiers">Identifiers</a></li>
20 <li><a href="#highlevel">High Level Structure</a>
21 <ol>
22 <li><a href="#modulestructure">Module Structure</a></li>
23 <li><a href="#linkage">Linkage Types</a></li>
24 <li><a href="#callingconv">Calling Conventions</a></li>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]25 <li><a href="#namedtypes">Named Types</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]26 <li><a href="#globalvars">Global Variables</a></li>
27 <li><a href="#functionstructure">Functions</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]28 <li><a href="#aliasstructure">Aliases</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]29 <li><a href="#paramattrs">Parameter Attributes</a></li>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]30 <li><a href="#fnattrs">Function Attributes</a></li>
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]31 <li><a href="#gc">Garbage Collector Names</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]32 <li><a href="#moduleasm">Module-Level Inline Assembly</a></li>
33 <li><a href="#datalayout">Data Layout</a></li>
34 </ol>
35 </li>
36 <li><a href="#typesystem">Type System</a>
37 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]38 <li><a href="#t_classifications">Type Classifications</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]39 <li><a href="#t_primitive">Primitive Types</a>
40 <ol>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]41 <li><a href="#t_floating">Floating Point Types</a></li>
42 <li><a href="#t_void">Void Type</a></li>
43 <li><a href="#t_label">Label Type</a></li>
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]44 <li><a href="#t_metadata">Metadata Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]45 </ol>
46 </li>
47 <li><a href="#t_derived">Derived Types</a>
48 <ol>
Chris Lattner251ab812007-12-18 06:18:21 +0000[diff] [blame]49 <li><a href="#t_integer">Integer Type</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]50 <li><a href="#t_array">Array Type</a></li>
51 <li><a href="#t_function">Function Type</a></li>
52 <li><a href="#t_pointer">Pointer Type</a></li>
53 <li><a href="#t_struct">Structure Type</a></li>
54 <li><a href="#t_pstruct">Packed Structure Type</a></li>
55 <li><a href="#t_vector">Vector Type</a></li>
56 <li><a href="#t_opaque">Opaque Type</a></li>
57 </ol>
58 </li>
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]59 <li><a href="#t_uprefs">Type Up-references</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]60 </ol>
61 </li>
62 <li><a href="#constants">Constants</a>
63 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]64 <li><a href="#simpleconstants">Simple Constants</a></li>
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]65 <li><a href="#complexconstants">Complex Constants</a></li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]66 <li><a href="#globalconstants">Global Variable and Function Addresses</a></li>
67 <li><a href="#undefvalues">Undefined Values</a></li>
68 <li><a href="#constantexprs">Constant Expressions</a></li>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]69 <li><a href="#metadata">Embedded Metadata</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]70 </ol>
71 </li>
72 <li><a href="#othervalues">Other Values</a>
73 <ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]74 <li><a href="#inlineasm">Inline Assembler Expressions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]75 </ol>
76 </li>
77 <li><a href="#instref">Instruction Reference</a>
78 <ol>
79 <li><a href="#terminators">Terminator Instructions</a>
80 <ol>
81 <li><a href="#i_ret">'<tt>ret</tt>' Instruction</a></li>
82 <li><a href="#i_br">'<tt>br</tt>' Instruction</a></li>
83 <li><a href="#i_switch">'<tt>switch</tt>' Instruction</a></li>
84 <li><a href="#i_invoke">'<tt>invoke</tt>' Instruction</a></li>
85 <li><a href="#i_unwind">'<tt>unwind</tt>' Instruction</a></li>
86 <li><a href="#i_unreachable">'<tt>unreachable</tt>' Instruction</a></li>
87 </ol>
88 </li>
89 <li><a href="#binaryops">Binary Operations</a>
90 <ol>
91 <li><a href="#i_add">'<tt>add</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]92 <li><a href="#i_fadd">'<tt>fadd</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]93 <li><a href="#i_sub">'<tt>sub</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]94 <li><a href="#i_fsub">'<tt>fsub</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]95 <li><a href="#i_mul">'<tt>mul</tt>' Instruction</a></li>
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]96 <li><a href="#i_fmul">'<tt>fmul</tt>' Instruction</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]97 <li><a href="#i_udiv">'<tt>udiv</tt>' Instruction</a></li>
98 <li><a href="#i_sdiv">'<tt>sdiv</tt>' Instruction</a></li>
99 <li><a href="#i_fdiv">'<tt>fdiv</tt>' Instruction</a></li>
100 <li><a href="#i_urem">'<tt>urem</tt>' Instruction</a></li>
101 <li><a href="#i_srem">'<tt>srem</tt>' Instruction</a></li>
102 <li><a href="#i_frem">'<tt>frem</tt>' Instruction</a></li>
103 </ol>
104 </li>
105 <li><a href="#bitwiseops">Bitwise Binary Operations</a>
106 <ol>
107 <li><a href="#i_shl">'<tt>shl</tt>' Instruction</a></li>
108 <li><a href="#i_lshr">'<tt>lshr</tt>' Instruction</a></li>
109 <li><a href="#i_ashr">'<tt>ashr</tt>' Instruction</a></li>
110 <li><a href="#i_and">'<tt>and</tt>' Instruction</a></li>
111 <li><a href="#i_or">'<tt>or</tt>' Instruction</a></li>
112 <li><a href="#i_xor">'<tt>xor</tt>' Instruction</a></li>
113 </ol>
114 </li>
115 <li><a href="#vectorops">Vector Operations</a>
116 <ol>
117 <li><a href="#i_extractelement">'<tt>extractelement</tt>' Instruction</a></li>
118 <li><a href="#i_insertelement">'<tt>insertelement</tt>' Instruction</a></li>
119 <li><a href="#i_shufflevector">'<tt>shufflevector</tt>' Instruction</a></li>
120 </ol>
121 </li>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]122 <li><a href="#aggregateops">Aggregate Operations</a>
123 <ol>
124 <li><a href="#i_extractvalue">'<tt>extractvalue</tt>' Instruction</a></li>
125 <li><a href="#i_insertvalue">'<tt>insertvalue</tt>' Instruction</a></li>
126 </ol>
127 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]128 <li><a href="#memoryops">Memory Access and Addressing Operations</a>
129 <ol>
130 <li><a href="#i_malloc">'<tt>malloc</tt>' Instruction</a></li>
131 <li><a href="#i_free">'<tt>free</tt>' Instruction</a></li>
132 <li><a href="#i_alloca">'<tt>alloca</tt>' Instruction</a></li>
133 <li><a href="#i_load">'<tt>load</tt>' Instruction</a></li>
134 <li><a href="#i_store">'<tt>store</tt>' Instruction</a></li>
135 <li><a href="#i_getelementptr">'<tt>getelementptr</tt>' Instruction</a></li>
136 </ol>
137 </li>
138 <li><a href="#convertops">Conversion Operations</a>
139 <ol>
140 <li><a href="#i_trunc">'<tt>trunc .. to</tt>' Instruction</a></li>
141 <li><a href="#i_zext">'<tt>zext .. to</tt>' Instruction</a></li>
142 <li><a href="#i_sext">'<tt>sext .. to</tt>' Instruction</a></li>
143 <li><a href="#i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a></li>
144 <li><a href="#i_fpext">'<tt>fpext .. to</tt>' Instruction</a></li>
145 <li><a href="#i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a></li>
146 <li><a href="#i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a></li>
147 <li><a href="#i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a></li>
148 <li><a href="#i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a></li>
149 <li><a href="#i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a></li>
150 <li><a href="#i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a></li>
151 <li><a href="#i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a></li>
152 </ol>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]153 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]154 <li><a href="#otherops">Other Operations</a>
155 <ol>
156 <li><a href="#i_icmp">'<tt>icmp</tt>' Instruction</a></li>
157 <li><a href="#i_fcmp">'<tt>fcmp</tt>' Instruction</a></li>
158 <li><a href="#i_phi">'<tt>phi</tt>' Instruction</a></li>
159 <li><a href="#i_select">'<tt>select</tt>' Instruction</a></li>
160 <li><a href="#i_call">'<tt>call</tt>' Instruction</a></li>
161 <li><a href="#i_va_arg">'<tt>va_arg</tt>' Instruction</a></li>
162 </ol>
163 </li>
164 </ol>
165 </li>
166 <li><a href="#intrinsics">Intrinsic Functions</a>
167 <ol>
168 <li><a href="#int_varargs">Variable Argument Handling Intrinsics</a>
169 <ol>
170 <li><a href="#int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a></li>
171 <li><a href="#int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a></li>
172 <li><a href="#int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a></li>
173 </ol>
174 </li>
175 <li><a href="#int_gc">Accurate Garbage Collection Intrinsics</a>
176 <ol>
177 <li><a href="#int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a></li>
178 <li><a href="#int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a></li>
179 <li><a href="#int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a></li>
180 </ol>
181 </li>
182 <li><a href="#int_codegen">Code Generator Intrinsics</a>
183 <ol>
184 <li><a href="#int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a></li>
185 <li><a href="#int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a></li>
186 <li><a href="#int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a></li>
187 <li><a href="#int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a></li>
188 <li><a href="#int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a></li>
189 <li><a href="#int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a></li>
190 <li><a href="#int_readcyclecounter"><tt>llvm.readcyclecounter</tt>' Intrinsic</a></li>
191 </ol>
192 </li>
193 <li><a href="#int_libc">Standard C Library Intrinsics</a>
194 <ol>
195 <li><a href="#int_memcpy">'<tt>llvm.memcpy.*</tt>' Intrinsic</a></li>
196 <li><a href="#int_memmove">'<tt>llvm.memmove.*</tt>' Intrinsic</a></li>
197 <li><a href="#int_memset">'<tt>llvm.memset.*</tt>' Intrinsic</a></li>
198 <li><a href="#int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a></li>
199 <li><a href="#int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a></li>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]200 <li><a href="#int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a></li>
201 <li><a href="#int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a></li>
202 <li><a href="#int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]203 </ol>
204 </li>
205 <li><a href="#int_manip">Bit Manipulation Intrinsics</a>
206 <ol>
207 <li><a href="#int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a></li>
208 <li><a href="#int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic </a></li>
209 <li><a href="#int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic </a></li>
210 <li><a href="#int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic </a></li>
211 <li><a href="#int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic </a></li>
212 <li><a href="#int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic </a></li>
213 </ol>
214 </li>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]215 <li><a href="#int_overflow">Arithmetic with Overflow Intrinsics</a>
216 <ol>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]217 <li><a href="#int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt> Intrinsics</a></li>
218 <li><a href="#int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt> Intrinsics</a></li>
219 <li><a href="#int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt> Intrinsics</a></li>
220 <li><a href="#int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt> Intrinsics</a></li>
221 <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]222 <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]223 </ol>
224 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]225 <li><a href="#int_debugger">Debugger intrinsics</a></li>
226 <li><a href="#int_eh">Exception Handling intrinsics</a></li>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]227 <li><a href="#int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]228 <ol>
229 <li><a href="#int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a></li>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]230 </ol>
231 </li>
Bill Wendling9127adb2008-11-18 22:10:53 +0000[diff] [blame]232 <li><a href="#int_atomics">Atomic intrinsics</a>
233 <ol>
234 <li><a href="#int_memory_barrier"><tt>llvm.memory_barrier</tt></a></li>
235 <li><a href="#int_atomic_cmp_swap"><tt>llvm.atomic.cmp.swap</tt></a></li>
236 <li><a href="#int_atomic_swap"><tt>llvm.atomic.swap</tt></a></li>
237 <li><a href="#int_atomic_load_add"><tt>llvm.atomic.load.add</tt></a></li>
238 <li><a href="#int_atomic_load_sub"><tt>llvm.atomic.load.sub</tt></a></li>
239 <li><a href="#int_atomic_load_and"><tt>llvm.atomic.load.and</tt></a></li>
240 <li><a href="#int_atomic_load_nand"><tt>llvm.atomic.load.nand</tt></a></li>
241 <li><a href="#int_atomic_load_or"><tt>llvm.atomic.load.or</tt></a></li>
242 <li><a href="#int_atomic_load_xor"><tt>llvm.atomic.load.xor</tt></a></li>
243 <li><a href="#int_atomic_load_max"><tt>llvm.atomic.load.max</tt></a></li>
244 <li><a href="#int_atomic_load_min"><tt>llvm.atomic.load.min</tt></a></li>
245 <li><a href="#int_atomic_load_umax"><tt>llvm.atomic.load.umax</tt></a></li>
246 <li><a href="#int_atomic_load_umin"><tt>llvm.atomic.load.umin</tt></a></li>
247 </ol>
248 </li>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]249 <li><a href="#int_general">General intrinsics</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]250 <ol>
Reid Spencerb043f672007-07-20 19:59:11 +0000[diff] [blame]251 <li><a href="#int_var_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]252 '<tt>llvm.var.annotation</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]253 <li><a href="#int_annotation">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]254 '<tt>llvm.annotation.*</tt>' Intrinsic</a></li>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]255 <li><a href="#int_trap">
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]256 '<tt>llvm.trap</tt>' Intrinsic</a></li>
257 <li><a href="#int_stackprotector">
258 '<tt>llvm.stackprotector</tt>' Intrinsic</a></li>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]259 </ol>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]260 </li>
261 </ol>
262 </li>
263</ol>
264
265<div class="doc_author">
266 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
267 and <a href="mailto:vadve@cs.uiuc.edu">Vikram Adve</a></p>
268</div>
269
270<!-- *********************************************************************** -->
271<div class="doc_section"> <a name="abstract">Abstract </a></div>
272<!-- *********************************************************************** -->
273
274<div class="doc_text">
275<p>This document is a reference manual for the LLVM assembly language.
Bill Wendlinge7846a52008-08-05 22:29:16 +0000[diff] [blame]276LLVM is a Static Single Assignment (SSA) based representation that provides
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]277type safety, low-level operations, flexibility, and the capability of
278representing 'all' high-level languages cleanly. It is the common code
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]279representation used throughout all phases of the LLVM compilation
280strategy.</p>
281</div>
282
283<!-- *********************************************************************** -->
284<div class="doc_section"> <a name="introduction">Introduction</a> </div>
285<!-- *********************************************************************** -->
286
287<div class="doc_text">
288
289<p>The LLVM code representation is designed to be used in three
290different forms: as an in-memory compiler IR, as an on-disk bitcode
291representation (suitable for fast loading by a Just-In-Time compiler),
292and as a human readable assembly language representation. This allows
293LLVM to provide a powerful intermediate representation for efficient
294compiler transformations and analysis, while providing a natural means
295to debug and visualize the transformations. The three different forms
296of LLVM are all equivalent. This document describes the human readable
297representation and notation.</p>
298
299<p>The LLVM representation aims to be light-weight and low-level
300while being expressive, typed, and extensible at the same time. It
301aims to be a "universal IR" of sorts, by being at a low enough level
302that high-level ideas may be cleanly mapped to it (similar to how
303microprocessors are "universal IR's", allowing many source languages to
304be mapped to them). By providing type information, LLVM can be used as
305the target of optimizations: for example, through pointer analysis, it
306can be proven that a C automatic variable is never accessed outside of
307the current function... allowing it to be promoted to a simple SSA
308value instead of a memory location.</p>
309
310</div>
311
312<!-- _______________________________________________________________________ -->
313<div class="doc_subsubsection"> <a name="wellformed">Well-Formedness</a> </div>
314
315<div class="doc_text">
316
317<p>It is important to note that this document describes 'well formed'
318LLVM assembly language. There is a difference between what the parser
319accepts and what is considered 'well formed'. For example, the
320following instruction is syntactically okay, but not well formed:</p>
321
322<div class="doc_code">
323<pre>
324%x = <a href="#i_add">add</a> i32 1, %x
325</pre>
326</div>
327
328<p>...because the definition of <tt>%x</tt> does not dominate all of
329its uses. The LLVM infrastructure provides a verification pass that may
330be used to verify that an LLVM module is well formed. This pass is
331automatically run by the parser after parsing input assembly and by
332the optimizer before it outputs bitcode. The violations pointed out
333by the verifier pass indicate bugs in transformation passes or input to
334the parser.</p>
335</div>
336
Chris Lattnera83fdc02007-10-03 17:34:29 +0000[diff] [blame]337<!-- Describe the typesetting conventions here. -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]338
339<!-- *********************************************************************** -->
340<div class="doc_section"> <a name="identifiers">Identifiers</a> </div>
341<!-- *********************************************************************** -->
342
343<div class="doc_text">
344
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]345 <p>LLVM identifiers come in two basic types: global and local. Global
346 identifiers (functions, global variables) begin with the @ character. Local
347 identifiers (register names, types) begin with the % character. Additionally,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]348 there are three different formats for identifiers, for different purposes:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]349
350<ol>
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]351 <li>Named values are represented as a string of characters with their prefix.
352 For example, %foo, @DivisionByZero, %a.really.long.identifier. The actual
353 regular expression used is '<tt>[%@][a-zA-Z$._][a-zA-Z$._0-9]*</tt>'.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]354 Identifiers which require other characters in their names can be surrounded
Daniel Dunbar73f9a772008-10-14 23:51:43 +0000[diff] [blame]355 with quotes. Special characters may be escaped using "\xx" where xx is the
356 ASCII code for the character in hexadecimal. In this way, any character can
357 be used in a name value, even quotes themselves.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]358
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]359 <li>Unnamed values are represented as an unsigned numeric value with their
360 prefix. For example, %12, @2, %44.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]361
362 <li>Constants, which are described in a <a href="#constants">section about
363 constants</a>, below.</li>
364</ol>
365
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]366<p>LLVM requires that values start with a prefix for two reasons: Compilers
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]367don't need to worry about name clashes with reserved words, and the set of
368reserved words may be expanded in the future without penalty. Additionally,
369unnamed identifiers allow a compiler to quickly come up with a temporary
370variable without having to avoid symbol table conflicts.</p>
371
372<p>Reserved words in LLVM are very similar to reserved words in other
373languages. There are keywords for different opcodes
374('<tt><a href="#i_add">add</a></tt>',
375 '<tt><a href="#i_bitcast">bitcast</a></tt>',
376 '<tt><a href="#i_ret">ret</a></tt>', etc...), for primitive type names ('<tt><a
377href="#t_void">void</a></tt>', '<tt><a href="#t_primitive">i32</a></tt>', etc...),
378and others. These reserved words cannot conflict with variable names, because
Reid Spencerc8245b02007-08-07 14:34:28 +0000[diff] [blame]379none of them start with a prefix character ('%' or '@').</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]380
381<p>Here is an example of LLVM code to multiply the integer variable
382'<tt>%X</tt>' by 8:</p>
383
384<p>The easy way:</p>
385
386<div class="doc_code">
387<pre>
388%result = <a href="#i_mul">mul</a> i32 %X, 8
389</pre>
390</div>
391
392<p>After strength reduction:</p>
393
394<div class="doc_code">
395<pre>
396%result = <a href="#i_shl">shl</a> i32 %X, i8 3
397</pre>
398</div>
399
400<p>And the hard way:</p>
401
402<div class="doc_code">
403<pre>
404<a href="#i_add">add</a> i32 %X, %X <i>; yields {i32}:%0</i>
405<a href="#i_add">add</a> i32 %0, %0 <i>; yields {i32}:%1</i>
406%result = <a href="#i_add">add</a> i32 %1, %1
407</pre>
408</div>
409
410<p>This last way of multiplying <tt>%X</tt> by 8 illustrates several
411important lexical features of LLVM:</p>
412
413<ol>
414
415 <li>Comments are delimited with a '<tt>;</tt>' and go until the end of
416 line.</li>
417
418 <li>Unnamed temporaries are created when the result of a computation is not
419 assigned to a named value.</li>
420
421 <li>Unnamed temporaries are numbered sequentially</li>
422
423</ol>
424
425<p>...and it also shows a convention that we follow in this document. When
426demonstrating instructions, we will follow an instruction with a comment that
427defines the type and name of value produced. Comments are shown in italic
428text.</p>
429
430</div>
431
432<!-- *********************************************************************** -->
433<div class="doc_section"> <a name="highlevel">High Level Structure</a> </div>
434<!-- *********************************************************************** -->
435
436<!-- ======================================================================= -->
437<div class="doc_subsection"> <a name="modulestructure">Module Structure</a>
438</div>
439
440<div class="doc_text">
441
442<p>LLVM programs are composed of "Module"s, each of which is a
443translation unit of the input programs. Each module consists of
444functions, global variables, and symbol table entries. Modules may be
445combined together with the LLVM linker, which merges function (and
446global variable) definitions, resolves forward declarations, and merges
447symbol table entries. Here is an example of the "hello world" module:</p>
448
449<div class="doc_code">
450<pre><i>; Declare the string constant as a global constant...</i>
451<a href="#identifiers">@.LC0</a> = <a href="#linkage_internal">internal</a> <a
452 href="#globalvars">constant</a> <a href="#t_array">[13 x i8]</a> c"hello world\0A\00" <i>; [13 x i8]*</i>
453
454<i>; External declaration of the puts function</i>
455<a href="#functionstructure">declare</a> i32 @puts(i8 *) <i>; i32(i8 *)* </i>
456
457<i>; Definition of main function</i>
458define i32 @main() { <i>; i32()* </i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]459 <i>; Convert [13 x i8]* to i8 *...</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]460 %cast210 = <a
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]461 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]462
463 <i>; Call puts function to write out the string to stdout...</i>
464 <a
465 href="#i_call">call</a> i32 @puts(i8 * %cast210) <i>; i32</i>
466 <a
467 href="#i_ret">ret</a> i32 0<br>}<br>
468</pre>
469</div>
470
471<p>This example is made up of a <a href="#globalvars">global variable</a>
472named "<tt>.LC0</tt>", an external declaration of the "<tt>puts</tt>"
473function, and a <a href="#functionstructure">function definition</a>
474for "<tt>main</tt>".</p>
475
476<p>In general, a module is made up of a list of global values,
477where both functions and global variables are global values. Global values are
478represented by a pointer to a memory location (in this case, a pointer to an
479array of char, and a pointer to a function), and have one of the following <a
480href="#linkage">linkage types</a>.</p>
481
482</div>
483
484<!-- ======================================================================= -->
485<div class="doc_subsection">
486 <a name="linkage">Linkage Types</a>
487</div>
488
489<div class="doc_text">
490
491<p>
492All Global Variables and Functions have one of the following types of linkage:
493</p>
494
495<dl>
496
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]497 <dt><tt><b><a name="linkage_private">private</a></b></tt>: </dt>
498
499 <dd>Global values with private linkage are only directly accessible by
500 objects in the current module. In particular, linking code into a module with
501 an private global value may cause the private to be renamed as necessary to
502 avoid collisions. Because the symbol is private to the module, all
503 references can be updated. This doesn't show up in any symbol table in the
504 object file.
505 </dd>
506
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]507 <dt><tt><b><a name="linkage_internal">internal</a></b></tt>: </dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]508
Duncan Sandsa75223a2009-01-16 09:29:46 +0000[diff] [blame]509 <dd> Similar to private, but the value shows as a local symbol (STB_LOCAL in
Rafael Espindolaa168fc92009-01-15 20:18:42 +0000[diff] [blame]510 the case of ELF) in the object file. This corresponds to the notion of the
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]511 '<tt>static</tt>' keyword in C.
512 </dd>
513
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]514 <dt><tt><b><a name="available_externally">available_externally</a></b></tt>:
515 </dt>
516
517 <dd>Globals with "<tt>available_externally</tt>" linkage are never emitted
518 into the object file corresponding to the LLVM module. They exist to
519 allow inlining and other optimizations to take place given knowledge of the
520 definition of the global, which is known to be somewhere outside the module.
521 Globals with <tt>available_externally</tt> linkage are allowed to be discarded
522 at will, and are otherwise the same as <tt>linkonce_odr</tt>. This linkage
523 type is only allowed on definitions, not declarations.</dd>
524
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]525 <dt><tt><b><a name="linkage_linkonce">linkonce</a></b></tt>: </dt>
526
527 <dd>Globals with "<tt>linkonce</tt>" linkage are merged with other globals of
528 the same name when linkage occurs. This is typically used to implement
529 inline functions, templates, or other code which must be generated in each
530 translation unit that uses it. Unreferenced <tt>linkonce</tt> globals are
531 allowed to be discarded.
532 </dd>
533
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]534 <dt><tt><b><a name="linkage_common">common</a></b></tt>: </dt>
535
536 <dd>"<tt>common</tt>" linkage is exactly the same as <tt>linkonce</tt>
537 linkage, except that unreferenced <tt>common</tt> globals may not be
538 discarded. This is used for globals that may be emitted in multiple
539 translation units, but that are not guaranteed to be emitted into every
540 translation unit that uses them. One example of this is tentative
541 definitions in C, such as "<tt>int X;</tt>" at global scope.
542 </dd>
543
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]544 <dt><tt><b><a name="linkage_weak">weak</a></b></tt>: </dt>
545
Dale Johannesen96e7e092008-05-23 23:13:41 +0000[diff] [blame]546 <dd>"<tt>weak</tt>" linkage is the same as <tt>common</tt> linkage, except
547 that some targets may choose to emit different assembly sequences for them
548 for target-dependent reasons. This is used for globals that are declared
549 "weak" in C source code.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]550 </dd>
551
552 <dt><tt><b><a name="linkage_appending">appending</a></b></tt>: </dt>
553
554 <dd>"<tt>appending</tt>" linkage may only be applied to global variables of
555 pointer to array type. When two global variables with appending linkage are
556 linked together, the two global arrays are appended together. This is the
557 LLVM, typesafe, equivalent of having the system linker append together
558 "sections" with identical names when .o files are linked.
559 </dd>
560
561 <dt><tt><b><a name="linkage_externweak">extern_weak</a></b></tt>: </dt>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]562
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]563 <dd>The semantics of this linkage follow the ELF object file model: the
564 symbol is weak until linked, if not linked, the symbol becomes null instead
565 of being an undefined reference.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]566 </dd>
567
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]568 <dt><tt><b><a name="linkage_linkonce">linkonce_odr</a></b></tt>: </dt>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]569 <dt><tt><b><a name="linkage_weak">weak_odr</a></b></tt>: </dt>
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]570 <dd>Some languages allow differing globals to be merged, such as two
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]571 functions with different semantics. Other languages, such as <tt>C++</tt>,
572 ensure that only equivalent globals are ever merged (the "one definition
Chris Lattner68433442009-04-13 05:44:34 +0000[diff] [blame]573 rule" - "ODR"). Such languages can use the <tt>linkonce_odr</tt>
Duncan Sandsb95df792009-03-11 20:14:15 +0000[diff] [blame]574 and <tt>weak_odr</tt> linkage types to indicate that the global will only
575 be merged with equivalent globals. These linkage types are otherwise the
576 same as their non-<tt>odr</tt> versions.
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]577 </dd>
578
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]579 <dt><tt><b><a name="linkage_external">externally visible</a></b></tt>:</dt>
580
581 <dd>If none of the above identifiers are used, the global is externally
582 visible, meaning that it participates in linkage and can be used to resolve
583 external symbol references.
584 </dd>
585</dl>
586
587 <p>
588 The next two types of linkage are targeted for Microsoft Windows platform
589 only. They are designed to support importing (exporting) symbols from (to)
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]590 DLLs (Dynamic Link Libraries).
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]591 </p>
592
593 <dl>
594 <dt><tt><b><a name="linkage_dllimport">dllimport</a></b></tt>: </dt>
595
596 <dd>"<tt>dllimport</tt>" linkage causes the compiler to reference a function
597 or variable via a global pointer to a pointer that is set up by the DLL
598 exporting the symbol. On Microsoft Windows targets, the pointer name is
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]599 formed by combining <code>__imp_</code> and the function or variable name.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]600 </dd>
601
602 <dt><tt><b><a name="linkage_dllexport">dllexport</a></b></tt>: </dt>
603
604 <dd>"<tt>dllexport</tt>" linkage causes the compiler to provide a global
605 pointer to a pointer in a DLL, so that it can be referenced with the
606 <tt>dllimport</tt> attribute. On Microsoft Windows targets, the pointer
Dan Gohman5ec99832009-01-12 21:35:55 +0000[diff] [blame]607 name is formed by combining <code>__imp_</code> and the function or variable
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]608 name.
609 </dd>
610
611</dl>
612
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]613<p>For example, since the "<tt>.LC0</tt>"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]614variable is defined to be internal, if another module defined a "<tt>.LC0</tt>"
615variable and was linked with this one, one of the two would be renamed,
616preventing a collision. Since "<tt>main</tt>" and "<tt>puts</tt>" are
617external (i.e., lacking any linkage declarations), they are accessible
618outside of the current module.</p>
619<p>It is illegal for a function <i>declaration</i>
Duncan Sands565f65d2009-03-11 08:08:06 +0000[diff] [blame]620to have any linkage type other than "externally visible", <tt>dllimport</tt>
621or <tt>extern_weak</tt>.</p>
Duncan Sands19d161f2009-03-07 15:45:40 +0000[diff] [blame]622<p>Aliases can have only <tt>external</tt>, <tt>internal</tt>, <tt>weak</tt>
623or <tt>weak_odr</tt> linkages.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]624</div>
625
626<!-- ======================================================================= -->
627<div class="doc_subsection">
628 <a name="callingconv">Calling Conventions</a>
629</div>
630
631<div class="doc_text">
632
633<p>LLVM <a href="#functionstructure">functions</a>, <a href="#i_call">calls</a>
634and <a href="#i_invoke">invokes</a> can all have an optional calling convention
635specified for the call. The calling convention of any pair of dynamic
636caller/callee must match, or the behavior of the program is undefined. The
637following calling conventions are supported by LLVM, and more may be added in
638the future:</p>
639
640<dl>
641 <dt><b>"<tt>ccc</tt>" - The C calling convention</b>:</dt>
642
643 <dd>This calling convention (the default if no other calling convention is
644 specified) matches the target C calling conventions. This calling convention
645 supports varargs function calls and tolerates some mismatch in the declared
646 prototype and implemented declaration of the function (as does normal C).
647 </dd>
648
649 <dt><b>"<tt>fastcc</tt>" - The fast calling convention</b>:</dt>
650
651 <dd>This calling convention attempts to make calls as fast as possible
652 (e.g. by passing things in registers). This calling convention allows the
653 target to use whatever tricks it wants to produce fast code for the target,
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]654 without having to conform to an externally specified ABI (Application Binary
655 Interface). Implementations of this convention should allow arbitrary
Arnold Schwaighofer07444922008-05-14 09:17:12 +0000[diff] [blame]656 <a href="CodeGenerator.html#tailcallopt">tail call optimization</a> to be
657 supported. This calling convention does not support varargs and requires the
658 prototype of all callees to exactly match the prototype of the function
659 definition.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]660 </dd>
661
662 <dt><b>"<tt>coldcc</tt>" - The cold calling convention</b>:</dt>
663
664 <dd>This calling convention attempts to make code in the caller as efficient
665 as possible under the assumption that the call is not commonly executed. As
666 such, these calls often preserve all registers so that the call does not break
667 any live ranges in the caller side. This calling convention does not support
668 varargs and requires the prototype of all callees to exactly match the
669 prototype of the function definition.
670 </dd>
671
672 <dt><b>"<tt>cc &lt;<em>n</em>&gt;</tt>" - Numbered convention</b>:</dt>
673
674 <dd>Any calling convention may be specified by number, allowing
675 target-specific calling conventions to be used. Target specific calling
676 conventions start at 64.
677 </dd>
678</dl>
679
680<p>More calling conventions can be added/defined on an as-needed basis, to
681support pascal conventions or any other well-known target-independent
682convention.</p>
683
684</div>
685
686<!-- ======================================================================= -->
687<div class="doc_subsection">
688 <a name="visibility">Visibility Styles</a>
689</div>
690
691<div class="doc_text">
692
693<p>
694All Global Variables and Functions have one of the following visibility styles:
695</p>
696
697<dl>
698 <dt><b>"<tt>default</tt>" - Default style</b>:</dt>
699
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]700 <dd>On targets that use the ELF object file format, default visibility means
701 that the declaration is visible to other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]702 modules and, in shared libraries, means that the declared entity may be
703 overridden. On Darwin, default visibility means that the declaration is
704 visible to other modules. Default visibility corresponds to "external
705 linkage" in the language.
706 </dd>
707
708 <dt><b>"<tt>hidden</tt>" - Hidden style</b>:</dt>
709
710 <dd>Two declarations of an object with hidden visibility refer to the same
711 object if they are in the same shared object. Usually, hidden visibility
712 indicates that the symbol will not be placed into the dynamic symbol table,
713 so no other module (executable or shared library) can reference it
714 directly.
715 </dd>
716
717 <dt><b>"<tt>protected</tt>" - Protected style</b>:</dt>
718
719 <dd>On ELF, protected visibility indicates that the symbol will be placed in
720 the dynamic symbol table, but that references within the defining module will
721 bind to the local symbol. That is, the symbol cannot be overridden by another
722 module.
723 </dd>
724</dl>
725
726</div>
727
728<!-- ======================================================================= -->
729<div class="doc_subsection">
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]730 <a name="namedtypes">Named Types</a>
731</div>
732
733<div class="doc_text">
734
735<p>LLVM IR allows you to specify name aliases for certain types. This can make
736it easier to read the IR and make the IR more condensed (particularly when
737recursive types are involved). An example of a name specification is:
738</p>
739
740<div class="doc_code">
741<pre>
742%mytype = type { %mytype*, i32 }
743</pre>
744</div>
745
746<p>You may give a name to any <a href="#typesystem">type</a> except "<a
747href="t_void">void</a>". Type name aliases may be used anywhere a type is
748expected with the syntax "%mytype".</p>
749
750<p>Note that type names are aliases for the structural type that they indicate,
751and that you can therefore specify multiple names for the same type. This often
752leads to confusing behavior when dumping out a .ll file. Since LLVM IR uses
753structural typing, the name is not part of the type. When printing out LLVM IR,
754the printer will pick <em>one name</em> to render all types of a particular
755shape. This means that if you have code where two different source types end up
756having the same LLVM type, that the dumper will sometimes print the "wrong" or
757unexpected type. This is an important design point and isn't going to
758change.</p>
759
760</div>
761
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]762<!-- ======================================================================= -->
763<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]764 <a name="globalvars">Global Variables</a>
765</div>
766
767<div class="doc_text">
768
769<p>Global variables define regions of memory allocated at compilation time
770instead of run-time. Global variables may optionally be initialized, may have
771an explicit section to be placed in, and may have an optional explicit alignment
772specified. A variable may be defined as "thread_local", which means that it
773will not be shared by threads (each thread will have a separated copy of the
774variable). A variable may be defined as a global "constant," which indicates
775that the contents of the variable will <b>never</b> be modified (enabling better
776optimization, allowing the global data to be placed in the read-only section of
777an executable, etc). Note that variables that need runtime initialization
778cannot be marked "constant" as there is a store to the variable.</p>
779
780<p>
781LLVM explicitly allows <em>declarations</em> of global variables to be marked
782constant, even if the final definition of the global is not. This capability
783can be used to enable slightly better optimization of the program, but requires
784the language definition to guarantee that optimizations based on the
785'constantness' are valid for the translation units that do not include the
786definition.
787</p>
788
789<p>As SSA values, global variables define pointer values that are in
790scope (i.e. they dominate) all basic blocks in the program. Global
791variables always define a pointer to their "content" type because they
792describe a region of memory, and all memory objects in LLVM are
793accessed through pointers.</p>
794
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]795<p>A global variable may be declared to reside in a target-specifc numbered
796address space. For targets that support them, address spaces may affect how
797optimizations are performed and/or what target instructions are used to access
Christopher Lamb20a39e92007-12-12 08:44:39 +0000[diff] [blame]798the variable. The default address space is zero. The address space qualifier
799must precede any other attributes.</p>
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]800
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]801<p>LLVM allows an explicit section to be specified for globals. If the target
802supports it, it will emit globals to the section specified.</p>
803
804<p>An explicit alignment may be specified for a global. If not present, or if
805the alignment is set to zero, the alignment of the global is set by the target
806to whatever it feels convenient. If an explicit alignment is specified, the
807global is forced to have at least that much alignment. All alignments must be
808a power of 2.</p>
809
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]810<p>For example, the following defines a global in a numbered address space with
811an initializer, section, and alignment:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]812
813<div class="doc_code">
814<pre>
Dan Gohman21ef02c2009-01-11 00:40:00 +0000[diff] [blame]815@G = addrspace(5) constant float 1.0, section "foo", align 4
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]816</pre>
817</div>
818
819</div>
820
821
822<!-- ======================================================================= -->
823<div class="doc_subsection">
824 <a name="functionstructure">Functions</a>
825</div>
826
827<div class="doc_text">
828
829<p>LLVM function definitions consist of the "<tt>define</tt>" keyord,
830an optional <a href="#linkage">linkage type</a>, an optional
831<a href="#visibility">visibility style</a>, an optional
832<a href="#callingconv">calling convention</a>, a return type, an optional
833<a href="#paramattrs">parameter attribute</a> for the return type, a function
834name, a (possibly empty) argument list (each with optional
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]835<a href="#paramattrs">parameter attributes</a>), optional
836<a href="#fnattrs">function attributes</a>, an optional section,
837an optional alignment, an optional <a href="#gc">garbage collector name</a>,
Chris Lattneree202542008-10-04 18:10:21 +0000[diff] [blame]838an opening curly brace, a list of basic blocks, and a closing curly brace.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]839
840LLVM function declarations consist of the "<tt>declare</tt>" keyword, an
841optional <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
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]845name, a possibly empty list of arguments, an optional alignment, and an optional
Gordon Henriksend606f5b2007-12-10 03:30:21 +0000[diff] [blame]846<a href="#gc">garbage collector name</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]847
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]848<p>A function definition contains a list of basic blocks, forming the CFG
849(Control Flow Graph) for
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]850the function. Each basic block may optionally start with a label (giving the
851basic block a symbol table entry), contains a list of instructions, and ends
852with a <a href="#terminators">terminator</a> instruction (such as a branch or
853function return).</p>
854
855<p>The first basic block in a function is special in two ways: it is immediately
856executed on entrance to the function, and it is not allowed to have predecessor
857basic blocks (i.e. there can not be any branches to the entry block of a
858function). Because the block can have no predecessors, it also cannot have any
859<a href="#i_phi">PHI nodes</a>.</p>
860
861<p>LLVM allows an explicit section to be specified for functions. If the target
862supports it, it will emit functions to the section specified.</p>
863
864<p>An explicit alignment may be specified for a function. If not present, or if
865the alignment is set to zero, the alignment of the function is set by the target
866to whatever it feels convenient. If an explicit alignment is specified, the
867function is forced to have at least that much alignment. All alignments must be
868a power of 2.</p>
869
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]870 <h5>Syntax:</h5>
871
872<div class="doc_code">
Chris Lattner1e5c5cd02008-10-13 16:55:18 +0000[diff] [blame]873<tt>
874define [<a href="#linkage">linkage</a>] [<a href="#visibility">visibility</a>]
875 [<a href="#callingconv">cconv</a>] [<a href="#paramattrs">ret attrs</a>]
876 &lt;ResultType&gt; @&lt;FunctionName&gt; ([argument list])
877 [<a href="#fnattrs">fn Attrs</a>] [section "name"] [align N]
878 [<a href="#gc">gc</a>] { ... }
879</tt>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]880</div>
881
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]882</div>
883
884
885<!-- ======================================================================= -->
886<div class="doc_subsection">
887 <a name="aliasstructure">Aliases</a>
888</div>
889<div class="doc_text">
890 <p>Aliases act as "second name" for the aliasee value (which can be either
Anton Korobeynikov96822812008-03-22 08:36:14 +0000[diff] [blame]891 function, global variable, another alias or bitcast of global value). Aliases
892 may have an optional <a href="#linkage">linkage type</a>, and an
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]893 optional <a href="#visibility">visibility style</a>.</p>
894
895 <h5>Syntax:</h5>
896
897<div class="doc_code">
898<pre>
Duncan Sandsd7bfabf2008-09-12 20:48:21 +0000[diff] [blame]899@&lt;Name&gt; = alias [Linkage] [Visibility] &lt;AliaseeTy&gt; @&lt;Aliasee&gt;
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]900</pre>
901</div>
902
903</div>
904
905
906
907<!-- ======================================================================= -->
908<div class="doc_subsection"><a name="paramattrs">Parameter Attributes</a></div>
909<div class="doc_text">
910 <p>The return type and each parameter of a function type may have a set of
911 <i>parameter attributes</i> associated with them. Parameter attributes are
912 used to communicate additional information about the result or parameters of
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]913 a function. Parameter attributes are considered to be part of the function,
914 not of the function type, so functions with different parameter attributes
915 can have the same function type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]916
917 <p>Parameter attributes are simple keywords that follow the type specified. If
918 multiple parameter attributes are needed, they are space separated. For
919 example:</p>
920
921<div class="doc_code">
922<pre>
Nick Lewycky3022a742009-02-15 23:06:14 +0000[diff] [blame]923declare i32 @printf(i8* noalias nocapture, ...)
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]924declare i32 @atoi(i8 zeroext)
925declare signext i8 @returns_signed_char()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]926</pre>
927</div>
928
Duncan Sandsf5588dc2007-11-27 13:23:08 +0000[diff] [blame]929 <p>Note that any attributes for the function result (<tt>nounwind</tt>,
930 <tt>readonly</tt>) come immediately after the argument list.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]931
932 <p>Currently, only the following parameter attributes are defined:</p>
933 <dl>
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]934 <dt><tt>zeroext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]935 <dd>This indicates to the code generator that the parameter or return value
936 should be zero-extended to a 32-bit value by the caller (for a parameter)
937 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]938
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]939 <dt><tt>signext</tt></dt>
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]940 <dd>This indicates to the code generator that the parameter or return value
941 should be sign-extended to a 32-bit value by the caller (for a parameter)
942 or the callee (for a return value).</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]943
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]944 <dt><tt>inreg</tt></dt>
Dale Johannesenc08a0e22008-09-25 20:47:45 +0000[diff] [blame]945 <dd>This indicates that this parameter or return value should be treated
946 in a special target-dependent fashion during while emitting code for a
947 function call or return (usually, by putting it in a register as opposed
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]948 to memory, though some targets use it to distinguish between two different
949 kinds of registers). Use of this attribute is target-specific.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]950
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]951 <dt><tt><a name="byval">byval</a></tt></dt>
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]952 <dd>This indicates that the pointer parameter should really be passed by
953 value to the function. The attribute implies that a hidden copy of the
954 pointee is made between the caller and the callee, so the callee is unable
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]955 to modify the value in the callee. This attribute is only valid on LLVM
Chris Lattner04c86182008-01-15 04:34:22 +0000[diff] [blame]956 pointer arguments. It is generally used to pass structs and arrays by
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]957 value, but is also valid on pointers to scalars. The copy is considered to
958 belong to the caller not the callee (for example,
959 <tt><a href="#readonly">readonly</a></tt> functions should not write to
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]960 <tt>byval</tt> parameters). This is not a valid attribute for return
Chris Lattnere39c5bf2009-02-05 05:42:28 +0000[diff] [blame]961 values. The byval attribute also supports specifying an alignment with the
962 align attribute. This has a target-specific effect on the code generator
963 that usually indicates a desired alignment for the synthesized stack
964 slot.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]965
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]966 <dt><tt>sret</tt></dt>
Duncan Sands616cc032008-02-18 04:19:38 +0000[diff] [blame]967 <dd>This indicates that the pointer parameter specifies the address of a
968 structure that is the return value of the function in the source program.
Chris Lattnerf33b8452008-10-04 18:33:34 +0000[diff] [blame]969 This pointer must be guaranteed by the caller to be valid: loads and stores
970 to the structure may be assumed by the callee to not to trap. This may only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]971 be applied to the first parameter. This is not a valid attribute for
972 return values. </dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]973
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]974 <dt><tt>noalias</tt></dt>
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]975 <dd>This indicates that the pointer does not alias any global or any other
976 parameter. The caller is responsible for ensuring that this is the
Nick Lewyckya604a942008-11-24 05:00:44 +0000[diff] [blame]977 case. On a function return value, <tt>noalias</tt> additionally indicates
978 that the pointer does not alias any other pointers visible to the
Nick Lewycky40fb6f22008-12-19 06:39:12 +0000[diff] [blame]979 caller. For further details, please see the discussion of the NoAlias
980 response in
981 <a href="http://llvm.org/docs/AliasAnalysis.html#MustMayNo">alias
982 analysis</a>.</dd>
983
984 <dt><tt>nocapture</tt></dt>
985 <dd>This indicates that the callee does not make any copies of the pointer
986 that outlive the callee itself. This is not a valid attribute for return
987 values.</dd>
Chris Lattner275e6be2008-01-11 06:20:47 +0000[diff] [blame]988
Duncan Sands4ee46812007-07-27 19:57:41 +0000[diff] [blame]989 <dt><tt>nest</tt></dt>
Duncan Sandsf1a7d4c2008-07-08 09:27:25 +0000[diff] [blame]990 <dd>This indicates that the pointer parameter can be excised using the
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]991 <a href="#int_trampoline">trampoline intrinsics</a>. This is not a valid
992 attribute for return values.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]993 </dl>
994
995</div>
996
997<!-- ======================================================================= -->
998<div class="doc_subsection">
Gordon Henriksen13fe5e32007-12-10 03:18:06 +0000[diff] [blame]999 <a name="gc">Garbage Collector Names</a>
1000</div>
1001
1002<div class="doc_text">
1003<p>Each function may specify a garbage collector name, which is simply a
1004string.</p>
1005
1006<div class="doc_code"><pre
1007>define void @f() gc "name" { ...</pre></div>
1008
1009<p>The compiler declares the supported values of <i>name</i>. Specifying a
1010collector which will cause the compiler to alter its output in order to support
1011the named garbage collection algorithm.</p>
1012</div>
1013
1014<!-- ======================================================================= -->
1015<div class="doc_subsection">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1016 <a name="fnattrs">Function Attributes</a>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1017</div>
1018
1019<div class="doc_text">
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1020
1021<p>Function attributes are set to communicate additional information about
1022 a function. Function attributes are considered to be part of the function,
1023 not of the function type, so functions with different parameter attributes
1024 can have the same function type.</p>
1025
1026 <p>Function attributes are simple keywords that follow the type specified. If
1027 multiple attributes are needed, they are space separated. For
1028 example:</p>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1029
1030<div class="doc_code">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1031<pre>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1032define void @f() noinline { ... }
1033define void @f() alwaysinline { ... }
1034define void @f() alwaysinline optsize { ... }
1035define void @f() optsize
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1036</pre>
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1037</div>
1038
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1039<dl>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1040<dt><tt>alwaysinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1041<dd>This attribute indicates that the inliner should attempt to inline this
1042function into callers whenever possible, ignoring any active inlining size
1043threshold for this caller.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1044
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1045<dt><tt>noinline</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1046<dd>This attribute indicates that the inliner should never inline this function
Chris Lattner377f3ae2008-10-05 17:14:59 +0000[diff] [blame]1047in any situation. This attribute may not be used together with the
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1048<tt>alwaysinline</tt> attribute.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1049
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1050<dt><tt>optsize</tt></dt>
Devang Patelc29099b2008-09-29 18:34:44 +0000[diff] [blame]1051<dd>This attribute suggests that optimization passes and code generator passes
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1052make choices that keep the code size of this function low, and otherwise do
1053optimizations specifically to reduce code size.</dd>
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1054
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1055<dt><tt>noreturn</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1056<dd>This function attribute indicates that the function never returns normally.
1057This produces undefined behavior at runtime if the function ever does
1058dynamically return.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1059
1060<dt><tt>nounwind</tt></dt>
Chris Lattner82d70f92008-10-04 18:23:17 +0000[diff] [blame]1061<dd>This function attribute indicates that the function never returns with an
1062unwind or exceptional control flow. If the function does unwind, its runtime
1063behavior is undefined.</dd>
1064
1065<dt><tt>readnone</tt></dt>
Duncan Sands2f500832009-05-06 06:49:50 +0000[diff] [blame]1066<dd>This attribute indicates that the function computes its result (or decides to
1067unwind an exception) based strictly on its arguments, without dereferencing any
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1068pointer arguments or otherwise accessing any mutable state (e.g. memory, control
1069registers, etc) visible to caller functions. It does not write through any
1070pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments) and
Duncan Sands2f500832009-05-06 06:49:50 +0000[diff] [blame]1071never changes any state visible to callers. This means that it cannot unwind
1072exceptions by calling the <tt>C++</tt> exception throwing methods, but could
1073use the <tt>unwind</tt> instruction.</dd>
Devang Patel008cd3e2008-09-26 23:51:19 +0000[diff] [blame]1074
Duncan Sands7d94e5a2008-10-06 08:14:18 +0000[diff] [blame]1075<dt><tt><a name="readonly">readonly</a></tt></dt>
1076<dd>This attribute indicates that the function does not write through any
1077pointer arguments (including <tt><a href="#byval">byval</a></tt> arguments)
1078or otherwise modify any state (e.g. memory, control registers, etc) visible to
1079caller functions. It may dereference pointer arguments and read state that may
Duncan Sands2f500832009-05-06 06:49:50 +0000[diff] [blame]1080be set in the caller. A readonly function always returns the same value (or
1081unwinds an exception identically) when called with the same set of arguments
1082and global state. It cannot unwind an exception by calling the <tt>C++</tt>
1083exception throwing methods, but may use the <tt>unwind</tt> instruction.</dd>
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1084
1085<dt><tt><a name="ssp">ssp</a></tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1086<dd>This attribute indicates that the function should emit a stack smashing
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1087protector. It is in the form of a "canary"&mdash;a random value placed on the
1088stack before the local variables that's checked upon return from the function to
1089see if it has been overwritten. A heuristic is used to determine if a function
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1090needs stack protectors or not.
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1091
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1092<br><br>If a function that has an <tt>ssp</tt> attribute is inlined into a function
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1093that doesn't have an <tt>ssp</tt> attribute, then the resulting function will
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1094have an <tt>ssp</tt> attribute.</dd>
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1095
1096<dt><tt>sspreq</tt></dt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1097<dd>This attribute indicates that the function should <em>always</em> emit a
Bill Wendlingdfaabba2008-11-13 01:02:51 +0000[diff] [blame]1098stack smashing protector. This overrides the <tt><a href="#ssp">ssp</a></tt>
Bill Wendling34f70d82008-11-26 19:19:05 +0000[diff] [blame]1099function attribute.
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1100
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1101If a function that has an <tt>sspreq</tt> attribute is inlined into a
Bill Wendlingbe9ec3f2008-11-26 19:07:40 +0000[diff] [blame]1102function that doesn't have an <tt>sspreq</tt> attribute or which has
1103an <tt>ssp</tt> attribute, then the resulting function will have
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1104an <tt>sspreq</tt> attribute.</dd>
1105
1106<dt><tt>noredzone</tt></dt>
Dan Gohman06c9b732009-06-15 17:37:09 +0000[diff] [blame]1107<dd>This attribute indicates that the code generator should not use a
Dan Gohmanf958d5c2009-06-15 21:18:01 +0000[diff] [blame]1108red zone, even if the target-specific ABI normally permits it.
Dan Gohman06c9b732009-06-15 17:37:09 +0000[diff] [blame]1109</dd>
Devang Patela2f9f412009-06-12 19:45:19 +0000[diff] [blame]1110
1111<dt><tt>noimplicitfloat</tt></dt>
1112<dd>This attributes disables implicit floating point instructions.</dd>
1113
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1114</dl>
1115
Devang Pateld468f1c2008-09-04 23:05:13 +0000[diff] [blame]1116</div>
1117
1118<!-- ======================================================================= -->
1119<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1120 <a name="moduleasm">Module-Level Inline Assembly</a>
1121</div>
1122
1123<div class="doc_text">
1124<p>
1125Modules may contain "module-level inline asm" blocks, which corresponds to the
1126GCC "file scope inline asm" blocks. These blocks are internally concatenated by
1127LLVM and treated as a single unit, but may be separated in the .ll file if
1128desired. The syntax is very simple:
1129</p>
1130
1131<div class="doc_code">
1132<pre>
1133module asm "inline asm code goes here"
1134module asm "more can go here"
1135</pre>
1136</div>
1137
1138<p>The strings can contain any character by escaping non-printable characters.
1139 The escape sequence used is simply "\xx" where "xx" is the two digit hex code
1140 for the number.
1141</p>
1142
1143<p>
1144 The inline asm code is simply printed to the machine code .s file when
1145 assembly code is generated.
1146</p>
1147</div>
1148
1149<!-- ======================================================================= -->
1150<div class="doc_subsection">
1151 <a name="datalayout">Data Layout</a>
1152</div>
1153
1154<div class="doc_text">
1155<p>A module may specify a target specific data layout string that specifies how
1156data is to be laid out in memory. The syntax for the data layout is simply:</p>
1157<pre> target datalayout = "<i>layout specification</i>"</pre>
1158<p>The <i>layout specification</i> consists of a list of specifications
1159separated by the minus sign character ('-'). Each specification starts with a
1160letter and may include other information after the letter to define some
1161aspect of the data layout. The specifications accepted are as follows: </p>
1162<dl>
1163 <dt><tt>E</tt></dt>
1164 <dd>Specifies that the target lays out data in big-endian form. That is, the
1165 bits with the most significance have the lowest address location.</dd>
1166 <dt><tt>e</tt></dt>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1167 <dd>Specifies that the target lays out data in little-endian form. That is,
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1168 the bits with the least significance have the lowest address location.</dd>
1169 <dt><tt>p:<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1170 <dd>This specifies the <i>size</i> of a pointer and its <i>abi</i> and
1171 <i>preferred</i> alignments. All sizes are in bits. Specifying the <i>pref</i>
1172 alignment is optional. If omitted, the preceding <tt>:</tt> should be omitted
1173 too.</dd>
1174 <dt><tt>i<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1175 <dd>This specifies the alignment for an integer type of a given bit
1176 <i>size</i>. The value of <i>size</i> must be in the range [1,2^23).</dd>
1177 <dt><tt>v<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1178 <dd>This specifies the alignment for a vector type of a given bit
1179 <i>size</i>.</dd>
1180 <dt><tt>f<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1181 <dd>This specifies the alignment for a floating point type of a given bit
1182 <i>size</i>. The value of <i>size</i> must be either 32 (float) or 64
1183 (double).</dd>
1184 <dt><tt>a<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1185 <dd>This specifies the alignment for an aggregate type of a given bit
1186 <i>size</i>.</dd>
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1187 <dt><tt>s<i>size</i>:<i>abi</i>:<i>pref</i></tt></dt>
1188 <dd>This specifies the alignment for a stack object of a given bit
1189 <i>size</i>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1190</dl>
1191<p>When constructing the data layout for a given target, LLVM starts with a
1192default set of specifications which are then (possibly) overriden by the
1193specifications in the <tt>datalayout</tt> keyword. The default specifications
1194are given in this list:</p>
1195<ul>
1196 <li><tt>E</tt> - big endian</li>
1197 <li><tt>p:32:64:64</tt> - 32-bit pointers with 64-bit alignment</li>
1198 <li><tt>i1:8:8</tt> - i1 is 8-bit (byte) aligned</li>
1199 <li><tt>i8:8:8</tt> - i8 is 8-bit (byte) aligned</li>
1200 <li><tt>i16:16:16</tt> - i16 is 16-bit aligned</li>
1201 <li><tt>i32:32:32</tt> - i32 is 32-bit aligned</li>
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1202 <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]1203 alignment of 64-bits</li>
1204 <li><tt>f32:32:32</tt> - float is 32-bit aligned</li>
1205 <li><tt>f64:64:64</tt> - double is 64-bit aligned</li>
1206 <li><tt>v64:64:64</tt> - 64-bit vector is 64-bit aligned</li>
1207 <li><tt>v128:128:128</tt> - 128-bit vector is 128-bit aligned</li>
1208 <li><tt>a0:0:1</tt> - aggregates are 8-bit aligned</li>
Daniel Dunbard88a97b2009-06-08 22:17:53 +0000[diff] [blame]1209 <li><tt>s0:64:64</tt> - stack objects are 64-bit aligned</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1210</ul>
Chris Lattner6a9f3c42008-08-05 18:21:08 +0000[diff] [blame]1211<p>When LLVM is determining the alignment for a given type, it uses the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1212following rules:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1213<ol>
1214 <li>If the type sought is an exact match for one of the specifications, that
1215 specification is used.</li>
1216 <li>If no match is found, and the type sought is an integer type, then the
1217 smallest integer type that is larger than the bitwidth of the sought type is
1218 used. If none of the specifications are larger than the bitwidth then the the
1219 largest integer type is used. For example, given the default specifications
1220 above, the i7 type will use the alignment of i8 (next largest) while both
1221 i65 and i256 will use the alignment of i64 (largest specified).</li>
1222 <li>If no match is found, and the type sought is a vector type, then the
1223 largest vector type that is smaller than the sought vector type will be used
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]1224 as a fall back. This happens because &lt;128 x double&gt; can be implemented
1225 in terms of 64 &lt;2 x double&gt;, for example.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1226</ol>
1227</div>
1228
1229<!-- *********************************************************************** -->
1230<div class="doc_section"> <a name="typesystem">Type System</a> </div>
1231<!-- *********************************************************************** -->
1232
1233<div class="doc_text">
1234
1235<p>The LLVM type system is one of the most important features of the
1236intermediate representation. Being typed enables a number of
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]1237optimizations to be performed on the intermediate representation directly,
1238without having to do
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1239extra analyses on the side before the transformation. A strong type
1240system makes it easier to read the generated code and enables novel
1241analyses and transformations that are not feasible to perform on normal
1242three address code representations.</p>
1243
1244</div>
1245
1246<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1247<div class="doc_subsection"> <a name="t_classifications">Type
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1248Classifications</a> </div>
1249<div class="doc_text">
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1250<p>The types fall into a few useful
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1251classifications:</p>
1252
1253<table border="1" cellspacing="0" cellpadding="4">
1254 <tbody>
1255 <tr><th>Classification</th><th>Types</th></tr>
1256 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1257 <td><a href="#t_integer">integer</a></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1258 <td><tt>i1, i2, i3, ... i8, ... i16, ... i32, ... i64, ... </tt></td>
1259 </tr>
1260 <tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1261 <td><a href="#t_floating">floating point</a></td>
1262 <td><tt>float, double, x86_fp80, fp128, ppc_fp128</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1263 </tr>
1264 <tr>
1265 <td><a name="t_firstclass">first class</a></td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1266 <td><a href="#t_integer">integer</a>,
1267 <a href="#t_floating">floating point</a>,
1268 <a href="#t_pointer">pointer</a>,
Dan Gohmanf6237db2008-06-18 18:42:13 +0000[diff] [blame]1269 <a href="#t_vector">vector</a>,
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]1270 <a href="#t_struct">structure</a>,
1271 <a href="#t_array">array</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1272 <a href="#t_label">label</a>,
1273 <a href="#t_metadata">metadata</a>.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1274 </td>
1275 </tr>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1276 <tr>
1277 <td><a href="#t_primitive">primitive</a></td>
1278 <td><a href="#t_label">label</a>,
1279 <a href="#t_void">void</a>,
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1280 <a href="#t_floating">floating point</a>,
1281 <a href="#t_metadata">metadata</a>.</td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1282 </tr>
1283 <tr>
1284 <td><a href="#t_derived">derived</a></td>
1285 <td><a href="#t_integer">integer</a>,
1286 <a href="#t_array">array</a>,
1287 <a href="#t_function">function</a>,
1288 <a href="#t_pointer">pointer</a>,
1289 <a href="#t_struct">structure</a>,
1290 <a href="#t_pstruct">packed structure</a>,
1291 <a href="#t_vector">vector</a>,
1292 <a href="#t_opaque">opaque</a>.
Dan Gohman032ba852008-10-14 16:32:04 +0000[diff] [blame]1293 </td>
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1294 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1295 </tbody>
1296</table>
1297
1298<p>The <a href="#t_firstclass">first class</a> types are perhaps the
1299most important. Values of these types are the only ones which can be
1300produced by instructions, passed as arguments, or used as operands to
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]1301instructions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1302</div>
1303
1304<!-- ======================================================================= -->
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1305<div class="doc_subsection"> <a name="t_primitive">Primitive Types</a> </div>
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1306
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1307<div class="doc_text">
1308<p>The primitive types are the fundamental building blocks of the LLVM
1309system.</p>
1310
Chris Lattner86437612008-01-04 04:34:14 +0000[diff] [blame]1311</div>
1312
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1313<!-- _______________________________________________________________________ -->
1314<div class="doc_subsubsection"> <a name="t_floating">Floating Point Types</a> </div>
1315
1316<div class="doc_text">
1317 <table>
1318 <tbody>
1319 <tr><th>Type</th><th>Description</th></tr>
1320 <tr><td><tt>float</tt></td><td>32-bit floating point value</td></tr>
1321 <tr><td><tt>double</tt></td><td>64-bit floating point value</td></tr>
1322 <tr><td><tt>fp128</tt></td><td>128-bit floating point value (112-bit mantissa)</td></tr>
1323 <tr><td><tt>x86_fp80</tt></td><td>80-bit floating point value (X87)</td></tr>
1324 <tr><td><tt>ppc_fp128</tt></td><td>128-bit floating point value (two 64-bits)</td></tr>
1325 </tbody>
1326 </table>
1327</div>
1328
1329<!-- _______________________________________________________________________ -->
1330<div class="doc_subsubsection"> <a name="t_void">Void Type</a> </div>
1331
1332<div class="doc_text">
1333<h5>Overview:</h5>
1334<p>The void type does not represent any value and has no size.</p>
1335
1336<h5>Syntax:</h5>
1337
1338<pre>
1339 void
1340</pre>
1341</div>
1342
1343<!-- _______________________________________________________________________ -->
1344<div class="doc_subsubsection"> <a name="t_label">Label Type</a> </div>
1345
1346<div class="doc_text">
1347<h5>Overview:</h5>
1348<p>The label type represents code labels.</p>
1349
1350<h5>Syntax:</h5>
1351
1352<pre>
1353 label
1354</pre>
1355</div>
1356
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]1357<!-- _______________________________________________________________________ -->
1358<div class="doc_subsubsection"> <a name="t_metadata">Metadata Type</a> </div>
1359
1360<div class="doc_text">
1361<h5>Overview:</h5>
1362<p>The metadata type represents embedded metadata. The only derived type that
1363may contain metadata is <tt>metadata*</tt> or a function type that returns or
1364takes metadata typed parameters, but not pointer to metadata types.</p>
1365
1366<h5>Syntax:</h5>
1367
1368<pre>
1369 metadata
1370</pre>
1371</div>
1372
Chris Lattner488772f2008-01-04 04:32:38 +0000[diff] [blame]1373
1374<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1375<div class="doc_subsection"> <a name="t_derived">Derived Types</a> </div>
1376
1377<div class="doc_text">
1378
1379<p>The real power in LLVM comes from the derived types in the system.
1380This is what allows a programmer to represent arrays, functions,
1381pointers, and other useful types. Note that these derived types may be
1382recursive: For example, it is possible to have a two dimensional array.</p>
1383
1384</div>
1385
1386<!-- _______________________________________________________________________ -->
1387<div class="doc_subsubsection"> <a name="t_integer">Integer Type</a> </div>
1388
1389<div class="doc_text">
1390
1391<h5>Overview:</h5>
1392<p>The integer type is a very simple derived type that simply specifies an
1393arbitrary bit width for the integer type desired. Any bit width from 1 bit to
13942^23-1 (about 8 million) can be specified.</p>
1395
1396<h5>Syntax:</h5>
1397
1398<pre>
1399 iN
1400</pre>
1401
1402<p>The number of bits the integer will occupy is specified by the <tt>N</tt>
1403value.</p>
1404
1405<h5>Examples:</h5>
1406<table class="layout">
Nick Lewycky39382d62009-05-24 02:46:06 +0000[diff] [blame]1407 <tr class="layout">
1408 <td class="left"><tt>i1</tt></td>
1409 <td class="left">a single-bit integer.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1410 </tr>
Nick Lewycky39382d62009-05-24 02:46:06 +0000[diff] [blame]1411 <tr class="layout">
1412 <td class="left"><tt>i32</tt></td>
1413 <td class="left">a 32-bit integer.</td>
1414 </tr>
1415 <tr class="layout">
1416 <td class="left"><tt>i1942652</tt></td>
1417 <td class="left">a really big integer of over 1 million bits.</td>
1418 </tr>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1419</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1420
1421<p>Note that the code generator does not yet support large integer types
1422to be used as function return types. The specific limit on how large a
1423return type the code generator can currently handle is target-dependent;
1424currently it's often 64 bits for 32-bit targets and 128 bits for 64-bit
1425targets.</p>
1426
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1427</div>
1428
1429<!-- _______________________________________________________________________ -->
1430<div class="doc_subsubsection"> <a name="t_array">Array Type</a> </div>
1431
1432<div class="doc_text">
1433
1434<h5>Overview:</h5>
1435
1436<p>The array type is a very simple derived type that arranges elements
1437sequentially in memory. The array type requires a size (number of
1438elements) and an underlying data type.</p>
1439
1440<h5>Syntax:</h5>
1441
1442<pre>
1443 [&lt;# elements&gt; x &lt;elementtype&gt;]
1444</pre>
1445
1446<p>The number of elements is a constant integer value; elementtype may
1447be any type with a size.</p>
1448
1449<h5>Examples:</h5>
1450<table class="layout">
1451 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1452 <td class="left"><tt>[40 x i32]</tt></td>
1453 <td class="left">Array of 40 32-bit integer values.</td>
1454 </tr>
1455 <tr class="layout">
1456 <td class="left"><tt>[41 x i32]</tt></td>
1457 <td class="left">Array of 41 32-bit integer values.</td>
1458 </tr>
1459 <tr class="layout">
1460 <td class="left"><tt>[4 x i8]</tt></td>
1461 <td class="left">Array of 4 8-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1462 </tr>
1463</table>
1464<p>Here are some examples of multidimensional arrays:</p>
1465<table class="layout">
1466 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1467 <td class="left"><tt>[3 x [4 x i32]]</tt></td>
1468 <td class="left">3x4 array of 32-bit integer values.</td>
1469 </tr>
1470 <tr class="layout">
1471 <td class="left"><tt>[12 x [10 x float]]</tt></td>
1472 <td class="left">12x10 array of single precision floating point values.</td>
1473 </tr>
1474 <tr class="layout">
1475 <td class="left"><tt>[2 x [3 x [4 x i16]]]</tt></td>
1476 <td class="left">2x3x4 array of 16-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1477 </tr>
1478</table>
1479
1480<p>Note that 'variable sized arrays' can be implemented in LLVM with a zero
1481length array. Normally, accesses past the end of an array are undefined in
1482LLVM (e.g. it is illegal to access the 5th element of a 3 element array).
1483As a special case, however, zero length arrays are recognized to be variable
1484length. This allows implementation of 'pascal style arrays' with the LLVM
1485type "{ i32, [0 x float]}", for example.</p>
1486
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1487<p>Note that the code generator does not yet support large aggregate types
1488to be used as function return types. The specific limit on how large an
1489aggregate return type the code generator can currently handle is
1490target-dependent, and also dependent on the aggregate element types.</p>
1491
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1492</div>
1493
1494<!-- _______________________________________________________________________ -->
1495<div class="doc_subsubsection"> <a name="t_function">Function Type</a> </div>
1496<div class="doc_text">
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1497
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1498<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1499
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1500<p>The function type can be thought of as a function signature. It
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1501consists of a return type and a list of formal parameter types. The
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1502return type of a function type is a scalar type, a void type, or a struct type.
Devang Pateld5404c02008-03-24 20:52:42 +0000[diff] [blame]1503If the return type is a struct type then all struct elements must be of first
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1504class types, and the struct must have at least one element.</p>
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1505
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1506<h5>Syntax:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1507
1508<pre>
1509 &lt;returntype list&gt; (&lt;parameter list&gt;)
1510</pre>
1511
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1512<p>...where '<tt>&lt;parameter list&gt;</tt>' is a comma-separated list of type
1513specifiers. Optionally, the parameter list may include a type <tt>...</tt>,
1514which indicates that the function takes a variable number of arguments.
1515Variable argument functions can access their arguments with the <a
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]1516 href="#int_varargs">variable argument handling intrinsic</a> functions.
1517'<tt>&lt;returntype list&gt;</tt>' is a comma-separated list of
1518<a href="#t_firstclass">first class</a> type specifiers.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]1519
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1520<h5>Examples:</h5>
1521<table class="layout">
1522 <tr class="layout">
1523 <td class="left"><tt>i32 (i32)</tt></td>
1524 <td class="left">function taking an <tt>i32</tt>, returning an <tt>i32</tt>
1525 </td>
1526 </tr><tr class="layout">
Reid Spencerf234bed2007-07-19 23:13:04 +0000[diff] [blame]1527 <td class="left"><tt>float&nbsp;(i16&nbsp;signext,&nbsp;i32&nbsp;*)&nbsp;*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1528 </tt></td>
1529 <td class="left"><a href="#t_pointer">Pointer</a> to a function that takes
1530 an <tt>i16</tt> that should be sign extended and a
1531 <a href="#t_pointer">pointer</a> to <tt>i32</tt>, returning
1532 <tt>float</tt>.
1533 </td>
1534 </tr><tr class="layout">
1535 <td class="left"><tt>i32 (i8*, ...)</tt></td>
1536 <td class="left">A vararg function that takes at least one
1537 <a href="#t_pointer">pointer</a> to <tt>i8 </tt> (char in C),
1538 which returns an integer. This is the signature for <tt>printf</tt> in
1539 LLVM.
1540 </td>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1541 </tr><tr class="layout">
1542 <td class="left"><tt>{i32, i32} (i32)</tt></td>
Misha Brukmanafc88b02008-11-27 06:41:20 +0000[diff] [blame]1543 <td class="left">A function taking an <tt>i32</tt>, returning two
1544 <tt>i32</tt> values as an aggregate of type <tt>{ i32, i32 }</tt>
Devang Pateld4ba41d2008-03-24 05:35:41 +0000[diff] [blame]1545 </td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1546 </tr>
1547</table>
1548
1549</div>
1550<!-- _______________________________________________________________________ -->
1551<div class="doc_subsubsection"> <a name="t_struct">Structure Type</a> </div>
1552<div class="doc_text">
1553<h5>Overview:</h5>
1554<p>The structure type is used to represent a collection of data members
1555together in memory. The packing of the field types is defined to match
1556the ABI of the underlying processor. The elements of a structure may
1557be any type that has a size.</p>
1558<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1559and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1560field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1561instruction.</p>
1562<h5>Syntax:</h5>
1563<pre> { &lt;type list&gt; }<br></pre>
1564<h5>Examples:</h5>
1565<table class="layout">
1566 <tr class="layout">
1567 <td class="left"><tt>{ i32, i32, i32 }</tt></td>
1568 <td class="left">A triple of three <tt>i32</tt> values</td>
1569 </tr><tr class="layout">
1570 <td class="left"><tt>{&nbsp;float,&nbsp;i32&nbsp;(i32)&nbsp;*&nbsp;}</tt></td>
1571 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1572 second element is a <a href="#t_pointer">pointer</a> to a
1573 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1574 an <tt>i32</tt>.</td>
1575 </tr>
1576</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1577
1578<p>Note that the code generator does not yet support large aggregate types
1579to be used as function return types. The specific limit on how large an
1580aggregate return type the code generator can currently handle is
1581target-dependent, and also dependent on the aggregate element types.</p>
1582
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1583</div>
1584
1585<!-- _______________________________________________________________________ -->
1586<div class="doc_subsubsection"> <a name="t_pstruct">Packed Structure Type</a>
1587</div>
1588<div class="doc_text">
1589<h5>Overview:</h5>
1590<p>The packed structure type is used to represent a collection of data members
1591together in memory. There is no padding between fields. Further, the alignment
1592of a packed structure is 1 byte. The elements of a packed structure may
1593be any type that has a size.</p>
1594<p>Structures are accessed using '<tt><a href="#i_load">load</a></tt>
1595and '<tt><a href="#i_store">store</a></tt>' by getting a pointer to a
1596field with the '<tt><a href="#i_getelementptr">getelementptr</a></tt>'
1597instruction.</p>
1598<h5>Syntax:</h5>
1599<pre> &lt; { &lt;type list&gt; } &gt; <br></pre>
1600<h5>Examples:</h5>
1601<table class="layout">
1602 <tr class="layout">
1603 <td class="left"><tt>&lt; { i32, i32, i32 } &gt;</tt></td>
1604 <td class="left">A triple of three <tt>i32</tt> values</td>
1605 </tr><tr class="layout">
Bill Wendling74d3eac2008-09-07 10:26:33 +0000[diff] [blame]1606 <td class="left">
1607<tt>&lt;&nbsp;{&nbsp;float,&nbsp;i32&nbsp;(i32)*&nbsp;}&nbsp;&gt;</tt></td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1608 <td class="left">A pair, where the first element is a <tt>float</tt> and the
1609 second element is a <a href="#t_pointer">pointer</a> to a
1610 <a href="#t_function">function</a> that takes an <tt>i32</tt>, returning
1611 an <tt>i32</tt>.</td>
1612 </tr>
1613</table>
1614</div>
1615
1616<!-- _______________________________________________________________________ -->
1617<div class="doc_subsubsection"> <a name="t_pointer">Pointer Type</a> </div>
1618<div class="doc_text">
1619<h5>Overview:</h5>
1620<p>As in many languages, the pointer type represents a pointer or
Christopher Lambdd0049d2007-12-11 09:31:00 +0000[diff] [blame]1621reference to another object, which must live in memory. Pointer types may have
1622an optional address space attribute defining the target-specific numbered
1623address space where the pointed-to object resides. The default address space is
1624zero.</p>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1625
1626<p>Note that LLVM does not permit pointers to void (<tt>void*</tt>) nor does
Chris Lattner0d879e02009-02-08 22:21:28 +0000[diff] [blame]1627it permit pointers to labels (<tt>label*</tt>). Use <tt>i8*</tt> instead.</p>
Chris Lattner96edbd32009-02-08 19:53:29 +0000[diff] [blame]1628
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1629<h5>Syntax:</h5>
1630<pre> &lt;type&gt; *<br></pre>
1631<h5>Examples:</h5>
1632<table class="layout">
1633 <tr class="layout">
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]1634 <td class="left"><tt>[4 x i32]*</tt></td>
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1635 <td class="left">A <a href="#t_pointer">pointer</a> to <a
1636 href="#t_array">array</a> of four <tt>i32</tt> values.</td>
1637 </tr>
1638 <tr class="layout">
1639 <td class="left"><tt>i32 (i32 *) *</tt></td>
1640 <td class="left"> A <a href="#t_pointer">pointer</a> to a <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1641 href="#t_function">function</a> that takes an <tt>i32*</tt>, returning an
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1642 <tt>i32</tt>.</td>
1643 </tr>
1644 <tr class="layout">
1645 <td class="left"><tt>i32 addrspace(5)*</tt></td>
1646 <td class="left">A <a href="#t_pointer">pointer</a> to an <tt>i32</tt> value
1647 that resides in address space #5.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1648 </tr>
1649</table>
1650</div>
1651
1652<!-- _______________________________________________________________________ -->
1653<div class="doc_subsubsection"> <a name="t_vector">Vector Type</a> </div>
1654<div class="doc_text">
1655
1656<h5>Overview:</h5>
1657
1658<p>A vector type is a simple derived type that represents a vector
1659of elements. Vector types are used when multiple primitive data
1660are operated in parallel using a single instruction (SIMD).
1661A vector type requires a size (number of
1662elements) and an underlying primitive data type. Vectors must have a power
1663of two length (1, 2, 4, 8, 16 ...). Vector types are
1664considered <a href="#t_firstclass">first class</a>.</p>
1665
1666<h5>Syntax:</h5>
1667
1668<pre>
1669 &lt; &lt;# elements&gt; x &lt;elementtype&gt; &gt;
1670</pre>
1671
1672<p>The number of elements is a constant integer value; elementtype may
1673be any integer or floating point type.</p>
1674
1675<h5>Examples:</h5>
1676
1677<table class="layout">
1678 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1679 <td class="left"><tt>&lt;4 x i32&gt;</tt></td>
1680 <td class="left">Vector of 4 32-bit integer values.</td>
1681 </tr>
1682 <tr class="layout">
1683 <td class="left"><tt>&lt;8 x float&gt;</tt></td>
1684 <td class="left">Vector of 8 32-bit floating-point values.</td>
1685 </tr>
1686 <tr class="layout">
1687 <td class="left"><tt>&lt;2 x i64&gt;</tt></td>
1688 <td class="left">Vector of 2 64-bit integer values.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1689 </tr>
1690</table>
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]1691
1692<p>Note that the code generator does not yet support large vector types
1693to be used as function return types. The specific limit on how large a
1694vector return type codegen can currently handle is target-dependent;
1695currently it's often a few times longer than a hardware vector register.</p>
1696
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1697</div>
1698
1699<!-- _______________________________________________________________________ -->
1700<div class="doc_subsubsection"> <a name="t_opaque">Opaque Type</a> </div>
1701<div class="doc_text">
1702
1703<h5>Overview:</h5>
1704
1705<p>Opaque types are used to represent unknown types in the system. This
Gordon Henriksenda0706e2007-10-14 00:34:53 +0000[diff] [blame]1706corresponds (for example) to the C notion of a forward declared structure type.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1707In LLVM, opaque types can eventually be resolved to any type (not just a
1708structure type).</p>
1709
1710<h5>Syntax:</h5>
1711
1712<pre>
1713 opaque
1714</pre>
1715
1716<h5>Examples:</h5>
1717
1718<table class="layout">
1719 <tr class="layout">
Chris Lattner7311d222007-12-19 05:04:11 +0000[diff] [blame]1720 <td class="left"><tt>opaque</tt></td>
1721 <td class="left">An opaque type.</td>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1722 </tr>
1723</table>
1724</div>
1725
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1726<!-- ======================================================================= -->
1727<div class="doc_subsection">
1728 <a name="t_uprefs">Type Up-references</a>
1729</div>
1730
1731<div class="doc_text">
1732<h5>Overview:</h5>
1733<p>
1734An "up reference" allows you to refer to a lexically enclosing type without
1735requiring it to have a name. For instance, a structure declaration may contain a
1736pointer to any of the types it is lexically a member of. Example of up
1737references (with their equivalent as named type declarations) include:</p>
1738
1739<pre>
Chris Lattner5ad632d2009-02-09 10:00:56 +0000[diff] [blame]1740 { \2 * } %x = type { %x* }
Chris Lattner515195a2009-02-02 07:32:36 +0000[diff] [blame]1741 { \2 }* %y = type { %y }*
1742 \1* %z = type %z*
1743</pre>
1744
1745<p>
1746An up reference is needed by the asmprinter for printing out cyclic types when
1747there is no declared name for a type in the cycle. Because the asmprinter does
1748not want to print out an infinite type string, it needs a syntax to handle
1749recursive types that have no names (all names are optional in llvm IR).
1750</p>
1751
1752<h5>Syntax:</h5>
1753<pre>
1754 \&lt;level&gt;
1755</pre>
1756
1757<p>
1758The level is the count of the lexical type that is being referred to.
1759</p>
1760
1761<h5>Examples:</h5>
1762
1763<table class="layout">
1764 <tr class="layout">
1765 <td class="left"><tt>\1*</tt></td>
1766 <td class="left">Self-referential pointer.</td>
1767 </tr>
1768 <tr class="layout">
1769 <td class="left"><tt>{ { \3*, i8 }, i32 }</tt></td>
1770 <td class="left">Recursive structure where the upref refers to the out-most
1771 structure.</td>
1772 </tr>
1773</table>
1774</div>
1775
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1776
1777<!-- *********************************************************************** -->
1778<div class="doc_section"> <a name="constants">Constants</a> </div>
1779<!-- *********************************************************************** -->
1780
1781<div class="doc_text">
1782
1783<p>LLVM has several different basic types of constants. This section describes
1784them all and their syntax.</p>
1785
1786</div>
1787
1788<!-- ======================================================================= -->
1789<div class="doc_subsection"><a name="simpleconstants">Simple Constants</a></div>
1790
1791<div class="doc_text">
1792
1793<dl>
1794 <dt><b>Boolean constants</b></dt>
1795
1796 <dd>The two strings '<tt>true</tt>' and '<tt>false</tt>' are both valid
1797 constants of the <tt><a href="#t_primitive">i1</a></tt> type.
1798 </dd>
1799
1800 <dt><b>Integer constants</b></dt>
1801
1802 <dd>Standard integers (such as '4') are constants of the <a
1803 href="#t_integer">integer</a> type. Negative numbers may be used with
1804 integer types.
1805 </dd>
1806
1807 <dt><b>Floating point constants</b></dt>
1808
1809 <dd>Floating point constants use standard decimal notation (e.g. 123.421),
1810 exponential notation (e.g. 1.23421e+2), or a more precise hexadecimal
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]1811 notation (see below). The assembler requires the exact decimal value of
1812 a floating-point constant. For example, the assembler accepts 1.25 but
1813 rejects 1.3 because 1.3 is a repeating decimal in binary. Floating point
1814 constants must have a <a href="#t_floating">floating point</a> type. </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1815
1816 <dt><b>Null pointer constants</b></dt>
1817
1818 <dd>The identifier '<tt>null</tt>' is recognized as a null pointer constant
1819 and must be of <a href="#t_pointer">pointer type</a>.</dd>
1820
1821</dl>
1822
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1823<p>The one non-intuitive notation for constants is the hexadecimal form
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1824of floating point constants. For example, the form '<tt>double
18250x432ff973cafa8000</tt>' is equivalent to (but harder to read than) '<tt>double
18264.5e+15</tt>'. The only time hexadecimal floating point constants are required
1827(and the only time that they are generated by the disassembler) is when a
1828floating point constant must be emitted but it cannot be represented as a
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1829decimal floating point number in a reasonable number of digits. For example,
1830NaN's, infinities, and other
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1831special values are represented in their IEEE hexadecimal format so that
1832assembly and disassembly do not cause any bits to change in the constants.</p>
Dale Johannesenf82a52f2009-02-11 22:14:51 +0000[diff] [blame]1833<p>When using the hexadecimal form, constants of types float and double are
1834represented using the 16-digit form shown above (which matches the IEEE754
1835representation for double); float values must, however, be exactly representable
1836as IEE754 single precision.
1837Hexadecimal format is always used for long
1838double, and there are three forms of long double. The 80-bit
1839format used by x86 is represented as <tt>0xK</tt>
1840followed by 20 hexadecimal digits.
1841The 128-bit format used by PowerPC (two adjacent doubles) is represented
1842by <tt>0xM</tt> followed by 32 hexadecimal digits. The IEEE 128-bit
1843format is represented
1844by <tt>0xL</tt> followed by 32 hexadecimal digits; no currently supported
1845target uses this format. Long doubles will only work if they match
1846the long double format on your target. All hexadecimal formats are big-endian
1847(sign bit at the left).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1848</div>
1849
1850<!-- ======================================================================= -->
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1851<div class="doc_subsection">
1852<a name="aggregateconstants"> <!-- old anchor -->
1853<a name="complexconstants">Complex Constants</a></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1854</div>
1855
1856<div class="doc_text">
Chris Lattner97063852009-02-28 18:32:25 +0000[diff] [blame]1857<p>Complex constants are a (potentially recursive) combination of simple
1858constants and smaller complex constants.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1859
1860<dl>
1861 <dt><b>Structure constants</b></dt>
1862
1863 <dd>Structure constants are represented with notation similar to structure
1864 type definitions (a comma separated list of elements, surrounded by braces
Chris Lattner6c8de962007-12-25 20:34:52 +0000[diff] [blame]1865 (<tt>{}</tt>)). For example: "<tt>{ i32 4, float 17.0, i32* @G }</tt>",
1866 where "<tt>@G</tt>" is declared as "<tt>@G = external global i32</tt>". Structure constants
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1867 must have <a href="#t_struct">structure type</a>, and the number and
1868 types of elements must match those specified by the type.
1869 </dd>
1870
1871 <dt><b>Array constants</b></dt>
1872
1873 <dd>Array constants are represented with notation similar to array type
1874 definitions (a comma separated list of elements, surrounded by square brackets
1875 (<tt>[]</tt>)). For example: "<tt>[ i32 42, i32 11, i32 74 ]</tt>". Array
1876 constants must have <a href="#t_array">array type</a>, and the number and
1877 types of elements must match those specified by the type.
1878 </dd>
1879
1880 <dt><b>Vector constants</b></dt>
1881
1882 <dd>Vector constants are represented with notation similar to vector type
1883 definitions (a comma separated list of elements, surrounded by
1884 less-than/greater-than's (<tt>&lt;&gt;</tt>)). For example: "<tt>&lt; i32 42,
1885 i32 11, i32 74, i32 100 &gt;</tt>". Vector constants must have <a
1886 href="#t_vector">vector type</a>, and the number and types of elements must
1887 match those specified by the type.
1888 </dd>
1889
1890 <dt><b>Zero initialization</b></dt>
1891
1892 <dd>The string '<tt>zeroinitializer</tt>' can be used to zero initialize a
1893 value to zero of <em>any</em> type, including scalar and aggregate types.
1894 This is often used to avoid having to print large zero initializers (e.g. for
1895 large arrays) and is always exactly equivalent to using explicit zero
1896 initializers.
1897 </dd>
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]1898
1899 <dt><b>Metadata node</b></dt>
1900
Nick Lewyckyf122c7e2009-05-30 16:08:30 +0000[diff] [blame]1901 <dd>A metadata node is a structure-like constant with
1902 <a href="#t_metadata">metadata type</a>. For example:
1903 "<tt>metadata !{ i32 0, metadata !"test" }</tt>". Unlike other constants
1904 that are meant to be interpreted as part of the instruction stream, metadata
1905 is a place to attach additional information such as debug info.
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]1906 </dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1907</dl>
1908
1909</div>
1910
1911<!-- ======================================================================= -->
1912<div class="doc_subsection">
1913 <a name="globalconstants">Global Variable and Function Addresses</a>
1914</div>
1915
1916<div class="doc_text">
1917
1918<p>The addresses of <a href="#globalvars">global variables</a> and <a
1919href="#functionstructure">functions</a> are always implicitly valid (link-time)
1920constants. These constants are explicitly referenced when the <a
1921href="#identifiers">identifier for the global</a> is used and always have <a
1922href="#t_pointer">pointer</a> type. For example, the following is a legal LLVM
1923file:</p>
1924
1925<div class="doc_code">
1926<pre>
1927@X = global i32 17
1928@Y = global i32 42
1929@Z = global [2 x i32*] [ i32* @X, i32* @Y ]
1930</pre>
1931</div>
1932
1933</div>
1934
1935<!-- ======================================================================= -->
1936<div class="doc_subsection"><a name="undefvalues">Undefined Values</a></div>
1937<div class="doc_text">
1938 <p>The string '<tt>undef</tt>' is recognized as a type-less constant that has
1939 no specific value. Undefined values may be of any type and be used anywhere
1940 a constant is permitted.</p>
1941
1942 <p>Undefined values indicate to the compiler that the program is well defined
1943 no matter what value is used, giving the compiler more freedom to optimize.
1944 </p>
1945</div>
1946
1947<!-- ======================================================================= -->
1948<div class="doc_subsection"><a name="constantexprs">Constant Expressions</a>
1949</div>
1950
1951<div class="doc_text">
1952
1953<p>Constant expressions are used to allow expressions involving other constants
1954to be used as constants. Constant expressions may be of any <a
1955href="#t_firstclass">first class</a> type and may involve any LLVM operation
1956that does not have side effects (e.g. load and call are not supported). The
1957following is the syntax for constant expressions:</p>
1958
1959<dl>
1960 <dt><b><tt>trunc ( CST to TYPE )</tt></b></dt>
1961 <dd>Truncate a constant to another type. The bit size of CST must be larger
1962 than the bit size of TYPE. Both types must be integers.</dd>
1963
1964 <dt><b><tt>zext ( CST to TYPE )</tt></b></dt>
1965 <dd>Zero extend a constant to another type. The bit size of CST must be
1966 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1967
1968 <dt><b><tt>sext ( CST to TYPE )</tt></b></dt>
1969 <dd>Sign extend a constant to another type. The bit size of CST must be
1970 smaller or equal to the bit size of TYPE. Both types must be integers.</dd>
1971
1972 <dt><b><tt>fptrunc ( CST to TYPE )</tt></b></dt>
1973 <dd>Truncate a floating point constant to another floating point type. The
1974 size of CST must be larger than the size of TYPE. Both types must be
1975 floating point.</dd>
1976
1977 <dt><b><tt>fpext ( CST to TYPE )</tt></b></dt>
1978 <dd>Floating point extend a constant to another type. The size of CST must be
1979 smaller or equal to the size of TYPE. Both types must be floating point.</dd>
1980
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]1981 <dt><b><tt>fptoui ( CST to TYPE )</tt></b></dt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1982 <dd>Convert a floating point constant to the corresponding unsigned integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1983 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1984 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1985 of the same number of elements. If the value won't fit in the integer type,
1986 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1987
1988 <dt><b><tt>fptosi ( CST to TYPE )</tt></b></dt>
1989 <dd>Convert a floating point constant to the corresponding signed integer
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1990 constant. TYPE must be a scalar or vector integer type. CST must be of scalar
1991 or vector floating point type. Both CST and TYPE must be scalars, or vectors
1992 of the same number of elements. If the value won't fit in the integer type,
1993 the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1994
1995 <dt><b><tt>uitofp ( CST to TYPE )</tt></b></dt>
1996 <dd>Convert an unsigned integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]1997 constant. TYPE must be a scalar or vector floating point type. CST must be of
1998 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
1999 of the same number of elements. If the value won't fit in the floating point
2000 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2001
2002 <dt><b><tt>sitofp ( CST to TYPE )</tt></b></dt>
2003 <dd>Convert a signed integer constant to the corresponding floating point
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]2004 constant. TYPE must be a scalar or vector floating point type. CST must be of
2005 scalar or vector integer type. Both CST and TYPE must be scalars, or vectors
2006 of the same number of elements. If the value won't fit in the floating point
2007 type, the results are undefined.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2008
2009 <dt><b><tt>ptrtoint ( CST to TYPE )</tt></b></dt>
2010 <dd>Convert a pointer typed constant to the corresponding integer constant
2011 TYPE must be an integer type. CST must be of pointer type. The CST value is
2012 zero extended, truncated, or unchanged to make it fit in TYPE.</dd>
2013
2014 <dt><b><tt>inttoptr ( CST to TYPE )</tt></b></dt>
2015 <dd>Convert a integer constant to a pointer constant. TYPE must be a
2016 pointer type. CST must be of integer type. The CST value is zero extended,
2017 truncated, or unchanged to make it fit in a pointer size. This one is
2018 <i>really</i> dangerous!</dd>
2019
2020 <dt><b><tt>bitcast ( CST to TYPE )</tt></b></dt>
Chris Lattner557bc5d2009-02-28 18:27:03 +0000[diff] [blame]2021 <dd>Convert a constant, CST, to another TYPE. The constraints of the operands
2022 are the same as those for the <a href="#i_bitcast">bitcast
2023 instruction</a>.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2024
2025 <dt><b><tt>getelementptr ( CSTPTR, IDX0, IDX1, ... )</tt></b></dt>
2026
2027 <dd>Perform the <a href="#i_getelementptr">getelementptr operation</a> on
2028 constants. As with the <a href="#i_getelementptr">getelementptr</a>
2029 instruction, the index list may have zero or more indexes, which are required
2030 to make sense for the type of "CSTPTR".</dd>
2031
2032 <dt><b><tt>select ( COND, VAL1, VAL2 )</tt></b></dt>
2033
2034 <dd>Perform the <a href="#i_select">select operation</a> on
2035 constants.</dd>
2036
2037 <dt><b><tt>icmp COND ( VAL1, VAL2 )</tt></b></dt>
2038 <dd>Performs the <a href="#i_icmp">icmp operation</a> on constants.</dd>
2039
2040 <dt><b><tt>fcmp COND ( VAL1, VAL2 )</tt></b></dt>
2041 <dd>Performs the <a href="#i_fcmp">fcmp operation</a> on constants.</dd>
2042
2043 <dt><b><tt>extractelement ( VAL, IDX )</tt></b></dt>
2044
2045 <dd>Perform the <a href="#i_extractelement">extractelement
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2046 operation</a> on constants.</dd>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2047
2048 <dt><b><tt>insertelement ( VAL, ELT, IDX )</tt></b></dt>
2049
2050 <dd>Perform the <a href="#i_insertelement">insertelement
2051 operation</a> on constants.</dd>
2052
2053
2054 <dt><b><tt>shufflevector ( VEC1, VEC2, IDXMASK )</tt></b></dt>
2055
2056 <dd>Perform the <a href="#i_shufflevector">shufflevector
2057 operation</a> on constants.</dd>
2058
2059 <dt><b><tt>OPCODE ( LHS, RHS )</tt></b></dt>
2060
2061 <dd>Perform the specified operation of the LHS and RHS constants. OPCODE may
2062 be any of the <a href="#binaryops">binary</a> or <a href="#bitwiseops">bitwise
2063 binary</a> operations. The constraints on operands are the same as those for
2064 the corresponding instruction (e.g. no bitwise operations on floating point
2065 values are allowed).</dd>
2066</dl>
2067</div>
2068
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2069<!-- ======================================================================= -->
2070<div class="doc_subsection"><a name="metadata">Embedded Metadata</a>
2071</div>
2072
2073<div class="doc_text">
2074
2075<p>Embedded metadata provides a way to attach arbitrary data to the
2076instruction stream without affecting the behaviour of the program. There are
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]2077two metadata primitives, strings and nodes. All metadata has the
2078<tt>metadata</tt> type and is identified in syntax by a preceding exclamation
2079point ('<tt>!</tt>').
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2080</p>
2081
2082<p>A metadata string is a string surrounded by double quotes. It can contain
2083any character by escaping non-printable characters with "\xx" where "xx" is
2084the two digit hex code. For example: "<tt>!"test\00"</tt>".
2085</p>
2086
2087<p>Metadata nodes are represented with notation similar to structure constants
2088(a comma separated list of elements, surrounded by braces and preceeded by an
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]2089exclamation point). For example: "<tt>!{ metadata !"test\00", i32 10}</tt>".
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2090</p>
2091
Nick Lewycky117f4382009-05-10 20:57:05 +0000[diff] [blame]2092<p>A metadata node will attempt to track changes to the values it holds. In
2093the event that a value is deleted, it will be replaced with a typeless
Nick Lewycky29aaef82009-05-30 05:06:04 +0000[diff] [blame]2094"<tt>null</tt>", such as "<tt>metadata !{null, i32 10}</tt>".</p>
Nick Lewycky117f4382009-05-10 20:57:05 +0000[diff] [blame]2095
Nick Lewycky4dcf8102009-04-04 07:22:01 +0000[diff] [blame]2096<p>Optimizations may rely on metadata to provide additional information about
2097the program that isn't available in the instructions, or that isn't easily
2098computable. Similarly, the code generator may expect a certain metadata format
2099to be used to express debugging information.</p>
2100</div>
2101
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2102<!-- *********************************************************************** -->
2103<div class="doc_section"> <a name="othervalues">Other Values</a> </div>
2104<!-- *********************************************************************** -->
2105
2106<!-- ======================================================================= -->
2107<div class="doc_subsection">
2108<a name="inlineasm">Inline Assembler Expressions</a>
2109</div>
2110
2111<div class="doc_text">
2112
2113<p>
2114LLVM supports inline assembler expressions (as opposed to <a href="#moduleasm">
2115Module-Level Inline Assembly</a>) through the use of a special value. This
2116value represents the inline assembler as a string (containing the instructions
2117to emit), a list of operand constraints (stored as a string), and a flag that
2118indicates whether or not the inline asm expression has side effects. An example
2119inline assembler expression is:
2120</p>
2121
2122<div class="doc_code">
2123<pre>
2124i32 (i32) asm "bswap $0", "=r,r"
2125</pre>
2126</div>
2127
2128<p>
2129Inline assembler expressions may <b>only</b> be used as the callee operand of
2130a <a href="#i_call"><tt>call</tt> instruction</a>. Thus, typically we have:
2131</p>
2132
2133<div class="doc_code">
2134<pre>
2135%X = call i32 asm "<a href="#int_bswap">bswap</a> $0", "=r,r"(i32 %Y)
2136</pre>
2137</div>
2138
2139<p>
2140Inline asms with side effects not visible in the constraint list must be marked
2141as having side effects. This is done through the use of the
2142'<tt>sideeffect</tt>' keyword, like so:
2143</p>
2144
2145<div class="doc_code">
2146<pre>
2147call void asm sideeffect "eieio", ""()
2148</pre>
2149</div>
2150
2151<p>TODO: The format of the asm and constraints string still need to be
2152documented here. Constraints on what can be done (e.g. duplication, moving, etc
Chris Lattner1cdd64e2008-10-04 18:36:02 +0000[diff] [blame]2153need to be documented). This is probably best done by reference to another
2154document that covers inline asm from a holistic perspective.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2155</p>
2156
2157</div>
2158
2159<!-- *********************************************************************** -->
2160<div class="doc_section"> <a name="instref">Instruction Reference</a> </div>
2161<!-- *********************************************************************** -->
2162
2163<div class="doc_text">
2164
2165<p>The LLVM instruction set consists of several different
2166classifications of instructions: <a href="#terminators">terminator
2167instructions</a>, <a href="#binaryops">binary instructions</a>,
2168<a href="#bitwiseops">bitwise binary instructions</a>, <a
2169 href="#memoryops">memory instructions</a>, and <a href="#otherops">other
2170instructions</a>.</p>
2171
2172</div>
2173
2174<!-- ======================================================================= -->
2175<div class="doc_subsection"> <a name="terminators">Terminator
2176Instructions</a> </div>
2177
2178<div class="doc_text">
2179
2180<p>As mentioned <a href="#functionstructure">previously</a>, every
2181basic block in a program ends with a "Terminator" instruction, which
2182indicates which block should be executed after the current block is
2183finished. These terminator instructions typically yield a '<tt>void</tt>'
2184value: they produce control flow, not values (the one exception being
2185the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction).</p>
2186<p>There are six different terminator instructions: the '<a
2187 href="#i_ret"><tt>ret</tt></a>' instruction, the '<a href="#i_br"><tt>br</tt></a>'
2188instruction, the '<a href="#i_switch"><tt>switch</tt></a>' instruction,
2189the '<a href="#i_invoke"><tt>invoke</tt></a>' instruction, the '<a
2190 href="#i_unwind"><tt>unwind</tt></a>' instruction, and the '<a
2191 href="#i_unreachable"><tt>unreachable</tt></a>' instruction.</p>
2192
2193</div>
2194
2195<!-- _______________________________________________________________________ -->
2196<div class="doc_subsubsection"> <a name="i_ret">'<tt>ret</tt>'
2197Instruction</a> </div>
2198<div class="doc_text">
2199<h5>Syntax:</h5>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2200<pre>
2201 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]2202 ret void <i>; Return from void function</i>
2203</pre>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2204
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2205<h5>Overview:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2206
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2207<p>The '<tt>ret</tt>' instruction is used to return control flow (and
2208optionally a value) from a function back to the caller.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2209<p>There are two forms of the '<tt>ret</tt>' instruction: one that
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2210returns a value and then causes control flow, and one that just causes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2211control flow to occur.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2212
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2213<h5>Arguments:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2214
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]2215<p>The '<tt>ret</tt>' instruction optionally accepts a single argument,
2216the return value. The type of the return value must be a
2217'<a href="#t_firstclass">first class</a>' type.</p>
2218
2219<p>A function is not <a href="#wellformed">well formed</a> if
2220it it has a non-void return type and contains a '<tt>ret</tt>'
2221instruction with no return value or a return value with a type that
2222does not match its type, or if it has a void return type and contains
2223a '<tt>ret</tt>' instruction with a return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2224
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2225<h5>Semantics:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2226
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2227<p>When the '<tt>ret</tt>' instruction is executed, control flow
2228returns back to the calling function's context. If the caller is a "<a
2229 href="#i_call"><tt>call</tt></a>" instruction, execution continues at
2230the instruction after the call. If the caller was an "<a
2231 href="#i_invoke"><tt>invoke</tt></a>" instruction, execution continues
2232at the beginning of the "normal" destination block. If the instruction
2233returns a value, that value shall set the call or invoke instruction's
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2234return value.</p>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2235
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2236<h5>Example:</h5>
Chris Lattner43030e72008-04-23 04:59:35 +0000[diff] [blame]2237
2238<pre>
2239 ret i32 5 <i>; Return an integer value of 5</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2240 ret void <i>; Return from a void function</i>
Bill Wendlingd163e2d2009-02-28 22:12:54 +0000[diff] [blame]2241 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]2242</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2243
djge93155c2009-01-24 15:58:40 +0000[diff] [blame]2244<p>Note that the code generator does not yet fully support large
2245 return values. The specific sizes that are currently supported are
2246 dependent on the target. For integers, on 32-bit targets the limit
2247 is often 64 bits, and on 64-bit targets the limit is often 128 bits.
2248 For aggregate types, the current limits are dependent on the element
2249 types; for example targets are often limited to 2 total integer
2250 elements and 2 total floating-point elements.</p>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]2251
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2252</div>
2253<!-- _______________________________________________________________________ -->
2254<div class="doc_subsubsection"> <a name="i_br">'<tt>br</tt>' Instruction</a> </div>
2255<div class="doc_text">
2256<h5>Syntax:</h5>
2257<pre> br i1 &lt;cond&gt;, label &lt;iftrue&gt;, label &lt;iffalse&gt;<br> br label &lt;dest&gt; <i>; Unconditional branch</i>
2258</pre>
2259<h5>Overview:</h5>
2260<p>The '<tt>br</tt>' instruction is used to cause control flow to
2261transfer to a different basic block in the current function. There are
2262two forms of this instruction, corresponding to a conditional branch
2263and an unconditional branch.</p>
2264<h5>Arguments:</h5>
2265<p>The conditional branch form of the '<tt>br</tt>' instruction takes a
2266single '<tt>i1</tt>' value and two '<tt>label</tt>' values. The
2267unconditional form of the '<tt>br</tt>' instruction takes a single
2268'<tt>label</tt>' value as a target.</p>
2269<h5>Semantics:</h5>
2270<p>Upon execution of a conditional '<tt>br</tt>' instruction, the '<tt>i1</tt>'
2271argument is evaluated. If the value is <tt>true</tt>, control flows
2272to the '<tt>iftrue</tt>' <tt>label</tt> argument. If "cond" is <tt>false</tt>,
2273control flows to the '<tt>iffalse</tt>' <tt>label</tt> argument.</p>
2274<h5>Example:</h5>
Chris Lattner95127832009-05-09 18:11:50 +0000[diff] [blame]2275<pre>Test:<br> %cond = <a href="#i_icmp">icmp</a> eq i32 %a, %b<br> br i1 %cond, label %IfEqual, label %IfUnequal<br>IfEqual:<br> <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2276 href="#i_ret">ret</a> i32 1<br>IfUnequal:<br> <a href="#i_ret">ret</a> i32 0<br></pre>
2277</div>
2278<!-- _______________________________________________________________________ -->
2279<div class="doc_subsubsection">
2280 <a name="i_switch">'<tt>switch</tt>' Instruction</a>
2281</div>
2282
2283<div class="doc_text">
2284<h5>Syntax:</h5>
2285
2286<pre>
2287 switch &lt;intty&gt; &lt;value&gt;, label &lt;defaultdest&gt; [ &lt;intty&gt; &lt;val&gt;, label &lt;dest&gt; ... ]
2288</pre>
2289
2290<h5>Overview:</h5>
2291
2292<p>The '<tt>switch</tt>' instruction is used to transfer control flow to one of
2293several different places. It is a generalization of the '<tt>br</tt>'
2294instruction, allowing a branch to occur to one of many possible
2295destinations.</p>
2296
2297
2298<h5>Arguments:</h5>
2299
2300<p>The '<tt>switch</tt>' instruction uses three parameters: an integer
2301comparison value '<tt>value</tt>', a default '<tt>label</tt>' destination, and
2302an array of pairs of comparison value constants and '<tt>label</tt>'s. The
2303table is not allowed to contain duplicate constant entries.</p>
2304
2305<h5>Semantics:</h5>
2306
2307<p>The <tt>switch</tt> instruction specifies a table of values and
2308destinations. When the '<tt>switch</tt>' instruction is executed, this
2309table is searched for the given value. If the value is found, control flow is
2310transfered to the corresponding destination; otherwise, control flow is
2311transfered to the default destination.</p>
2312
2313<h5>Implementation:</h5>
2314
2315<p>Depending on properties of the target machine and the particular
2316<tt>switch</tt> instruction, this instruction may be code generated in different
2317ways. For example, it could be generated as a series of chained conditional
2318branches or with a lookup table.</p>
2319
2320<h5>Example:</h5>
2321
2322<pre>
2323 <i>; Emulate a conditional br instruction</i>
2324 %Val = <a href="#i_zext">zext</a> i1 %value to i32
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2325 switch i32 %Val, label %truedest [ i32 0, label %falsedest ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2326
2327 <i>; Emulate an unconditional br instruction</i>
2328 switch i32 0, label %dest [ ]
2329
2330 <i>; Implement a jump table:</i>
Dan Gohman01852382009-01-04 23:44:43 +0000[diff] [blame]2331 switch i32 %val, label %otherwise [ i32 0, label %onzero
2332 i32 1, label %onone
2333 i32 2, label %ontwo ]
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2334</pre>
2335</div>
2336
2337<!-- _______________________________________________________________________ -->
2338<div class="doc_subsubsection">
2339 <a name="i_invoke">'<tt>invoke</tt>' Instruction</a>
2340</div>
2341
2342<div class="doc_text">
2343
2344<h5>Syntax:</h5>
2345
2346<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2347 &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]2348 to label &lt;normal label&gt; unwind label &lt;exception label&gt;
2349</pre>
2350
2351<h5>Overview:</h5>
2352
2353<p>The '<tt>invoke</tt>' instruction causes control to transfer to a specified
2354function, with the possibility of control flow transfer to either the
2355'<tt>normal</tt>' label or the
2356'<tt>exception</tt>' label. If the callee function returns with the
2357"<tt><a href="#i_ret">ret</a></tt>" instruction, control flow will return to the
2358"normal" label. If the callee (or any indirect callees) returns with the "<a
2359href="#i_unwind"><tt>unwind</tt></a>" instruction, control is interrupted and
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]2360continued at the dynamically nearest "exception" label.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2361
2362<h5>Arguments:</h5>
2363
2364<p>This instruction requires several arguments:</p>
2365
2366<ol>
2367 <li>
2368 The optional "cconv" marker indicates which <a href="#callingconv">calling
2369 convention</a> the call should use. If none is specified, the call defaults
2370 to using C calling conventions.
2371 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2372
2373 <li>The optional <a href="#paramattrs">Parameter Attributes</a> list for
2374 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
2375 and '<tt>inreg</tt>' attributes are valid here.</li>
2376
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2377 <li>'<tt>ptr to function ty</tt>': shall be the signature of the pointer to
2378 function value being invoked. In most cases, this is a direct function
2379 invocation, but indirect <tt>invoke</tt>s are just as possible, branching off
2380 an arbitrary pointer to function value.
2381 </li>
2382
2383 <li>'<tt>function ptr val</tt>': An LLVM value containing a pointer to a
2384 function to be invoked. </li>
2385
2386 <li>'<tt>function args</tt>': argument list whose types match the function
2387 signature argument types. If the function signature indicates the function
2388 accepts a variable number of arguments, the extra arguments can be
2389 specified. </li>
2390
2391 <li>'<tt>normal label</tt>': the label reached when the called function
2392 executes a '<tt><a href="#i_ret">ret</a></tt>' instruction. </li>
2393
2394 <li>'<tt>exception label</tt>': the label reached when a callee returns with
2395 the <a href="#i_unwind"><tt>unwind</tt></a> instruction. </li>
2396
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]2397 <li>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]2398 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
2399 '<tt>readnone</tt>' attributes are valid here.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2400</ol>
2401
2402<h5>Semantics:</h5>
2403
2404<p>This instruction is designed to operate as a standard '<tt><a
2405href="#i_call">call</a></tt>' instruction in most regards. The primary
2406difference is that it establishes an association with a label, which is used by
2407the runtime library to unwind the stack.</p>
2408
2409<p>This instruction is used in languages with destructors to ensure that proper
2410cleanup is performed in the case of either a <tt>longjmp</tt> or a thrown
2411exception. Additionally, this is important for implementation of
2412'<tt>catch</tt>' clauses in high-level languages that support them.</p>
2413
Jay Foad8e2fd2c2009-06-03 10:20:10 +0000[diff] [blame]2414<p>For the purposes of the SSA form, the definition of the value
2415returned by the '<tt>invoke</tt>' instruction is deemed to occur on
2416the edge from the current block to the "normal" label. If the callee
2417unwinds then no return value is available.</p>
Dan Gohman140ba5d2009-05-22 21:47:08 +0000[diff] [blame]2418
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2419<h5>Example:</h5>
2420<pre>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2421 %retval = invoke i32 @Test(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2422 unwind label %TestCleanup <i>; {i32}:retval set</i>
Nick Lewyckya1c11a12008-03-16 07:18:12 +0000[diff] [blame]2423 %retval = invoke <a href="#callingconv">coldcc</a> i32 %Testfnptr(i32 15) to label %Continue
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2424 unwind label %TestCleanup <i>; {i32}:retval set</i>
2425</pre>
2426</div>
2427
2428
2429<!-- _______________________________________________________________________ -->
2430
2431<div class="doc_subsubsection"> <a name="i_unwind">'<tt>unwind</tt>'
2432Instruction</a> </div>
2433
2434<div class="doc_text">
2435
2436<h5>Syntax:</h5>
2437<pre>
2438 unwind
2439</pre>
2440
2441<h5>Overview:</h5>
2442
2443<p>The '<tt>unwind</tt>' instruction unwinds the stack, continuing control flow
2444at the first callee in the dynamic call stack which used an <a
2445href="#i_invoke"><tt>invoke</tt></a> instruction to perform the call. This is
2446primarily used to implement exception handling.</p>
2447
2448<h5>Semantics:</h5>
2449
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]2450<p>The '<tt>unwind</tt>' instruction causes execution of the current function to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2451immediately halt. The dynamic call stack is then searched for the first <a
2452href="#i_invoke"><tt>invoke</tt></a> instruction on the call stack. Once found,
2453execution continues at the "exceptional" destination block specified by the
2454<tt>invoke</tt> instruction. If there is no <tt>invoke</tt> instruction in the
2455dynamic call chain, undefined behavior results.</p>
2456</div>
2457
2458<!-- _______________________________________________________________________ -->
2459
2460<div class="doc_subsubsection"> <a name="i_unreachable">'<tt>unreachable</tt>'
2461Instruction</a> </div>
2462
2463<div class="doc_text">
2464
2465<h5>Syntax:</h5>
2466<pre>
2467 unreachable
2468</pre>
2469
2470<h5>Overview:</h5>
2471
2472<p>The '<tt>unreachable</tt>' instruction has no defined semantics. This
2473instruction is used to inform the optimizer that a particular portion of the
2474code is not reachable. This can be used to indicate that the code after a
2475no-return function cannot be reached, and other facts.</p>
2476
2477<h5>Semantics:</h5>
2478
2479<p>The '<tt>unreachable</tt>' instruction has no defined semantics.</p>
2480</div>
2481
2482
2483
2484<!-- ======================================================================= -->
2485<div class="doc_subsection"> <a name="binaryops">Binary Operations</a> </div>
2486<div class="doc_text">
2487<p>Binary operators are used to do most of the computation in a
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2488program. They require two operands of the same type, execute an operation on them, and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2489produce a single value. The operands might represent
2490multiple data, as is the case with the <a href="#t_vector">vector</a> data type.
Chris Lattnerab596d92008-04-01 18:47:32 +0000[diff] [blame]2491The result value has the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2492<p>There are several different binary operators:</p>
2493</div>
2494<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2495<div class="doc_subsubsection">
2496 <a name="i_add">'<tt>add</tt>' Instruction</a>
2497</div>
2498
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2499<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2500
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2501<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2502
2503<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2504 &lt;result&gt; = add &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2505</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2506
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2507<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2508
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2509<p>The '<tt>add</tt>' instruction returns the sum of its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2510
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2511<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2512
2513<p>The two arguments to the '<tt>add</tt>' instruction must be <a
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2514 href="#t_integer">integer</a> or
2515 <a href="#t_vector">vector</a> of integer values. Both arguments must
2516 have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2517
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2518<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2519
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2520<p>The value produced is the integer sum of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2521
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2522<p>If the sum has unsigned overflow, the result returned is the
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2523mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2524the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2525
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2526<p>Because LLVM integers use a two's complement representation, this
2527instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2528
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2529<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2530
2531<pre>
2532 &lt;result&gt; = add i32 4, %var <i>; yields {i32}:result = 4 + %var</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2533</pre>
2534</div>
2535<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2536<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2537 <a name="i_fadd">'<tt>fadd</tt>' Instruction</a>
2538</div>
2539
2540<div class="doc_text">
2541
2542<h5>Syntax:</h5>
2543
2544<pre>
2545 &lt;result&gt; = fadd &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2546</pre>
2547
2548<h5>Overview:</h5>
2549
2550<p>The '<tt>fadd</tt>' instruction returns the sum of its two operands.</p>
2551
2552<h5>Arguments:</h5>
2553
2554<p>The two arguments to the '<tt>fadd</tt>' instruction must be
2555<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a> of
2556floating point values. Both arguments must have identical types.</p>
2557
2558<h5>Semantics:</h5>
2559
2560<p>The value produced is the floating point sum of the two operands.</p>
2561
2562<h5>Example:</h5>
2563
2564<pre>
2565 &lt;result&gt; = fadd float 4.0, %var <i>; yields {float}:result = 4.0 + %var</i>
2566</pre>
2567</div>
2568<!-- _______________________________________________________________________ -->
2569<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2570 <a name="i_sub">'<tt>sub</tt>' Instruction</a>
2571</div>
2572
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2573<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2574
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2575<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2576
2577<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2578 &lt;result&gt; = sub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2579</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2580
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2581<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2582
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2583<p>The '<tt>sub</tt>' instruction returns the difference of its two
2584operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2585
2586<p>Note that the '<tt>sub</tt>' instruction is used to represent the
2587'<tt>neg</tt>' instruction present in most other intermediate
2588representations.</p>
2589
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2590<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2591
2592<p>The two arguments to the '<tt>sub</tt>' instruction must be <a
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2593 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of
2594 integer values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2595
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2596<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2597
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2598<p>The value produced is the integer difference of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2599
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2600<p>If the difference has unsigned overflow, the result returned is the
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2601mathematical result modulo 2<sup>n</sup>, where n is the bit width of
2602the result.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2603
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2604<p>Because LLVM integers use a two's complement representation, this
2605instruction is appropriate for both signed and unsigned integers.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2606
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2607<h5>Example:</h5>
2608<pre>
2609 &lt;result&gt; = sub i32 4, %var <i>; yields {i32}:result = 4 - %var</i>
2610 &lt;result&gt; = sub i32 0, %val <i>; yields {i32}:result = -%var</i>
2611</pre>
2612</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2613
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2614<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2615<div class="doc_subsubsection">
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2616 <a name="i_fsub">'<tt>fsub</tt>' Instruction</a>
2617</div>
2618
2619<div class="doc_text">
2620
2621<h5>Syntax:</h5>
2622
2623<pre>
2624 &lt;result&gt; = fsub &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2625</pre>
2626
2627<h5>Overview:</h5>
2628
2629<p>The '<tt>fsub</tt>' instruction returns the difference of its two
2630operands.</p>
2631
2632<p>Note that the '<tt>fsub</tt>' instruction is used to represent the
2633'<tt>fneg</tt>' instruction present in most other intermediate
2634representations.</p>
2635
2636<h5>Arguments:</h5>
2637
2638<p>The two arguments to the '<tt>fsub</tt>' instruction must be <a
2639 <a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2640 of floating point values. Both arguments must have identical types.</p>
2641
2642<h5>Semantics:</h5>
2643
2644<p>The value produced is the floating point difference of the two operands.</p>
2645
2646<h5>Example:</h5>
2647<pre>
2648 &lt;result&gt; = fsub float 4.0, %var <i>; yields {float}:result = 4.0 - %var</i>
2649 &lt;result&gt; = fsub float -0.0, %val <i>; yields {float}:result = -%var</i>
2650</pre>
2651</div>
2652
2653<!-- _______________________________________________________________________ -->
2654<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2655 <a name="i_mul">'<tt>mul</tt>' Instruction</a>
2656</div>
2657
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2658<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2659
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2660<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2661<pre> &lt;result&gt; = mul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2662</pre>
2663<h5>Overview:</h5>
2664<p>The '<tt>mul</tt>' instruction returns the product of its two
2665operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2666
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2667<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2668
2669<p>The two arguments to the '<tt>mul</tt>' instruction must be <a
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2670href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2671values. Both arguments must have identical types.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2672
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2673<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2674
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2675<p>The value produced is the integer product of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2676
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2677<p>If the result of the multiplication has unsigned overflow,
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2678the result returned is the mathematical result modulo
26792<sup>n</sup>, where n is the bit width of the result.</p>
2680<p>Because LLVM integers use a two's complement representation, and the
2681result is the same width as the operands, this instruction returns the
2682correct result for both signed and unsigned integers. If a full product
2683(e.g. <tt>i32</tt>x<tt>i32</tt>-><tt>i64</tt>) is needed, the operands
2684should be sign-extended or zero-extended as appropriate to the
2685width of the full product.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2686<h5>Example:</h5>
2687<pre> &lt;result&gt; = mul i32 4, %var <i>; yields {i32}:result = 4 * %var</i>
2688</pre>
2689</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2690
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2691<!-- _______________________________________________________________________ -->
Dan Gohman7ce405e2009-06-04 22:49:04 +0000[diff] [blame]2692<div class="doc_subsubsection">
2693 <a name="i_fmul">'<tt>fmul</tt>' Instruction</a>
2694</div>
2695
2696<div class="doc_text">
2697
2698<h5>Syntax:</h5>
2699<pre> &lt;result&gt; = fmul &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
2700</pre>
2701<h5>Overview:</h5>
2702<p>The '<tt>fmul</tt>' instruction returns the product of its two
2703operands.</p>
2704
2705<h5>Arguments:</h5>
2706
2707<p>The two arguments to the '<tt>fmul</tt>' instruction must be
2708<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2709of floating point values. Both arguments must have identical types.</p>
2710
2711<h5>Semantics:</h5>
2712
2713<p>The value produced is the floating point product of the two operands.</p>
2714
2715<h5>Example:</h5>
2716<pre> &lt;result&gt; = fmul float 4.0, %var <i>; yields {float}:result = 4.0 * %var</i>
2717</pre>
2718</div>
2719
2720<!-- _______________________________________________________________________ -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2721<div class="doc_subsubsection"> <a name="i_udiv">'<tt>udiv</tt>' Instruction
2722</a></div>
2723<div class="doc_text">
2724<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2725<pre> &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]2726</pre>
2727<h5>Overview:</h5>
2728<p>The '<tt>udiv</tt>' instruction returns the quotient of its two
2729operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2730
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2731<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2732
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2733<p>The two arguments to the '<tt>udiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2734<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2735values. Both arguments must have identical types.</p>
2736
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2737<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2738
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2739<p>The value produced is the unsigned integer quotient of the two operands.</p>
2740<p>Note that unsigned integer division and signed integer division are distinct
2741operations; for signed integer division, use '<tt>sdiv</tt>'.</p>
2742<p>Division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2743<h5>Example:</h5>
2744<pre> &lt;result&gt; = udiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2745</pre>
2746</div>
2747<!-- _______________________________________________________________________ -->
2748<div class="doc_subsubsection"> <a name="i_sdiv">'<tt>sdiv</tt>' Instruction
2749</a> </div>
2750<div class="doc_text">
2751<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2752<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2753 &lt;result&gt; = sdiv &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2754</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2755
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2756<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2757
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2758<p>The '<tt>sdiv</tt>' instruction returns the quotient of its two
2759operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2760
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2761<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2762
2763<p>The two arguments to the '<tt>sdiv</tt>' instruction must be
2764<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2765values. Both arguments must have identical types.</p>
2766
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2767<h5>Semantics:</h5>
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2768<p>The value produced is the signed integer quotient of the two operands rounded towards zero.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2769<p>Note that signed integer division and unsigned integer division are distinct
2770operations; for unsigned integer division, use '<tt>udiv</tt>'.</p>
2771<p>Division by zero leads to undefined behavior. Overflow also leads to
2772undefined behavior; this is a rare case, but can occur, for example,
2773by doing a 32-bit division of -2147483648 by -1.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2774<h5>Example:</h5>
2775<pre> &lt;result&gt; = sdiv i32 4, %var <i>; yields {i32}:result = 4 / %var</i>
2776</pre>
2777</div>
2778<!-- _______________________________________________________________________ -->
2779<div class="doc_subsubsection"> <a name="i_fdiv">'<tt>fdiv</tt>'
2780Instruction</a> </div>
2781<div class="doc_text">
2782<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2783<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2784 &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]2785</pre>
2786<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2787
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2788<p>The '<tt>fdiv</tt>' instruction returns the quotient of its two
2789operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2790
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2791<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2792
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2793<p>The two arguments to the '<tt>fdiv</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2794<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2795of floating point values. Both arguments must have identical types.</p>
2796
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2797<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2798
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2799<p>The value produced is the floating point quotient of the two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2800
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2801<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2802
2803<pre>
2804 &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]2805</pre>
2806</div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2807
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2808<!-- _______________________________________________________________________ -->
2809<div class="doc_subsubsection"> <a name="i_urem">'<tt>urem</tt>' Instruction</a>
2810</div>
2811<div class="doc_text">
2812<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2813<pre> &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]2814</pre>
2815<h5>Overview:</h5>
2816<p>The '<tt>urem</tt>' instruction returns the remainder from the
2817unsigned division of its two arguments.</p>
2818<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2819<p>The two arguments to the '<tt>urem</tt>' instruction must be
2820<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2821values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2822<h5>Semantics:</h5>
2823<p>This instruction returns the unsigned integer <i>remainder</i> of a division.
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2824This instruction always performs an unsigned division to get the remainder.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2825<p>Note that unsigned integer remainder and signed integer remainder are
2826distinct operations; for signed integer remainder, use '<tt>srem</tt>'.</p>
2827<p>Taking the remainder of a division by zero leads to undefined behavior.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2828<h5>Example:</h5>
2829<pre> &lt;result&gt; = urem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2830</pre>
2831
2832</div>
2833<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2834<div class="doc_subsubsection">
2835 <a name="i_srem">'<tt>srem</tt>' Instruction</a>
2836</div>
2837
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2838<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2839
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2840<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2841
2842<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2843 &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]2844</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2845
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2846<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2847
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2848<p>The '<tt>srem</tt>' instruction returns the remainder from the
Dan Gohman3e3fd8c2007-11-05 23:35:22 +0000[diff] [blame]2849signed division of its two operands. This instruction can also take
2850<a href="#t_vector">vector</a> versions of the values in which case
2851the elements must be integers.</p>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]2852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2853<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2854
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2855<p>The two arguments to the '<tt>srem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2856<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
2857values. Both arguments must have identical types.</p>
2858
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2859<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2860
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2861<p>This instruction returns the <i>remainder</i> of a division (where the result
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2862has the same sign as the dividend, <tt>op1</tt>), not the <i>modulo</i>
2863operator (where the result has the same sign as the divisor, <tt>op2</tt>) of
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2864a value. For more information about the difference, see <a
2865 href="http://mathforum.org/dr.math/problems/anne.4.28.99.html">The
2866Math Forum</a>. For a table of how this is implemented in various languages,
2867please see <a href="http://en.wikipedia.org/wiki/Modulo_operation">
2868Wikipedia: modulo operation</a>.</p>
Chris Lattner9aba1e22008-01-28 00:36:27 +0000[diff] [blame]2869<p>Note that signed integer remainder and unsigned integer remainder are
2870distinct operations; for unsigned integer remainder, use '<tt>urem</tt>'.</p>
2871<p>Taking the remainder of a division by zero leads to undefined behavior.
2872Overflow also leads to undefined behavior; this is a rare case, but can occur,
2873for example, by taking the remainder of a 32-bit division of -2147483648 by -1.
2874(The remainder doesn't actually overflow, but this rule lets srem be
2875implemented using instructions that return both the result of the division
2876and the remainder.)</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2877<h5>Example:</h5>
2878<pre> &lt;result&gt; = srem i32 4, %var <i>; yields {i32}:result = 4 % %var</i>
2879</pre>
2880
2881</div>
2882<!-- _______________________________________________________________________ -->
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2883<div class="doc_subsubsection">
2884 <a name="i_frem">'<tt>frem</tt>' Instruction</a> </div>
2885
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2886<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2887
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2888<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2889<pre> &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]2890</pre>
2891<h5>Overview:</h5>
2892<p>The '<tt>frem</tt>' instruction returns the remainder from the
2893division of its two operands.</p>
2894<h5>Arguments:</h5>
2895<p>The two arguments to the '<tt>frem</tt>' instruction must be
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2896<a href="#t_floating">floating point</a> or <a href="#t_vector">vector</a>
2897of floating point values. Both arguments must have identical types.</p>
2898
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2899<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2900
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2901<p>This instruction returns the <i>remainder</i> of a division.
2902The remainder has the same sign as the dividend.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2903
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2904<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]2905
2906<pre>
2907 &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]2908</pre>
2909</div>
2910
2911<!-- ======================================================================= -->
2912<div class="doc_subsection"> <a name="bitwiseops">Bitwise Binary
2913Operations</a> </div>
2914<div class="doc_text">
2915<p>Bitwise binary operators are used to do various forms of
2916bit-twiddling in a program. They are generally very efficient
2917instructions and can commonly be strength reduced from other
Chris Lattnerdc15b1d2008-04-01 18:45:27 +0000[diff] [blame]2918instructions. They require two operands of the same type, execute an operation on them,
2919and produce a single value. The resulting value is the same type as its operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2920</div>
2921
2922<!-- _______________________________________________________________________ -->
2923<div class="doc_subsubsection"> <a name="i_shl">'<tt>shl</tt>'
2924Instruction</a> </div>
2925<div class="doc_text">
2926<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2927<pre> &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]2928</pre>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2929
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2930<h5>Overview:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2931
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2932<p>The '<tt>shl</tt>' instruction returns the first operand shifted to
2933the left a specified number of bits.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2934
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2935<h5>Arguments:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2936
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2937<p>Both arguments to the '<tt>shl</tt>' instruction must be the same <a
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2938 href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2939type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2940
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2941<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2942
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2943<p>The value produced is <tt>op1</tt> * 2<sup><tt>op2</tt></sup> mod 2<sup>n</sup>,
2944where n is the width of the result. If <tt>op2</tt> is (statically or dynamically) negative or
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2945equal to or larger than the number of bits in <tt>op1</tt>, the result is undefined.
2946If the arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
2947corresponding shift amount in <tt>op2</tt>.</p>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2948
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2949<h5>Example:</h5><pre>
2950 &lt;result&gt; = shl i32 4, %var <i>; yields {i32}: 4 &lt;&lt; %var</i>
2951 &lt;result&gt; = shl i32 4, 2 <i>; yields {i32}: 16</i>
2952 &lt;result&gt; = shl i32 1, 10 <i>; yields {i32}: 1024</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2953 &lt;result&gt; = shl i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2954 &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]2955</pre>
2956</div>
2957<!-- _______________________________________________________________________ -->
2958<div class="doc_subsubsection"> <a name="i_lshr">'<tt>lshr</tt>'
2959Instruction</a> </div>
2960<div class="doc_text">
2961<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2962<pre> &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]2963</pre>
2964
2965<h5>Overview:</h5>
2966<p>The '<tt>lshr</tt>' instruction (logical shift right) returns the first
2967operand shifted to the right a specified number of bits with zero fill.</p>
2968
2969<h5>Arguments:</h5>
2970<p>Both arguments to the '<tt>lshr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]2971<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2972type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2973
2974<h5>Semantics:</h5>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2975
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2976<p>This instruction always performs a logical shift right operation. The most
2977significant bits of the result will be filled with zero bits after the
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]2978shift. If <tt>op2</tt> is (statically or dynamically) equal to or larger than
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]2979the number of bits in <tt>op1</tt>, the result is undefined. If the arguments are
2980vectors, each vector element of <tt>op1</tt> is shifted by the corresponding shift
2981amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2982
2983<h5>Example:</h5>
2984<pre>
2985 &lt;result&gt; = lshr i32 4, 1 <i>; yields {i32}:result = 2</i>
2986 &lt;result&gt; = lshr i32 4, 2 <i>; yields {i32}:result = 1</i>
2987 &lt;result&gt; = lshr i8 4, 3 <i>; yields {i8}:result = 0</i>
2988 &lt;result&gt; = lshr i8 -2, 1 <i>; yields {i8}:result = 0x7FFFFFFF </i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]2989 &lt;result&gt; = lshr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]2990 &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]2991</pre>
2992</div>
2993
2994<!-- _______________________________________________________________________ -->
2995<div class="doc_subsubsection"> <a name="i_ashr">'<tt>ashr</tt>'
2996Instruction</a> </div>
2997<div class="doc_text">
2998
2999<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3000<pre> &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]3001</pre>
3002
3003<h5>Overview:</h5>
3004<p>The '<tt>ashr</tt>' instruction (arithmetic shift right) returns the first
3005operand shifted to the right a specified number of bits with sign extension.</p>
3006
3007<h5>Arguments:</h5>
3008<p>Both arguments to the '<tt>ashr</tt>' instruction must be the same
Nate Begemanbb1ce942008-07-29 15:49:41 +0000[diff] [blame]3009<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3010type. '<tt>op2</tt>' is treated as an unsigned value.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3011
3012<h5>Semantics:</h5>
3013<p>This instruction always performs an arithmetic shift right operation,
3014The most significant bits of the result will be filled with the sign bit
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3015of <tt>op1</tt>. If <tt>op2</tt> is (statically or dynamically) equal to or
Mon P Wangb0f51822008-12-10 08:55:09 +0000[diff] [blame]3016larger than the number of bits in <tt>op1</tt>, the result is undefined. If the
3017arguments are vectors, each vector element of <tt>op1</tt> is shifted by the
3018corresponding shift amount in <tt>op2</tt>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3019
3020<h5>Example:</h5>
3021<pre>
3022 &lt;result&gt; = ashr i32 4, 1 <i>; yields {i32}:result = 2</i>
3023 &lt;result&gt; = ashr i32 4, 2 <i>; yields {i32}:result = 1</i>
3024 &lt;result&gt; = ashr i8 4, 3 <i>; yields {i8}:result = 0</i>
3025 &lt;result&gt; = ashr i8 -2, 1 <i>; yields {i8}:result = -1</i>
Chris Lattnerd939d9f2007-10-03 21:01:14 +0000[diff] [blame]3026 &lt;result&gt; = ashr i32 1, 32 <i>; undefined</i>
Mon P Wang9901e732008-12-09 05:46:39 +0000[diff] [blame]3027 &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]3028</pre>
3029</div>
3030
3031<!-- _______________________________________________________________________ -->
3032<div class="doc_subsubsection"> <a name="i_and">'<tt>and</tt>'
3033Instruction</a> </div>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3034
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3035<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3036
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3037<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3038
3039<pre>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3040 &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]3041</pre>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3042
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3043<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3044
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3045<p>The '<tt>and</tt>' instruction returns the bitwise logical and of
3046its two operands.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3047
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3048<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3049
3050<p>The two arguments to the '<tt>and</tt>' instruction must be
3051<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3052values. Both arguments must have identical types.</p>
3053
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3054<h5>Semantics:</h5>
3055<p>The truth table used for the '<tt>and</tt>' instruction is:</p>
3056<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]3057<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3058<table border="1" cellspacing="0" cellpadding="4">
3059 <tbody>
3060 <tr>
3061 <td>In0</td>
3062 <td>In1</td>
3063 <td>Out</td>
3064 </tr>
3065 <tr>
3066 <td>0</td>
3067 <td>0</td>
3068 <td>0</td>
3069 </tr>
3070 <tr>
3071 <td>0</td>
3072 <td>1</td>
3073 <td>0</td>
3074 </tr>
3075 <tr>
3076 <td>1</td>
3077 <td>0</td>
3078 <td>0</td>
3079 </tr>
3080 <tr>
3081 <td>1</td>
3082 <td>1</td>
3083 <td>1</td>
3084 </tr>
3085 </tbody>
3086</table>
3087</div>
3088<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3089<pre>
3090 &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]3091 &lt;result&gt; = and i32 15, 40 <i>; yields {i32}:result = 8</i>
3092 &lt;result&gt; = and i32 4, 8 <i>; yields {i32}:result = 0</i>
3093</pre>
3094</div>
3095<!-- _______________________________________________________________________ -->
3096<div class="doc_subsubsection"> <a name="i_or">'<tt>or</tt>' Instruction</a> </div>
3097<div class="doc_text">
3098<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3099<pre> &lt;result&gt; = or &lt;ty&gt; &lt;op1&gt;, &lt;op2&gt; <i>; yields {ty}:result</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3100</pre>
3101<h5>Overview:</h5>
3102<p>The '<tt>or</tt>' instruction returns the bitwise logical inclusive
3103or of its two operands.</p>
3104<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3105
3106<p>The two arguments to the '<tt>or</tt>' instruction must be
3107<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3108values. Both arguments must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3109<h5>Semantics:</h5>
3110<p>The truth table used for the '<tt>or</tt>' instruction is:</p>
3111<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]3112<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3113<table border="1" cellspacing="0" cellpadding="4">
3114 <tbody>
3115 <tr>
3116 <td>In0</td>
3117 <td>In1</td>
3118 <td>Out</td>
3119 </tr>
3120 <tr>
3121 <td>0</td>
3122 <td>0</td>
3123 <td>0</td>
3124 </tr>
3125 <tr>
3126 <td>0</td>
3127 <td>1</td>
3128 <td>1</td>
3129 </tr>
3130 <tr>
3131 <td>1</td>
3132 <td>0</td>
3133 <td>1</td>
3134 </tr>
3135 <tr>
3136 <td>1</td>
3137 <td>1</td>
3138 <td>1</td>
3139 </tr>
3140 </tbody>
3141</table>
3142</div>
3143<h5>Example:</h5>
3144<pre> &lt;result&gt; = or i32 4, %var <i>; yields {i32}:result = 4 | %var</i>
3145 &lt;result&gt; = or i32 15, 40 <i>; yields {i32}:result = 47</i>
3146 &lt;result&gt; = or i32 4, 8 <i>; yields {i32}:result = 12</i>
3147</pre>
3148</div>
3149<!-- _______________________________________________________________________ -->
3150<div class="doc_subsubsection"> <a name="i_xor">'<tt>xor</tt>'
3151Instruction</a> </div>
3152<div class="doc_text">
3153<h5>Syntax:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]3154<pre> &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]3155</pre>
3156<h5>Overview:</h5>
3157<p>The '<tt>xor</tt>' instruction returns the bitwise logical exclusive
3158or of its two operands. The <tt>xor</tt> is used to implement the
3159"one's complement" operation, which is the "~" operator in C.</p>
3160<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3161<p>The two arguments to the '<tt>xor</tt>' instruction must be
3162<a href="#t_integer">integer</a> or <a href="#t_vector">vector</a> of integer
3163values. Both arguments must have identical types.</p>
3164
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3165<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]3166
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3167<p>The truth table used for the '<tt>xor</tt>' instruction is:</p>
3168<p> </p>
Bill Wendling1e19eac2008-09-07 10:29:20 +0000[diff] [blame]3169<div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3170<table border="1" cellspacing="0" cellpadding="4">
3171 <tbody>
3172 <tr>
3173 <td>In0</td>
3174 <td>In1</td>
3175 <td>Out</td>
3176 </tr>
3177 <tr>
3178 <td>0</td>
3179 <td>0</td>
3180 <td>0</td>
3181 </tr>
3182 <tr>
3183 <td>0</td>
3184 <td>1</td>
3185 <td>1</td>
3186 </tr>
3187 <tr>
3188 <td>1</td>
3189 <td>0</td>
3190 <td>1</td>
3191 </tr>
3192 <tr>
3193 <td>1</td>
3194 <td>1</td>
3195 <td>0</td>
3196 </tr>
3197 </tbody>
3198</table>
3199</div>
3200<p> </p>
3201<h5>Example:</h5>
3202<pre> &lt;result&gt; = xor i32 4, %var <i>; yields {i32}:result = 4 ^ %var</i>
3203 &lt;result&gt; = xor i32 15, 40 <i>; yields {i32}:result = 39</i>
3204 &lt;result&gt; = xor i32 4, 8 <i>; yields {i32}:result = 12</i>
3205 &lt;result&gt; = xor i32 %V, -1 <i>; yields {i32}:result = ~%V</i>
3206</pre>
3207</div>
3208
3209<!-- ======================================================================= -->
3210<div class="doc_subsection">
3211 <a name="vectorops">Vector Operations</a>
3212</div>
3213
3214<div class="doc_text">
3215
3216<p>LLVM supports several instructions to represent vector operations in a
3217target-independent manner. These instructions cover the element-access and
3218vector-specific operations needed to process vectors effectively. While LLVM
3219does directly support these vector operations, many sophisticated algorithms
3220will want to use target-specific intrinsics to take full advantage of a specific
3221target.</p>
3222
3223</div>
3224
3225<!-- _______________________________________________________________________ -->
3226<div class="doc_subsubsection">
3227 <a name="i_extractelement">'<tt>extractelement</tt>' Instruction</a>
3228</div>
3229
3230<div class="doc_text">
3231
3232<h5>Syntax:</h5>
3233
3234<pre>
3235 &lt;result&gt; = extractelement &lt;n x &lt;ty&gt;&gt; &lt;val&gt;, i32 &lt;idx&gt; <i>; yields &lt;ty&gt;</i>
3236</pre>
3237
3238<h5>Overview:</h5>
3239
3240<p>
3241The '<tt>extractelement</tt>' instruction extracts a single scalar
3242element from a vector at a specified index.
3243</p>
3244
3245
3246<h5>Arguments:</h5>
3247
3248<p>
3249The first operand of an '<tt>extractelement</tt>' instruction is a
3250value of <a href="#t_vector">vector</a> type. The second operand is
3251an index indicating the position from which to extract the element.
3252The index may be a variable.</p>
3253
3254<h5>Semantics:</h5>
3255
3256<p>
3257The result is a scalar of the same type as the element type of
3258<tt>val</tt>. Its value is the value at position <tt>idx</tt> of
3259<tt>val</tt>. If <tt>idx</tt> exceeds the length of <tt>val</tt>, the
3260results are undefined.
3261</p>
3262
3263<h5>Example:</h5>
3264
3265<pre>
3266 %result = extractelement &lt;4 x i32&gt; %vec, i32 0 <i>; yields i32</i>
3267</pre>
3268</div>
3269
3270
3271<!-- _______________________________________________________________________ -->
3272<div class="doc_subsubsection">
3273 <a name="i_insertelement">'<tt>insertelement</tt>' Instruction</a>
3274</div>
3275
3276<div class="doc_text">
3277
3278<h5>Syntax:</h5>
3279
3280<pre>
Dan Gohmanbcc3c502008-05-12 23:38:42 +0000[diff] [blame]3281 &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]3282</pre>
3283
3284<h5>Overview:</h5>
3285
3286<p>
3287The '<tt>insertelement</tt>' instruction inserts a scalar
3288element into a vector at a specified index.
3289</p>
3290
3291
3292<h5>Arguments:</h5>
3293
3294<p>
3295The first operand of an '<tt>insertelement</tt>' instruction is a
3296value of <a href="#t_vector">vector</a> type. The second operand is a
3297scalar value whose type must equal the element type of the first
3298operand. The third operand is an index indicating the position at
3299which to insert the value. The index may be a variable.</p>
3300
3301<h5>Semantics:</h5>
3302
3303<p>
3304The result is a vector of the same type as <tt>val</tt>. Its
3305element values are those of <tt>val</tt> except at position
3306<tt>idx</tt>, where it gets the value <tt>elt</tt>. If <tt>idx</tt>
3307exceeds the length of <tt>val</tt>, the results are undefined.
3308</p>
3309
3310<h5>Example:</h5>
3311
3312<pre>
3313 %result = insertelement &lt;4 x i32&gt; %vec, i32 1, i32 0 <i>; yields &lt;4 x i32&gt;</i>
3314</pre>
3315</div>
3316
3317<!-- _______________________________________________________________________ -->
3318<div class="doc_subsubsection">
3319 <a name="i_shufflevector">'<tt>shufflevector</tt>' Instruction</a>
3320</div>
3321
3322<div class="doc_text">
3323
3324<h5>Syntax:</h5>
3325
3326<pre>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3327 &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]3328</pre>
3329
3330<h5>Overview:</h5>
3331
3332<p>
3333The '<tt>shufflevector</tt>' instruction constructs a permutation of elements
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3334from two input vectors, returning a vector with the same element type as
3335the input and length that is the same as the shuffle mask.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3336</p>
3337
3338<h5>Arguments:</h5>
3339
3340<p>
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3341The first two operands of a '<tt>shufflevector</tt>' instruction are vectors
3342with types that match each other. The third argument is a shuffle mask whose
3343element type is always 'i32'. The result of the instruction is a vector whose
3344length is the same as the shuffle mask and whose element type is the same as
3345the element type of the first two operands.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3346</p>
3347
3348<p>
3349The shuffle mask operand is required to be a constant vector with either
3350constant integer or undef values.
3351</p>
3352
3353<h5>Semantics:</h5>
3354
3355<p>
3356The elements of the two input vectors are numbered from left to right across
3357both of the vectors. The shuffle mask operand specifies, for each element of
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3358the result vector, which element of the two input vectors the result element
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3359gets. The element selector may be undef (meaning "don't care") and the second
3360operand may be undef if performing a shuffle from only one vector.
3361</p>
3362
3363<h5>Example:</h5>
3364
3365<pre>
3366 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3367 &lt;4 x i32&gt; &lt;i32 0, i32 4, i32 1, i32 5&gt; <i>; yields &lt;4 x i32&gt;</i>
3368 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; undef,
3369 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i> - Identity shuffle.
Mon P Wangbff5d9c2008-11-10 04:46:22 +0000[diff] [blame]3370 %result = shufflevector &lt;8 x i32&gt; %v1, &lt;8 x i32&gt; undef,
3371 &lt;4 x i32&gt; &lt;i32 0, i32 1, i32 2, i32 3&gt; <i>; yields &lt;4 x i32&gt;</i>
3372 %result = shufflevector &lt;4 x i32&gt; %v1, &lt;4 x i32&gt; %v2,
3373 &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]3374</pre>
3375</div>
3376
3377
3378<!-- ======================================================================= -->
3379<div class="doc_subsection">
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3380 <a name="aggregateops">Aggregate Operations</a>
3381</div>
3382
3383<div class="doc_text">
3384
3385<p>LLVM supports several instructions for working with aggregate values.
3386</p>
3387
3388</div>
3389
3390<!-- _______________________________________________________________________ -->
3391<div class="doc_subsubsection">
3392 <a name="i_extractvalue">'<tt>extractvalue</tt>' Instruction</a>
3393</div>
3394
3395<div class="doc_text">
3396
3397<h5>Syntax:</h5>
3398
3399<pre>
3400 &lt;result&gt; = extractvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;idx&gt;{, &lt;idx&gt;}*
3401</pre>
3402
3403<h5>Overview:</h5>
3404
3405<p>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3406The '<tt>extractvalue</tt>' instruction extracts the value of a struct field
3407or array element from an aggregate value.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3408</p>
3409
3410
3411<h5>Arguments:</h5>
3412
3413<p>
3414The first operand of an '<tt>extractvalue</tt>' instruction is a
3415value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a>
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3416type. The operands are constant indices to specify which value to extract
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3417in a similar manner as indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3418'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3419</p>
3420
3421<h5>Semantics:</h5>
3422
3423<p>
3424The result is the value at the position in the aggregate specified by
3425the index operands.
3426</p>
3427
3428<h5>Example:</h5>
3429
3430<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3431 %result = extractvalue {i32, float} %agg, 0 <i>; yields i32</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3432</pre>
3433</div>
3434
3435
3436<!-- _______________________________________________________________________ -->
3437<div class="doc_subsubsection">
3438 <a name="i_insertvalue">'<tt>insertvalue</tt>' Instruction</a>
3439</div>
3440
3441<div class="doc_text">
3442
3443<h5>Syntax:</h5>
3444
3445<pre>
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3446 &lt;result&gt; = insertvalue &lt;aggregate type&gt; &lt;val&gt;, &lt;ty&gt; &lt;val&gt;, &lt;idx&gt; <i>; yields &lt;n x &lt;ty&gt;&gt;</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3447</pre>
3448
3449<h5>Overview:</h5>
3450
3451<p>
3452The '<tt>insertvalue</tt>' instruction inserts a value
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3453into a struct field or array element in an aggregate.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3454</p>
3455
3456
3457<h5>Arguments:</h5>
3458
3459<p>
3460The first operand of an '<tt>insertvalue</tt>' instruction is a
3461value of <a href="#t_struct">struct</a> or <a href="#t_array">array</a> type.
3462The second operand is a first-class value to insert.
Dan Gohman4f29e422008-05-23 21:53:15 +0000[diff] [blame]3463The following operands are constant indices
Dan Gohmane5febe42008-05-31 00:58:22 +0000[diff] [blame]3464indicating the position at which to insert the value in a similar manner as
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3465indices in a
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3466'<tt><a href="#i_getelementptr">getelementptr</a></tt>' instruction.
3467The value to insert must have the same type as the value identified
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3468by the indices.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3469</p>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3470
3471<h5>Semantics:</h5>
3472
3473<p>
3474The result is an aggregate of the same type as <tt>val</tt>. Its
3475value is that of <tt>val</tt> except that the value at the position
Dan Gohman6c6dea02008-05-13 18:16:06 +0000[diff] [blame]3476specified by the indices is that of <tt>elt</tt>.
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3477</p>
3478
3479<h5>Example:</h5>
3480
3481<pre>
Dan Gohmanb1aab4e2008-06-23 15:26:37 +0000[diff] [blame]3482 %result = insertvalue {i32, float} %agg, i32 1, 0 <i>; yields {i32, float}</i>
Dan Gohman74d6faf2008-05-12 23:51:09 +0000[diff] [blame]3483</pre>
3484</div>
3485
3486
3487<!-- ======================================================================= -->
3488<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3489 <a name="memoryops">Memory Access and Addressing Operations</a>
3490</div>
3491
3492<div class="doc_text">
3493
3494<p>A key design point of an SSA-based representation is how it
3495represents memory. In LLVM, no memory locations are in SSA form, which
3496makes things very simple. This section describes how to read, write,
3497allocate, and free memory in LLVM.</p>
3498
3499</div>
3500
3501<!-- _______________________________________________________________________ -->
3502<div class="doc_subsubsection">
3503 <a name="i_malloc">'<tt>malloc</tt>' Instruction</a>
3504</div>
3505
3506<div class="doc_text">
3507
3508<h5>Syntax:</h5>
3509
3510<pre>
3511 &lt;result&gt; = malloc &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3512</pre>
3513
3514<h5>Overview:</h5>
3515
3516<p>The '<tt>malloc</tt>' instruction allocates memory from the system
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3517heap and returns a pointer to it. The object is always allocated in the generic
3518address space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3519
3520<h5>Arguments:</h5>
3521
3522<p>The '<tt>malloc</tt>' instruction allocates
3523<tt>sizeof(&lt;type&gt;)*NumElements</tt>
3524bytes of memory from the operating system and returns a pointer of the
3525appropriate type to the program. If "NumElements" is specified, it is the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3526number of elements allocated, otherwise "NumElements" is defaulted to be one.
Duncan Sandsb38ce6f2009-06-20 13:26:06 +0000[diff] [blame]3527If a constant alignment is specified, the value result of the allocation is
3528guaranteed to be aligned to at least that boundary. If not specified, or if
3529zero, the target can choose to align the allocation on any convenient boundary
3530compatible with the type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3531
3532<p>'<tt>type</tt>' must be a sized type.</p>
3533
3534<h5>Semantics:</h5>
3535
3536<p>Memory is allocated using the system "<tt>malloc</tt>" function, and
Nick Lewyckyff384472008-11-24 03:41:24 +0000[diff] [blame]3537a pointer is returned. The result of a zero byte allocation is undefined. The
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3538result is null if there is insufficient memory available.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3539
3540<h5>Example:</h5>
3541
3542<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3543 %array = malloc [4 x i8] <i>; yields {[%4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3544
3545 %size = <a href="#i_add">add</a> i32 2, 2 <i>; yields {i32}:size = i32 4</i>
3546 %array1 = malloc i8, i32 4 <i>; yields {i8*}:array1</i>
3547 %array2 = malloc [12 x i8], i32 %size <i>; yields {[12 x i8]*}:array2</i>
3548 %array3 = malloc i32, i32 4, align 1024 <i>; yields {i32*}:array3</i>
3549 %array4 = malloc i32, align 1024 <i>; yields {i32*}:array4</i>
3550</pre>
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]3551
3552<p>Note that the code generator does not yet respect the
3553 alignment value.</p>
3554
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3555</div>
3556
3557<!-- _______________________________________________________________________ -->
3558<div class="doc_subsubsection">
3559 <a name="i_free">'<tt>free</tt>' Instruction</a>
3560</div>
3561
3562<div class="doc_text">
3563
3564<h5>Syntax:</h5>
3565
3566<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3567 free &lt;type&gt; &lt;value&gt; <i>; yields {void}</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3568</pre>
3569
3570<h5>Overview:</h5>
3571
3572<p>The '<tt>free</tt>' instruction returns memory back to the unused
3573memory heap to be reallocated in the future.</p>
3574
3575<h5>Arguments:</h5>
3576
3577<p>'<tt>value</tt>' shall be a pointer value that points to a value
3578that was allocated with the '<tt><a href="#i_malloc">malloc</a></tt>'
3579instruction.</p>
3580
3581<h5>Semantics:</h5>
3582
3583<p>Access to the memory pointed to by the pointer is no longer defined
Chris Lattner329d6532008-04-19 22:41:32 +0000[diff] [blame]3584after this instruction executes. If the pointer is null, the operation
3585is a noop.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3586
3587<h5>Example:</h5>
3588
3589<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3590 %array = <a href="#i_malloc">malloc</a> [4 x i8] <i>; yields {[4 x i8]*}:array</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3591 free [4 x i8]* %array
3592</pre>
3593</div>
3594
3595<!-- _______________________________________________________________________ -->
3596<div class="doc_subsubsection">
3597 <a name="i_alloca">'<tt>alloca</tt>' Instruction</a>
3598</div>
3599
3600<div class="doc_text">
3601
3602<h5>Syntax:</h5>
3603
3604<pre>
3605 &lt;result&gt; = alloca &lt;type&gt;[, i32 &lt;NumElements&gt;][, align &lt;alignment&gt;] <i>; yields {type*}:result</i>
3606</pre>
3607
3608<h5>Overview:</h5>
3609
3610<p>The '<tt>alloca</tt>' instruction allocates memory on the stack frame of the
3611currently executing function, to be automatically released when this function
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]3612returns to its caller. The object is always allocated in the generic address
3613space (address space zero).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3614
3615<h5>Arguments:</h5>
3616
3617<p>The '<tt>alloca</tt>' instruction allocates <tt>sizeof(&lt;type&gt;)*NumElements</tt>
3618bytes of memory on the runtime stack, returning a pointer of the
Gabor Greif5082cf42008-02-09 22:24:34 +0000[diff] [blame]3619appropriate type to the program. If "NumElements" is specified, it is the
3620number of elements allocated, otherwise "NumElements" is defaulted to be one.
Duncan Sandsb38ce6f2009-06-20 13:26:06 +0000[diff] [blame]3621If a constant alignment is specified, the value result of the allocation is
3622guaranteed to be aligned to at least that boundary. If not specified, or if
3623zero, the target can choose to align the allocation on any convenient boundary
3624compatible with the type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3625
3626<p>'<tt>type</tt>' may be any sized type.</p>
3627
3628<h5>Semantics:</h5>
3629
Bill Wendling2a454572009-05-08 20:49:29 +0000[diff] [blame]3630<p>Memory is allocated; a pointer is returned. The operation is undefined if
Chris Lattner8b094fc2008-04-19 21:01:16 +0000[diff] [blame]3631there is insufficient stack space for the allocation. '<tt>alloca</tt>'d
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3632memory is automatically released when the function returns. The '<tt>alloca</tt>'
3633instruction is commonly used to represent automatic variables that must
3634have an address available. When the function returns (either with the <tt><a
3635 href="#i_ret">ret</a></tt> or <tt><a href="#i_unwind">unwind</a></tt>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3636instructions), the memory is reclaimed. Allocating zero bytes
3637is legal, but the result is undefined.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3638
3639<h5>Example:</h5>
3640
3641<pre>
Dan Gohmanf54f50a2009-01-04 23:49:44 +0000[diff] [blame]3642 %ptr = alloca i32 <i>; yields {i32*}:ptr</i>
3643 %ptr = alloca i32, i32 4 <i>; yields {i32*}:ptr</i>
3644 %ptr = alloca i32, i32 4, align 1024 <i>; yields {i32*}:ptr</i>
3645 %ptr = alloca i32, align 1024 <i>; yields {i32*}:ptr</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3646</pre>
3647</div>
3648
3649<!-- _______________________________________________________________________ -->
3650<div class="doc_subsubsection"> <a name="i_load">'<tt>load</tt>'
3651Instruction</a> </div>
3652<div class="doc_text">
3653<h5>Syntax:</h5>
3654<pre> &lt;result&gt; = load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]<br> &lt;result&gt; = volatile load &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;]<br></pre>
3655<h5>Overview:</h5>
3656<p>The '<tt>load</tt>' instruction is used to read from memory.</p>
3657<h5>Arguments:</h5>
3658<p>The argument to the '<tt>load</tt>' instruction specifies the memory
3659address from which to load. The pointer must point to a <a
3660 href="#t_firstclass">first class</a> type. If the <tt>load</tt> is
3661marked as <tt>volatile</tt>, then the optimizer is not allowed to modify
3662the number or order of execution of this <tt>load</tt> with other
3663volatile <tt>load</tt> and <tt><a href="#i_store">store</a></tt>
3664instructions. </p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3665<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3666The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3667(that is, the alignment of the memory address). A value of 0 or an
3668omitted "align" argument means that the operation has the preferential
3669alignment for the target. It is the responsibility of the code emitter
3670to ensure that the alignment information is correct. Overestimating
3671the alignment results in an undefined behavior. Underestimating the
3672alignment may produce less efficient code. An alignment of 1 is always
3673safe.
3674</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3675<h5>Semantics:</h5>
Duncan Sandsfe87fff2009-03-22 11:33:16 +0000[diff] [blame]3676<p>The location of memory pointed to is loaded. If the value being loaded
3677is of scalar type then the number of bytes read does not exceed the minimum
3678number of bytes needed to hold all bits of the type. For example, loading an
3679<tt>i24</tt> reads at most three bytes. When loading a value of a type like
3680<tt>i20</tt> with a size that is not an integral number of bytes, the result
3681is undefined if the value was not originally written using a store of the
3682same type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3683<h5>Examples:</h5>
3684<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
3685 <a
3686 href="#i_store">store</a> i32 3, i32* %ptr <i>; yields {void}</i>
3687 %val = load i32* %ptr <i>; yields {i32}:val = i32 3</i>
3688</pre>
3689</div>
3690<!-- _______________________________________________________________________ -->
3691<div class="doc_subsubsection"> <a name="i_store">'<tt>store</tt>'
3692Instruction</a> </div>
3693<div class="doc_text">
3694<h5>Syntax:</h5>
3695<pre> store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3696 volatile store &lt;ty&gt; &lt;value&gt;, &lt;ty&gt;* &lt;pointer&gt;[, align &lt;alignment&gt;] <i>; yields {void}</i>
3697</pre>
3698<h5>Overview:</h5>
3699<p>The '<tt>store</tt>' instruction is used to write to memory.</p>
3700<h5>Arguments:</h5>
3701<p>There are two arguments to the '<tt>store</tt>' instruction: a value
3702to store and an address at which to store it. The type of the '<tt>&lt;pointer&gt;</tt>'
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3703operand must be a pointer to the <a href="#t_firstclass">first class</a> type
3704of the '<tt>&lt;value&gt;</tt>'
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3705operand. If the <tt>store</tt> is marked as <tt>volatile</tt>, then the
3706optimizer is not allowed to modify the number or order of execution of
3707this <tt>store</tt> with other volatile <tt>load</tt> and <tt><a
3708 href="#i_store">store</a></tt> instructions.</p>
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3709<p>
Chris Lattner10368b62008-04-02 00:38:26 +0000[diff] [blame]3710The optional constant "align" argument specifies the alignment of the operation
Chris Lattner21003c82008-01-06 21:04:43 +0000[diff] [blame]3711(that is, the alignment of the memory address). A value of 0 or an
3712omitted "align" argument means that the operation has the preferential
3713alignment for the target. It is the responsibility of the code emitter
3714to ensure that the alignment information is correct. Overestimating
3715the alignment results in an undefined behavior. Underestimating the
3716alignment may produce less efficient code. An alignment of 1 is always
3717safe.
3718</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3719<h5>Semantics:</h5>
3720<p>The contents of memory are updated to contain '<tt>&lt;value&gt;</tt>'
Duncan Sandsfe87fff2009-03-22 11:33:16 +0000[diff] [blame]3721at the location specified by the '<tt>&lt;pointer&gt;</tt>' operand.
3722If '<tt>&lt;value&gt;</tt>' is of scalar type then the number of bytes
3723written does not exceed the minimum number of bytes needed to hold all
3724bits of the type. For example, storing an <tt>i24</tt> writes at most
3725three bytes. When writing a value of a type like <tt>i20</tt> with a
3726size that is not an integral number of bytes, it is unspecified what
3727happens to the extra bits that do not belong to the type, but they will
3728typically be overwritten.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3729<h5>Example:</h5>
3730<pre> %ptr = <a href="#i_alloca">alloca</a> i32 <i>; yields {i32*}:ptr</i>
Bill Wendling63ffa142007-10-22 05:10:05 +0000[diff] [blame]3731 store i32 3, i32* %ptr <i>; yields {void}</i>
3732 %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]3733</pre>
3734</div>
3735
3736<!-- _______________________________________________________________________ -->
3737<div class="doc_subsubsection">
3738 <a name="i_getelementptr">'<tt>getelementptr</tt>' Instruction</a>
3739</div>
3740
3741<div class="doc_text">
3742<h5>Syntax:</h5>
3743<pre>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3744 &lt;result&gt; = getelementptr &lt;pty&gt;* &lt;ptrval&gt;{, &lt;ty&gt; &lt;idx&gt;}*
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3745</pre>
3746
3747<h5>Overview:</h5>
3748
3749<p>
3750The '<tt>getelementptr</tt>' instruction is used to get the address of a
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3751subelement of an aggregate data structure. It performs address calculation only
3752and does not access memory.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3753
3754<h5>Arguments:</h5>
3755
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3756<p>The first argument is always a pointer, and forms the basis of the
3757calculation. The remaining arguments are indices, that indicate which of the
3758elements of the aggregate object are indexed. The interpretation of each index
3759is dependent on the type being indexed into. The first index always indexes the
3760pointer value given as the first argument, the second index indexes a value of
3761the type pointed to (not necessarily the value directly pointed to, since the
3762first index can be non-zero), etc. The first type indexed into must be a pointer
3763value, subsequent types can be arrays, vectors and structs. Note that subsequent
3764types being indexed into can never be pointers, since that would require loading
3765the pointer before continuing calculation.</p>
3766
3767<p>The type of each index argument depends on the type it is indexing into.
3768When indexing into a (packed) structure, only <tt>i32</tt> integer
3769<b>constants</b> are allowed. When indexing into an array, pointer or vector,
Sanjiv Guptaa2e72d92009-04-27 03:21:00 +0000[diff] [blame]3770integers of any width are allowed (also non-constants).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3771
3772<p>For example, let's consider a C code fragment and how it gets
3773compiled to LLVM:</p>
3774
3775<div class="doc_code">
3776<pre>
3777struct RT {
3778 char A;
3779 int B[10][20];
3780 char C;
3781};
3782struct ST {
3783 int X;
3784 double Y;
3785 struct RT Z;
3786};
3787
3788int *foo(struct ST *s) {
3789 return &amp;s[1].Z.B[5][13];
3790}
3791</pre>
3792</div>
3793
3794<p>The LLVM code generated by the GCC frontend is:</p>
3795
3796<div class="doc_code">
3797<pre>
Chris Lattner5b6dc6e2009-01-11 20:53:49 +0000[diff] [blame]3798%RT = <a href="#namedtypes">type</a> { i8 , [10 x [20 x i32]], i8 }
3799%ST = <a href="#namedtypes">type</a> { i32, double, %RT }
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3800
3801define i32* %foo(%ST* %s) {
3802entry:
3803 %reg = getelementptr %ST* %s, i32 1, i32 2, i32 1, i32 5, i32 13
3804 ret i32* %reg
3805}
3806</pre>
3807</div>
3808
3809<h5>Semantics:</h5>
3810
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3811<p>In the example above, the first index is indexing into the '<tt>%ST*</tt>'
3812type, which is a pointer, yielding a '<tt>%ST</tt>' = '<tt>{ i32, double, %RT
3813}</tt>' type, a structure. The second index indexes into the third element of
3814the structure, yielding a '<tt>%RT</tt>' = '<tt>{ i8 , [10 x [20 x i32]],
3815i8 }</tt>' type, another structure. The third index indexes into the second
3816element of the structure, yielding a '<tt>[10 x [20 x i32]]</tt>' type, an
3817array. The two dimensions of the array are subscripted into, yielding an
3818'<tt>i32</tt>' type. The '<tt>getelementptr</tt>' instruction returns a pointer
3819to this element, thus computing a value of '<tt>i32*</tt>' type.</p>
3820
3821<p>Note that it is perfectly legal to index partially through a
3822structure, returning a pointer to an inner element. Because of this,
3823the LLVM code for the given testcase is equivalent to:</p>
3824
3825<pre>
3826 define i32* %foo(%ST* %s) {
3827 %t1 = getelementptr %ST* %s, i32 1 <i>; yields %ST*:%t1</i>
3828 %t2 = getelementptr %ST* %t1, i32 0, i32 2 <i>; yields %RT*:%t2</i>
3829 %t3 = getelementptr %RT* %t2, i32 0, i32 1 <i>; yields [10 x [20 x i32]]*:%t3</i>
3830 %t4 = getelementptr [10 x [20 x i32]]* %t3, i32 0, i32 5 <i>; yields [20 x i32]*:%t4</i>
3831 %t5 = getelementptr [20 x i32]* %t4, i32 0, i32 13 <i>; yields i32*:%t5</i>
3832 ret i32* %t5
3833 }
3834</pre>
3835
Chris Lattner50609942009-03-09 20:55:18 +0000[diff] [blame]3836<p>Note that it is undefined to access an array out of bounds: array
3837and pointer indexes must always be within the defined bounds of the
3838array type when accessed with an instruction that dereferences the
3839pointer (e.g. a load or store instruction). The one exception for
3840this rule is zero length arrays. These arrays are defined to be
3841accessible as variable length arrays, which requires access beyond the
3842zero'th element.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3843
3844<p>The getelementptr instruction is often confusing. For some more insight
3845into how it works, see <a href="GetElementPtr.html">the getelementptr
3846FAQ</a>.</p>
3847
3848<h5>Example:</h5>
3849
3850<pre>
3851 <i>; yields [12 x i8]*:aptr</i>
Matthijs Kooijman32a080e2008-10-13 13:44:15 +0000[diff] [blame]3852 %aptr = getelementptr {i32, [12 x i8]}* %saptr, i64 0, i32 1
3853 <i>; yields i8*:vptr</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]3854 %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]3855 <i>; yields i8*:eptr</i>
3856 %eptr = getelementptr [12 x i8]* %aptr, i64 0, i32 1
Sanjiv Gupta4f9a0dc2009-04-25 07:27:44 +0000[diff] [blame]3857 <i>; yields i32*:iptr</i>
Sanjiv Gupta1e46c582009-04-24 16:38:13 +0000[diff] [blame]3858 %iptr = getelementptr [10 x i32]* @arr, i16 0, i16 0
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]3859</pre>
3860</div>
3861
3862<!-- ======================================================================= -->
3863<div class="doc_subsection"> <a name="convertops">Conversion Operations</a>
3864</div>
3865<div class="doc_text">
3866<p>The instructions in this category are the conversion instructions (casting)
3867which all take a single operand and a type. They perform various bit conversions
3868on the operand.</p>
3869</div>
3870
3871<!-- _______________________________________________________________________ -->
3872<div class="doc_subsubsection">
3873 <a name="i_trunc">'<tt>trunc .. to</tt>' Instruction</a>
3874</div>
3875<div class="doc_text">
3876
3877<h5>Syntax:</h5>
3878<pre>
3879 &lt;result&gt; = trunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3880</pre>
3881
3882<h5>Overview:</h5>
3883<p>
3884The '<tt>trunc</tt>' instruction truncates its operand to the type <tt>ty2</tt>.
3885</p>
3886
3887<h5>Arguments:</h5>
3888<p>
3889The '<tt>trunc</tt>' instruction takes a <tt>value</tt> to trunc, which must
3890be an <a href="#t_integer">integer</a> type, and a type that specifies the size
3891and type of the result, which must be an <a href="#t_integer">integer</a>
3892type. The bit size of <tt>value</tt> must be larger than the bit size of
3893<tt>ty2</tt>. Equal sized types are not allowed.</p>
3894
3895<h5>Semantics:</h5>
3896<p>
3897The '<tt>trunc</tt>' instruction truncates the high order bits in <tt>value</tt>
3898and converts the remaining bits to <tt>ty2</tt>. Since the source size must be
3899larger than the destination size, <tt>trunc</tt> cannot be a <i>no-op cast</i>.
3900It will always truncate bits.</p>
3901
3902<h5>Example:</h5>
3903<pre>
3904 %X = trunc i32 257 to i8 <i>; yields i8:1</i>
3905 %Y = trunc i32 123 to i1 <i>; yields i1:true</i>
3906 %Y = trunc i32 122 to i1 <i>; yields i1:false</i>
3907</pre>
3908</div>
3909
3910<!-- _______________________________________________________________________ -->
3911<div class="doc_subsubsection">
3912 <a name="i_zext">'<tt>zext .. to</tt>' Instruction</a>
3913</div>
3914<div class="doc_text">
3915
3916<h5>Syntax:</h5>
3917<pre>
3918 &lt;result&gt; = zext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3919</pre>
3920
3921<h5>Overview:</h5>
3922<p>The '<tt>zext</tt>' instruction zero extends its operand to type
3923<tt>ty2</tt>.</p>
3924
3925
3926<h5>Arguments:</h5>
3927<p>The '<tt>zext</tt>' instruction takes a value to cast, which must be of
3928<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3929also be of <a href="#t_integer">integer</a> type. The bit size of the
3930<tt>value</tt> must be smaller than the bit size of the destination type,
3931<tt>ty2</tt>.</p>
3932
3933<h5>Semantics:</h5>
3934<p>The <tt>zext</tt> fills the high order bits of the <tt>value</tt> with zero
3935bits until it reaches the size of the destination type, <tt>ty2</tt>.</p>
3936
3937<p>When zero extending from i1, the result will always be either 0 or 1.</p>
3938
3939<h5>Example:</h5>
3940<pre>
3941 %X = zext i32 257 to i64 <i>; yields i64:257</i>
3942 %Y = zext i1 true to i32 <i>; yields i32:1</i>
3943</pre>
3944</div>
3945
3946<!-- _______________________________________________________________________ -->
3947<div class="doc_subsubsection">
3948 <a name="i_sext">'<tt>sext .. to</tt>' Instruction</a>
3949</div>
3950<div class="doc_text">
3951
3952<h5>Syntax:</h5>
3953<pre>
3954 &lt;result&gt; = sext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3955</pre>
3956
3957<h5>Overview:</h5>
3958<p>The '<tt>sext</tt>' sign extends <tt>value</tt> to the type <tt>ty2</tt>.</p>
3959
3960<h5>Arguments:</h5>
3961<p>
3962The '<tt>sext</tt>' instruction takes a value to cast, which must be of
3963<a href="#t_integer">integer</a> type, and a type to cast it to, which must
3964also be of <a href="#t_integer">integer</a> type. The bit size of the
3965<tt>value</tt> must be smaller than the bit size of the destination type,
3966<tt>ty2</tt>.</p>
3967
3968<h5>Semantics:</h5>
3969<p>
3970The '<tt>sext</tt>' instruction performs a sign extension by copying the sign
3971bit (highest order bit) of the <tt>value</tt> until it reaches the bit size of
3972the type <tt>ty2</tt>.</p>
3973
3974<p>When sign extending from i1, the extension always results in -1 or 0.</p>
3975
3976<h5>Example:</h5>
3977<pre>
3978 %X = sext i8 -1 to i16 <i>; yields i16 :65535</i>
3979 %Y = sext i1 true to i32 <i>; yields i32:-1</i>
3980</pre>
3981</div>
3982
3983<!-- _______________________________________________________________________ -->
3984<div class="doc_subsubsection">
3985 <a name="i_fptrunc">'<tt>fptrunc .. to</tt>' Instruction</a>
3986</div>
3987
3988<div class="doc_text">
3989
3990<h5>Syntax:</h5>
3991
3992<pre>
3993 &lt;result&gt; = fptrunc &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
3994</pre>
3995
3996<h5>Overview:</h5>
3997<p>The '<tt>fptrunc</tt>' instruction truncates <tt>value</tt> to type
3998<tt>ty2</tt>.</p>
3999
4000
4001<h5>Arguments:</h5>
4002<p>The '<tt>fptrunc</tt>' instruction takes a <a href="#t_floating">floating
4003 point</a> value to cast and a <a href="#t_floating">floating point</a> type to
4004cast it to. The size of <tt>value</tt> must be larger than the size of
4005<tt>ty2</tt>. This implies that <tt>fptrunc</tt> cannot be used to make a
4006<i>no-op cast</i>.</p>
4007
4008<h5>Semantics:</h5>
4009<p> The '<tt>fptrunc</tt>' instruction truncates a <tt>value</tt> from a larger
4010<a href="#t_floating">floating point</a> type to a smaller
4011<a href="#t_floating">floating point</a> type. If the value cannot fit within
4012the destination type, <tt>ty2</tt>, then the results are undefined.</p>
4013
4014<h5>Example:</h5>
4015<pre>
4016 %X = fptrunc double 123.0 to float <i>; yields float:123.0</i>
4017 %Y = fptrunc double 1.0E+300 to float <i>; yields undefined</i>
4018</pre>
4019</div>
4020
4021<!-- _______________________________________________________________________ -->
4022<div class="doc_subsubsection">
4023 <a name="i_fpext">'<tt>fpext .. to</tt>' Instruction</a>
4024</div>
4025<div class="doc_text">
4026
4027<h5>Syntax:</h5>
4028<pre>
4029 &lt;result&gt; = fpext &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4030</pre>
4031
4032<h5>Overview:</h5>
4033<p>The '<tt>fpext</tt>' extends a floating point <tt>value</tt> to a larger
4034floating point value.</p>
4035
4036<h5>Arguments:</h5>
4037<p>The '<tt>fpext</tt>' instruction takes a
4038<a href="#t_floating">floating point</a> <tt>value</tt> to cast,
4039and a <a href="#t_floating">floating point</a> type to cast it to. The source
4040type must be smaller than the destination type.</p>
4041
4042<h5>Semantics:</h5>
4043<p>The '<tt>fpext</tt>' instruction extends the <tt>value</tt> from a smaller
4044<a href="#t_floating">floating point</a> type to a larger
4045<a href="#t_floating">floating point</a> type. The <tt>fpext</tt> cannot be
4046used to make a <i>no-op cast</i> because it always changes bits. Use
4047<tt>bitcast</tt> to make a <i>no-op cast</i> for a floating point cast.</p>
4048
4049<h5>Example:</h5>
4050<pre>
4051 %X = fpext float 3.1415 to double <i>; yields double:3.1415</i>
4052 %Y = fpext float 1.0 to float <i>; yields float:1.0 (no-op)</i>
4053</pre>
4054</div>
4055
4056<!-- _______________________________________________________________________ -->
4057<div class="doc_subsubsection">
4058 <a name="i_fptoui">'<tt>fptoui .. to</tt>' Instruction</a>
4059</div>
4060<div class="doc_text">
4061
4062<h5>Syntax:</h5>
4063<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4064 &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]4065</pre>
4066
4067<h5>Overview:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4068<p>The '<tt>fptoui</tt>' converts a floating point <tt>value</tt> to its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4069unsigned integer equivalent of type <tt>ty2</tt>.
4070</p>
4071
4072<h5>Arguments:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4073<p>The '<tt>fptoui</tt>' instruction takes a value to cast, which must be a
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4074scalar or vector <a href="#t_floating">floating point</a> value, and a type
4075to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4076type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4077vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4078
4079<h5>Semantics:</h5>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4080<p> The '<tt>fptoui</tt>' instruction converts its
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4081<a href="#t_floating">floating point</a> operand into the nearest (rounding
4082towards zero) unsigned integer value. If the value cannot fit in <tt>ty2</tt>,
4083the results are undefined.</p>
4084
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4085<h5>Example:</h5>
4086<pre>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4087 %X = fptoui double 123.0 to i32 <i>; yields i32:123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4088 %Y = fptoui float 1.0E+300 to i1 <i>; yields undefined:1</i>
Reid Spencere6adee82007-07-31 14:40:14 +0000[diff] [blame]4089 %X = fptoui float 1.04E+17 to i8 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4090</pre>
4091</div>
4092
4093<!-- _______________________________________________________________________ -->
4094<div class="doc_subsubsection">
4095 <a name="i_fptosi">'<tt>fptosi .. to</tt>' Instruction</a>
4096</div>
4097<div class="doc_text">
4098
4099<h5>Syntax:</h5>
4100<pre>
4101 &lt;result&gt; = fptosi &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4102</pre>
4103
4104<h5>Overview:</h5>
4105<p>The '<tt>fptosi</tt>' instruction converts
4106<a href="#t_floating">floating point</a> <tt>value</tt> to type <tt>ty2</tt>.
4107</p>
4108
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4109<h5>Arguments:</h5>
4110<p> The '<tt>fptosi</tt>' instruction takes a value to cast, which must be a
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4111scalar or vector <a href="#t_floating">floating point</a> value, and a type
4112to cast it to <tt>ty2</tt>, which must be an <a href="#t_integer">integer</a>
4113type. If <tt>ty</tt> is a vector floating point type, <tt>ty2</tt> must be a
4114vector integer type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4115
4116<h5>Semantics:</h5>
4117<p>The '<tt>fptosi</tt>' instruction converts its
4118<a href="#t_floating">floating point</a> operand into the nearest (rounding
4119towards zero) signed integer value. If the value cannot fit in <tt>ty2</tt>,
4120the results are undefined.</p>
4121
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4122<h5>Example:</h5>
4123<pre>
4124 %X = fptosi double -123.0 to i32 <i>; yields i32:-123</i>
Chris Lattner681f1e82007-09-22 03:17:52 +0000[diff] [blame]4125 %Y = fptosi float 1.0E-247 to i1 <i>; yields undefined:1</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4126 %X = fptosi float 1.04E+17 to i8 <i>; yields undefined:1</i>
4127</pre>
4128</div>
4129
4130<!-- _______________________________________________________________________ -->
4131<div class="doc_subsubsection">
4132 <a name="i_uitofp">'<tt>uitofp .. to</tt>' Instruction</a>
4133</div>
4134<div class="doc_text">
4135
4136<h5>Syntax:</h5>
4137<pre>
4138 &lt;result&gt; = uitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4139</pre>
4140
4141<h5>Overview:</h5>
4142<p>The '<tt>uitofp</tt>' instruction regards <tt>value</tt> as an unsigned
4143integer and converts that value to the <tt>ty2</tt> type.</p>
4144
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4145<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4146<p>The '<tt>uitofp</tt>' instruction takes a value to cast, which must be a
4147scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
4148to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4149type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4150floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4151
4152<h5>Semantics:</h5>
4153<p>The '<tt>uitofp</tt>' instruction interprets its operand as an unsigned
4154integer quantity and converts it to the corresponding floating point value. If
4155the value cannot fit in the floating point value, the results are undefined.</p>
4156
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4157<h5>Example:</h5>
4158<pre>
4159 %X = uitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4160 %Y = uitofp i8 -1 to double <i>; yields double:255.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4161</pre>
4162</div>
4163
4164<!-- _______________________________________________________________________ -->
4165<div class="doc_subsubsection">
4166 <a name="i_sitofp">'<tt>sitofp .. to</tt>' Instruction</a>
4167</div>
4168<div class="doc_text">
4169
4170<h5>Syntax:</h5>
4171<pre>
4172 &lt;result&gt; = sitofp &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4173</pre>
4174
4175<h5>Overview:</h5>
4176<p>The '<tt>sitofp</tt>' instruction regards <tt>value</tt> as a signed
4177integer and converts that value to the <tt>ty2</tt> type.</p>
4178
4179<h5>Arguments:</h5>
Nate Begeman78246ca2007-11-17 03:58:34 +0000[diff] [blame]4180<p>The '<tt>sitofp</tt>' instruction takes a value to cast, which must be a
4181scalar or vector <a href="#t_integer">integer</a> value, and a type to cast it
4182to <tt>ty2</tt>, which must be an <a href="#t_floating">floating point</a>
4183type. If <tt>ty</tt> is a vector integer type, <tt>ty2</tt> must be a vector
4184floating point type with the same number of elements as <tt>ty</tt></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4185
4186<h5>Semantics:</h5>
4187<p>The '<tt>sitofp</tt>' instruction interprets its operand as a signed
4188integer quantity and converts it to the corresponding floating point value. If
4189the value cannot fit in the floating point value, the results are undefined.</p>
4190
4191<h5>Example:</h5>
4192<pre>
4193 %X = sitofp i32 257 to float <i>; yields float:257.0</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4194 %Y = sitofp i8 -1 to double <i>; yields double:-1.0</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4195</pre>
4196</div>
4197
4198<!-- _______________________________________________________________________ -->
4199<div class="doc_subsubsection">
4200 <a name="i_ptrtoint">'<tt>ptrtoint .. to</tt>' Instruction</a>
4201</div>
4202<div class="doc_text">
4203
4204<h5>Syntax:</h5>
4205<pre>
4206 &lt;result&gt; = ptrtoint &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4207</pre>
4208
4209<h5>Overview:</h5>
4210<p>The '<tt>ptrtoint</tt>' instruction converts the pointer <tt>value</tt> to
4211the integer type <tt>ty2</tt>.</p>
4212
4213<h5>Arguments:</h5>
4214<p>The '<tt>ptrtoint</tt>' instruction takes a <tt>value</tt> to cast, which
4215must be a <a href="#t_pointer">pointer</a> value, and a type to cast it to
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4216<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]4217
4218<h5>Semantics:</h5>
4219<p>The '<tt>ptrtoint</tt>' instruction converts <tt>value</tt> to integer type
4220<tt>ty2</tt> by interpreting the pointer value as an integer and either
4221truncating or zero extending that value to the size of the integer type. If
4222<tt>value</tt> is smaller than <tt>ty2</tt> then a zero extension is done. If
4223<tt>value</tt> is larger than <tt>ty2</tt> then a truncation is done. If they
4224are the same size, then nothing is done (<i>no-op cast</i>) other than a type
4225change.</p>
4226
4227<h5>Example:</h5>
4228<pre>
4229 %X = ptrtoint i32* %X to i8 <i>; yields truncation on 32-bit architecture</i>
4230 %Y = ptrtoint i32* %x to i64 <i>; yields zero extension on 32-bit architecture</i>
4231</pre>
4232</div>
4233
4234<!-- _______________________________________________________________________ -->
4235<div class="doc_subsubsection">
4236 <a name="i_inttoptr">'<tt>inttoptr .. to</tt>' Instruction</a>
4237</div>
4238<div class="doc_text">
4239
4240<h5>Syntax:</h5>
4241<pre>
4242 &lt;result&gt; = inttoptr &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4243</pre>
4244
4245<h5>Overview:</h5>
4246<p>The '<tt>inttoptr</tt>' instruction converts an integer <tt>value</tt> to
4247a pointer type, <tt>ty2</tt>.</p>
4248
4249<h5>Arguments:</h5>
4250<p>The '<tt>inttoptr</tt>' instruction takes an <a href="#t_integer">integer</a>
4251value to cast, and a type to cast it to, which must be a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4252<a href="#t_pointer">pointer</a> type.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4253
4254<h5>Semantics:</h5>
4255<p>The '<tt>inttoptr</tt>' instruction converts <tt>value</tt> to type
4256<tt>ty2</tt> by applying either a zero extension or a truncation depending on
4257the size of the integer <tt>value</tt>. If <tt>value</tt> is larger than the
4258size of a pointer then a truncation is done. If <tt>value</tt> is smaller than
4259the size of a pointer then a zero extension is done. If they are the same size,
4260nothing is done (<i>no-op cast</i>).</p>
4261
4262<h5>Example:</h5>
4263<pre>
4264 %X = inttoptr i32 255 to i32* <i>; yields zero extension on 64-bit architecture</i>
4265 %X = inttoptr i32 255 to i32* <i>; yields no-op on 32-bit architecture</i>
4266 %Y = inttoptr i64 0 to i32* <i>; yields truncation on 32-bit architecture</i>
4267</pre>
4268</div>
4269
4270<!-- _______________________________________________________________________ -->
4271<div class="doc_subsubsection">
4272 <a name="i_bitcast">'<tt>bitcast .. to</tt>' Instruction</a>
4273</div>
4274<div class="doc_text">
4275
4276<h5>Syntax:</h5>
4277<pre>
4278 &lt;result&gt; = bitcast &lt;ty&gt; &lt;value&gt; to &lt;ty2&gt; <i>; yields ty2</i>
4279</pre>
4280
4281<h5>Overview:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4282
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4283<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4284<tt>ty2</tt> without changing any bits.</p>
4285
4286<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4287
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4288<p>The '<tt>bitcast</tt>' instruction takes a value to cast, which must be
Dan Gohman7305fa02008-09-08 16:45:59 +0000[diff] [blame]4289a non-aggregate first class value, and a type to cast it to, which must also be
4290a non-aggregate <a href="#t_firstclass">first class</a> type. The bit sizes of
4291<tt>value</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4292and the destination type, <tt>ty2</tt>, must be identical. If the source
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4293type is a pointer, the destination type must also be a pointer. This
4294instruction supports bitwise conversion of vectors to integers and to vectors
4295of other types (as long as they have the same size).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4296
4297<h5>Semantics:</h5>
4298<p>The '<tt>bitcast</tt>' instruction converts <tt>value</tt> to type
4299<tt>ty2</tt>. It is always a <i>no-op cast</i> because no bits change with
4300this conversion. The conversion is done as if the <tt>value</tt> had been
4301stored to memory and read back as type <tt>ty2</tt>. Pointer types may only be
4302converted to other pointer types with this instruction. To convert pointers to
4303other types, use the <a href="#i_inttoptr">inttoptr</a> or
4304<a href="#i_ptrtoint">ptrtoint</a> instructions first.</p>
4305
4306<h5>Example:</h5>
4307<pre>
4308 %X = bitcast i8 255 to i8 <i>; yields i8 :-1</i>
4309 %Y = bitcast i32* %x to sint* <i>; yields sint*:%x</i>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4310 %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]4311</pre>
4312</div>
4313
4314<!-- ======================================================================= -->
4315<div class="doc_subsection"> <a name="otherops">Other Operations</a> </div>
4316<div class="doc_text">
4317<p>The instructions in this category are the "miscellaneous"
4318instructions, which defy better classification.</p>
4319</div>
4320
4321<!-- _______________________________________________________________________ -->
4322<div class="doc_subsubsection"><a name="i_icmp">'<tt>icmp</tt>' Instruction</a>
4323</div>
4324<div class="doc_text">
4325<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4326<pre> &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]4327</pre>
4328<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4329<p>The '<tt>icmp</tt>' instruction returns a boolean value or
4330a vector of boolean values based on comparison
4331of its two integer, integer vector, or pointer operands.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4332<h5>Arguments:</h5>
4333<p>The '<tt>icmp</tt>' instruction takes three operands. The first operand is
4334the condition code indicating the kind of comparison to perform. It is not
4335a value, just a keyword. The possible condition code are:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4336</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4337<ol>
4338 <li><tt>eq</tt>: equal</li>
4339 <li><tt>ne</tt>: not equal </li>
4340 <li><tt>ugt</tt>: unsigned greater than</li>
4341 <li><tt>uge</tt>: unsigned greater or equal</li>
4342 <li><tt>ult</tt>: unsigned less than</li>
4343 <li><tt>ule</tt>: unsigned less or equal</li>
4344 <li><tt>sgt</tt>: signed greater than</li>
4345 <li><tt>sge</tt>: signed greater or equal</li>
4346 <li><tt>slt</tt>: signed less than</li>
4347 <li><tt>sle</tt>: signed less or equal</li>
4348</ol>
4349<p>The remaining two arguments must be <a href="#t_integer">integer</a> or
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4350<a href="#t_pointer">pointer</a>
4351or integer <a href="#t_vector">vector</a> typed.
4352They must also be identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4353<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4354<p>The '<tt>icmp</tt>' compares <tt>op1</tt> and <tt>op2</tt> according to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4355the condition code given as <tt>cond</tt>. The comparison performed always
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4356yields either an <a href="#t_primitive"><tt>i1</tt></a> or vector of <tt>i1</tt> result, as follows:
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4357</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4358<ol>
4359 <li><tt>eq</tt>: yields <tt>true</tt> if the operands are equal,
4360 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.
4361 </li>
4362 <li><tt>ne</tt>: yields <tt>true</tt> if the operands are unequal,
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4363 <tt>false</tt> otherwise. No sign interpretation is necessary or performed.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4364 <li><tt>ugt</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4365 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4366 <li><tt>uge</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4367 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4368 <li><tt>ult</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4369 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4370 <li><tt>ule</tt>: interprets the operands as unsigned values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4371 <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]4372 <li><tt>sgt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4373 <tt>true</tt> if <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4374 <li><tt>sge</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4375 <tt>true</tt> if <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4376 <li><tt>slt</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4377 <tt>true</tt> if <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4378 <li><tt>sle</tt>: interprets the operands as signed values and yields
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4379 <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]4380</ol>
4381<p>If the operands are <a href="#t_pointer">pointer</a> typed, the pointer
4382values are compared as if they were integers.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4383<p>If the operands are integer vectors, then they are compared
4384element by element. The result is an <tt>i1</tt> vector with
4385the same number of elements as the values being compared.
4386Otherwise, the result is an <tt>i1</tt>.
4387</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4388
4389<h5>Example:</h5>
4390<pre> &lt;result&gt; = icmp eq i32 4, 5 <i>; yields: result=false</i>
4391 &lt;result&gt; = icmp ne float* %X, %X <i>; yields: result=false</i>
4392 &lt;result&gt; = icmp ult i16 4, 5 <i>; yields: result=true</i>
4393 &lt;result&gt; = icmp sgt i16 4, 5 <i>; yields: result=false</i>
4394 &lt;result&gt; = icmp ule i16 -4, 5 <i>; yields: result=false</i>
4395 &lt;result&gt; = icmp sge i16 4, 5 <i>; yields: result=false</i>
4396</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4397
4398<p>Note that the code generator does not yet support vector types with
4399 the <tt>icmp</tt> instruction.</p>
4400
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4401</div>
4402
4403<!-- _______________________________________________________________________ -->
4404<div class="doc_subsubsection"><a name="i_fcmp">'<tt>fcmp</tt>' Instruction</a>
4405</div>
4406<div class="doc_text">
4407<h5>Syntax:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4408<pre> &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]4409</pre>
4410<h5>Overview:</h5>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4411<p>The '<tt>fcmp</tt>' instruction returns a boolean value
4412or vector of boolean values based on comparison
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4413of its operands.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4414<p>
4415If the operands are floating point scalars, then the result
4416type is a boolean (<a href="#t_primitive"><tt>i1</tt></a>).
4417</p>
4418<p>If the operands are floating point vectors, then the result type
4419is a vector of boolean with the same number of elements as the
4420operands being compared.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4421<h5>Arguments:</h5>
4422<p>The '<tt>fcmp</tt>' instruction takes three operands. The first operand is
4423the condition code indicating the kind of comparison to perform. It is not
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4424a value, just a keyword. The possible condition code are:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4425<ol>
4426 <li><tt>false</tt>: no comparison, always returns false</li>
4427 <li><tt>oeq</tt>: ordered and equal</li>
4428 <li><tt>ogt</tt>: ordered and greater than </li>
4429 <li><tt>oge</tt>: ordered and greater than or equal</li>
4430 <li><tt>olt</tt>: ordered and less than </li>
4431 <li><tt>ole</tt>: ordered and less than or equal</li>
4432 <li><tt>one</tt>: ordered and not equal</li>
4433 <li><tt>ord</tt>: ordered (no nans)</li>
4434 <li><tt>ueq</tt>: unordered or equal</li>
4435 <li><tt>ugt</tt>: unordered or greater than </li>
4436 <li><tt>uge</tt>: unordered or greater than or equal</li>
4437 <li><tt>ult</tt>: unordered or less than </li>
4438 <li><tt>ule</tt>: unordered or less than or equal</li>
4439 <li><tt>une</tt>: unordered or not equal</li>
4440 <li><tt>uno</tt>: unordered (either nans)</li>
4441 <li><tt>true</tt>: no comparison, always returns true</li>
4442</ol>
4443<p><i>Ordered</i> means that neither operand is a QNAN while
4444<i>unordered</i> means that either operand may be a QNAN.</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4445<p>Each of <tt>val1</tt> and <tt>val2</tt> arguments must be
4446either a <a href="#t_floating">floating point</a> type
4447or a <a href="#t_vector">vector</a> of floating point type.
4448They must have identical types.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4449<h5>Semantics:</h5>
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4450<p>The '<tt>fcmp</tt>' instruction compares <tt>op1</tt> and <tt>op2</tt>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4451according to the condition code given as <tt>cond</tt>.
4452If the operands are vectors, then the vectors are compared
4453element by element.
4454Each comparison performed
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4455always yields an <a href="#t_primitive">i1</a> result, as follows:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4456<ol>
4457 <li><tt>false</tt>: always yields <tt>false</tt>, regardless of operands.</li>
4458 <li><tt>oeq</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4459 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4460 <li><tt>ogt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4461 <tt>op1</tt> is greather than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4462 <li><tt>oge</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4463 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4464 <li><tt>olt</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4465 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4466 <li><tt>ole</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4467 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4468 <li><tt>one</tt>: yields <tt>true</tt> if both operands are not a QNAN and
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4469 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4470 <li><tt>ord</tt>: yields <tt>true</tt> if both operands are not a QNAN.</li>
4471 <li><tt>ueq</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4472 <tt>op1</tt> is equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4473 <li><tt>ugt</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4474 <tt>op1</tt> is greater than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4475 <li><tt>uge</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4476 <tt>op1</tt> is greater than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4477 <li><tt>ult</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4478 <tt>op1</tt> is less than <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4479 <li><tt>ule</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4480 <tt>op1</tt> is less than or equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4481 <li><tt>une</tt>: yields <tt>true</tt> if either operand is a QNAN or
Gabor Greifd9068fe2008-08-07 21:46:00 +0000[diff] [blame]4482 <tt>op1</tt> is not equal to <tt>op2</tt>.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4483 <li><tt>uno</tt>: yields <tt>true</tt> if either operand is a QNAN.</li>
4484 <li><tt>true</tt>: always yields <tt>true</tt>, regardless of operands.</li>
4485</ol>
4486
4487<h5>Example:</h5>
4488<pre> &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]4489 &lt;result&gt; = fcmp one float 4.0, 5.0 <i>; yields: result=true</i>
4490 &lt;result&gt; = fcmp olt float 4.0, 5.0 <i>; yields: result=true</i>
4491 &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]4492</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4493
4494<p>Note that the code generator does not yet support vector types with
4495 the <tt>fcmp</tt> instruction.</p>
4496
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4497</div>
4498
4499<!-- _______________________________________________________________________ -->
Nate Begeman646fa482008-05-12 19:01:56 +0000[diff] [blame]4500<div class="doc_subsubsection">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4501 <a name="i_phi">'<tt>phi</tt>' Instruction</a>
4502</div>
4503
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4504<div class="doc_text">
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4505
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4506<h5>Syntax:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4507
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4508<pre> &lt;result&gt; = phi &lt;ty&gt; [ &lt;val0&gt;, &lt;label0&gt;], ...<br></pre>
4509<h5>Overview:</h5>
4510<p>The '<tt>phi</tt>' instruction is used to implement the &#966; node in
4511the SSA graph representing the function.</p>
4512<h5>Arguments:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4513
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4514<p>The type of the incoming values is specified with the first type
4515field. After this, the '<tt>phi</tt>' instruction takes a list of pairs
4516as arguments, with one pair for each predecessor basic block of the
4517current block. Only values of <a href="#t_firstclass">first class</a>
4518type may be used as the value arguments to the PHI node. Only labels
4519may be used as the label arguments.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4520
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4521<p>There must be no non-phi instructions between the start of a basic
4522block and the PHI instructions: i.e. PHI instructions must be first in
4523a basic block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4524
Jay Foad8e2fd2c2009-06-03 10:20:10 +0000[diff] [blame]4525<p>For the purposes of the SSA form, the use of each incoming value is
4526deemed to occur on the edge from the corresponding predecessor block
4527to the current block (but after any definition of an '<tt>invoke</tt>'
4528instruction's return value on the same edge).</p>
4529
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4530<h5>Semantics:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4531
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4532<p>At runtime, the '<tt>phi</tt>' instruction logically takes on the value
4533specified by the pair corresponding to the predecessor basic block that executed
4534just prior to the current block.</p>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4535
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4536<h5>Example:</h5>
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4537<pre>
4538Loop: ; Infinite loop that counts from 0 on up...
4539 %indvar = phi i32 [ 0, %LoopHeader ], [ %nextindvar, %Loop ]
4540 %nextindvar = add i32 %indvar, 1
4541 br label %Loop
4542</pre>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4543</div>
4544
4545<!-- _______________________________________________________________________ -->
4546<div class="doc_subsubsection">
4547 <a name="i_select">'<tt>select</tt>' Instruction</a>
4548</div>
4549
4550<div class="doc_text">
4551
4552<h5>Syntax:</h5>
4553
4554<pre>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4555 &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>
4556
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4557 <i>selty</i> is either i1 or {&lt;N x i1&gt;}
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4558</pre>
4559
4560<h5>Overview:</h5>
4561
4562<p>
4563The '<tt>select</tt>' instruction is used to choose one value based on a
4564condition, without branching.
4565</p>
4566
4567
4568<h5>Arguments:</h5>
4569
4570<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4571The '<tt>select</tt>' instruction requires an 'i1' value or
4572a vector of 'i1' values indicating the
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4573condition, and two values of the same <a href="#t_firstclass">first class</a>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4574type. If the val1/val2 are vectors and
4575the condition is a scalar, then entire vectors are selected, not
Chris Lattner6704c212008-05-20 20:48:21 +0000[diff] [blame]4576individual elements.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4577</p>
4578
4579<h5>Semantics:</h5>
4580
4581<p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4582If the condition is an i1 and it evaluates to 1, the instruction returns the first
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4583value argument; otherwise, it returns the second value argument.
4584</p>
Dan Gohmanb60ca3c2008-09-09 01:02:47 +0000[diff] [blame]4585<p>
4586If the condition is a vector of i1, then the value arguments must
4587be vectors of the same size, and the selection is done element
4588by element.
4589</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4590
4591<h5>Example:</h5>
4592
4593<pre>
4594 %X = select i1 true, i8 17, i8 42 <i>; yields i8:17</i>
4595</pre>
Dan Gohmana53eb382009-01-22 01:39:38 +0000[diff] [blame]4596
4597<p>Note that the code generator does not yet support conditions
4598 with vector type.</p>
4599
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4600</div>
4601
4602
4603<!-- _______________________________________________________________________ -->
4604<div class="doc_subsubsection">
4605 <a name="i_call">'<tt>call</tt>' Instruction</a>
4606</div>
4607
4608<div class="doc_text">
4609
4610<h5>Syntax:</h5>
4611<pre>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4612 &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]4613</pre>
4614
4615<h5>Overview:</h5>
4616
4617<p>The '<tt>call</tt>' instruction represents a simple function call.</p>
4618
4619<h5>Arguments:</h5>
4620
4621<p>This instruction requires several arguments:</p>
4622
4623<ol>
4624 <li>
4625 <p>The optional "tail" marker indicates whether the callee function accesses
4626 any allocas or varargs in the caller. If the "tail" marker is present, the
4627 function call is eligible for tail call optimization. Note that calls may
4628 be marked "tail" even if they do not occur before a <a
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4629 href="#i_ret"><tt>ret</tt></a> instruction.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4630 </li>
4631 <li>
4632 <p>The optional "cconv" marker indicates which <a href="#callingconv">calling
4633 convention</a> the call should use. If none is specified, the call defaults
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4634 to using C calling conventions.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4635 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4636
4637 <li>
4638 <p>The optional <a href="#paramattrs">Parameter Attributes</a> list for
4639 return values. Only '<tt>zeroext</tt>', '<tt>signext</tt>',
4640 and '<tt>inreg</tt>' attributes are valid here.</p>
4641 </li>
4642
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4643 <li>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4644 <p>'<tt>ty</tt>': the type of the call instruction itself which is also
4645 the type of the return value. Functions that return no value are marked
4646 <tt><a href="#t_void">void</a></tt>.</p>
4647 </li>
4648 <li>
4649 <p>'<tt>fnty</tt>': shall be the signature of the pointer to function
4650 value being invoked. The argument types must match the types implied by
4651 this signature. This type can be omitted if the function is not varargs
4652 and if the function type does not return a pointer to a function.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4653 </li>
4654 <li>
4655 <p>'<tt>fnptrval</tt>': An LLVM value containing a pointer to a function to
4656 be invoked. In most cases, this is a direct function invocation, but
4657 indirect <tt>call</tt>s are just as possible, calling an arbitrary pointer
4658 to function value.</p>
4659 </li>
4660 <li>
4661 <p>'<tt>function args</tt>': argument list whose types match the
4662 function signature argument types. All arguments must be of
4663 <a href="#t_firstclass">first class</a> type. If the function signature
4664 indicates the function accepts a variable number of arguments, the extra
4665 arguments can be specified.</p>
4666 </li>
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4667 <li>
Devang Pateld0bfcc72008-10-07 17:48:33 +0000[diff] [blame]4668 <p>The optional <a href="#fnattrs">function attributes</a> list. Only
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4669 '<tt>noreturn</tt>', '<tt>nounwind</tt>', '<tt>readonly</tt>' and
4670 '<tt>readnone</tt>' attributes are valid here.</p>
4671 </li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4672</ol>
4673
4674<h5>Semantics:</h5>
4675
4676<p>The '<tt>call</tt>' instruction is used to cause control flow to
4677transfer to a specified function, with its incoming arguments bound to
4678the specified values. Upon a '<tt><a href="#i_ret">ret</a></tt>'
4679instruction in the called function, control flow continues with the
4680instruction after the function call, and the return value of the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4681function is bound to the result argument.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4682
4683<h5>Example:</h5>
4684
4685<pre>
Nick Lewycky93082fc2007-09-08 13:57:50 +0000[diff] [blame]4686 %retval = call i32 @test(i32 %argc)
Chris Lattner5e893ef2008-03-21 17:24:17 +0000[diff] [blame]4687 call i32 (i8 *, ...)* @printf(i8 * %msg, i32 12, i8 42) <i>; yields i32</i>
4688 %X = tail call i32 @foo() <i>; yields i32</i>
4689 %Y = tail call <a href="#callingconv">fastcc</a> i32 @foo() <i>; yields i32</i>
4690 call void %foo(i8 97 signext)
Devang Patela3cc5372008-03-10 20:49:15 +0000[diff] [blame]4691
4692 %struct.A = type { i32, i8 }
Devang Patelac2fc272008-10-06 18:50:38 +0000[diff] [blame]4693 %r = call %struct.A @foo() <i>; yields { 32, i8 }</i>
Dan Gohman3e700032008-10-04 19:00:07 +0000[diff] [blame]4694 %gr = extractvalue %struct.A %r, 0 <i>; yields i32</i>
4695 %gr1 = extractvalue %struct.A %r, 1 <i>; yields i8</i>
Chris Lattnerac454b32008-10-08 06:26:11 +0000[diff] [blame]4696 %Z = call void @foo() noreturn <i>; indicates that %foo never returns normally</i>
Matthijs Kooijman2c4e05a2008-10-07 10:03:45 +0000[diff] [blame]4697 %ZZ = call zeroext i32 @bar() <i>; Return value is %zero extended</i>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4698</pre>
4699
4700</div>
4701
4702<!-- _______________________________________________________________________ -->
4703<div class="doc_subsubsection">
4704 <a name="i_va_arg">'<tt>va_arg</tt>' Instruction</a>
4705</div>
4706
4707<div class="doc_text">
4708
4709<h5>Syntax:</h5>
4710
4711<pre>
4712 &lt;resultval&gt; = va_arg &lt;va_list*&gt; &lt;arglist&gt;, &lt;argty&gt;
4713</pre>
4714
4715<h5>Overview:</h5>
4716
4717<p>The '<tt>va_arg</tt>' instruction is used to access arguments passed through
4718the "variable argument" area of a function call. It is used to implement the
4719<tt>va_arg</tt> macro in C.</p>
4720
4721<h5>Arguments:</h5>
4722
4723<p>This instruction takes a <tt>va_list*</tt> value and the type of
4724the argument. It returns a value of the specified argument type and
4725increments the <tt>va_list</tt> to point to the next argument. The
4726actual type of <tt>va_list</tt> is target specific.</p>
4727
4728<h5>Semantics:</h5>
4729
4730<p>The '<tt>va_arg</tt>' instruction loads an argument of the specified
4731type from the specified <tt>va_list</tt> and causes the
4732<tt>va_list</tt> to point to the next argument. For more information,
4733see the variable argument handling <a href="#int_varargs">Intrinsic
4734Functions</a>.</p>
4735
4736<p>It is legal for this instruction to be called in a function which does not
4737take a variable number of arguments, for example, the <tt>vfprintf</tt>
4738function.</p>
4739
4740<p><tt>va_arg</tt> is an LLVM instruction instead of an <a
4741href="#intrinsics">intrinsic function</a> because it takes a type as an
4742argument.</p>
4743
4744<h5>Example:</h5>
4745
4746<p>See the <a href="#int_varargs">variable argument processing</a> section.</p>
4747
Dan Gohman60967192009-01-12 23:12:39 +0000[diff] [blame]4748<p>Note that the code generator does not yet fully support va_arg
4749 on many targets. Also, it does not currently support va_arg with
4750 aggregate types on any target.</p>
4751
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4752</div>
4753
4754<!-- *********************************************************************** -->
4755<div class="doc_section"> <a name="intrinsics">Intrinsic Functions</a> </div>
4756<!-- *********************************************************************** -->
4757
4758<div class="doc_text">
4759
4760<p>LLVM supports the notion of an "intrinsic function". These functions have
4761well known names and semantics and are required to follow certain restrictions.
4762Overall, these intrinsics represent an extension mechanism for the LLVM
4763language that does not require changing all of the transformations in LLVM when
4764adding to the language (or the bitcode reader/writer, the parser, etc...).</p>
4765
4766<p>Intrinsic function names must all start with an "<tt>llvm.</tt>" prefix. This
4767prefix is reserved in LLVM for intrinsic names; thus, function names may not
4768begin with this prefix. Intrinsic functions must always be external functions:
4769you cannot define the body of intrinsic functions. Intrinsic functions may
4770only be used in call or invoke instructions: it is illegal to take the address
4771of an intrinsic function. Additionally, because intrinsic functions are part
4772of the LLVM language, it is required if any are added that they be documented
4773here.</p>
4774
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4775<p>Some intrinsic functions can be overloaded, i.e., the intrinsic represents
4776a family of functions that perform the same operation but on different data
4777types. Because LLVM can represent over 8 million different integer types,
4778overloading is used commonly to allow an intrinsic function to operate on any
4779integer type. One or more of the argument types or the result type can be
4780overloaded to accept any integer type. Argument types may also be defined as
4781exactly matching a previous argument's type or the result type. This allows an
4782intrinsic function which accepts multiple arguments, but needs all of them to
4783be of the same type, to only be overloaded with respect to a single argument or
4784the result.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4785
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]4786<p>Overloaded intrinsics will have the names of its overloaded argument types
4787encoded into its function name, each preceded by a period. Only those types
4788which are overloaded result in a name suffix. Arguments whose type is matched
4789against another type do not. For example, the <tt>llvm.ctpop</tt> function can
4790take an integer of any width and returns an integer of exactly the same integer
4791width. This leads to a family of functions such as
4792<tt>i8 @llvm.ctpop.i8(i8 %val)</tt> and <tt>i29 @llvm.ctpop.i29(i29 %val)</tt>.
4793Only one type, the return type, is overloaded, and only one type suffix is
4794required. Because the argument's type is matched against the return type, it
4795does not require its own name suffix.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4796
4797<p>To learn how to add an intrinsic function, please see the
4798<a href="ExtendingLLVM.html">Extending LLVM Guide</a>.
4799</p>
4800
4801</div>
4802
4803<!-- ======================================================================= -->
4804<div class="doc_subsection">
4805 <a name="int_varargs">Variable Argument Handling Intrinsics</a>
4806</div>
4807
4808<div class="doc_text">
4809
4810<p>Variable argument support is defined in LLVM with the <a
4811 href="#i_va_arg"><tt>va_arg</tt></a> instruction and these three
4812intrinsic functions. These functions are related to the similarly
4813named macros defined in the <tt>&lt;stdarg.h&gt;</tt> header file.</p>
4814
4815<p>All of these functions operate on arguments that use a
4816target-specific value type "<tt>va_list</tt>". The LLVM assembly
4817language reference manual does not define what this type is, so all
4818transformations should be prepared to handle these functions regardless of
4819the type used.</p>
4820
4821<p>This example shows how the <a href="#i_va_arg"><tt>va_arg</tt></a>
4822instruction and the variable argument handling intrinsic functions are
4823used.</p>
4824
4825<div class="doc_code">
4826<pre>
4827define i32 @test(i32 %X, ...) {
4828 ; Initialize variable argument processing
4829 %ap = alloca i8*
4830 %ap2 = bitcast i8** %ap to i8*
4831 call void @llvm.va_start(i8* %ap2)
4832
4833 ; Read a single integer argument
4834 %tmp = va_arg i8** %ap, i32
4835
4836 ; Demonstrate usage of llvm.va_copy and llvm.va_end
4837 %aq = alloca i8*
4838 %aq2 = bitcast i8** %aq to i8*
4839 call void @llvm.va_copy(i8* %aq2, i8* %ap2)
4840 call void @llvm.va_end(i8* %aq2)
4841
4842 ; Stop processing of arguments.
4843 call void @llvm.va_end(i8* %ap2)
4844 ret i32 %tmp
4845}
4846
4847declare void @llvm.va_start(i8*)
4848declare void @llvm.va_copy(i8*, i8*)
4849declare void @llvm.va_end(i8*)
4850</pre>
4851</div>
4852
4853</div>
4854
4855<!-- _______________________________________________________________________ -->
4856<div class="doc_subsubsection">
4857 <a name="int_va_start">'<tt>llvm.va_start</tt>' Intrinsic</a>
4858</div>
4859
4860
4861<div class="doc_text">
4862<h5>Syntax:</h5>
4863<pre> declare void %llvm.va_start(i8* &lt;arglist&gt;)<br></pre>
4864<h5>Overview:</h5>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4865<p>The '<tt>llvm.va_start</tt>' intrinsic initializes
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4866<tt>*&lt;arglist&gt;</tt> for subsequent use by <tt><a
4867href="#i_va_arg">va_arg</a></tt>.</p>
4868
4869<h5>Arguments:</h5>
4870
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4871<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]4872
4873<h5>Semantics:</h5>
4874
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]4875<p>The '<tt>llvm.va_start</tt>' intrinsic works just like the <tt>va_start</tt>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4876macro available in C. In a target-dependent way, it initializes the
4877<tt>va_list</tt> element to which the argument points, so that the next call to
4878<tt>va_arg</tt> will produce the first variable argument passed to the function.
4879Unlike the C <tt>va_start</tt> macro, this intrinsic does not need to know the
4880last argument of the function as the compiler can figure that out.</p>
4881
4882</div>
4883
4884<!-- _______________________________________________________________________ -->
4885<div class="doc_subsubsection">
4886 <a name="int_va_end">'<tt>llvm.va_end</tt>' Intrinsic</a>
4887</div>
4888
4889<div class="doc_text">
4890<h5>Syntax:</h5>
4891<pre> declare void @llvm.va_end(i8* &lt;arglist&gt;)<br></pre>
4892<h5>Overview:</h5>
4893
4894<p>The '<tt>llvm.va_end</tt>' intrinsic destroys <tt>*&lt;arglist&gt;</tt>,
4895which has been initialized previously with <tt><a href="#int_va_start">llvm.va_start</a></tt>
4896or <tt><a href="#i_va_copy">llvm.va_copy</a></tt>.</p>
4897
4898<h5>Arguments:</h5>
4899
4900<p>The argument is a pointer to a <tt>va_list</tt> to destroy.</p>
4901
4902<h5>Semantics:</h5>
4903
4904<p>The '<tt>llvm.va_end</tt>' intrinsic works just like the <tt>va_end</tt>
4905macro available in C. In a target-dependent way, it destroys the
4906<tt>va_list</tt> element to which the argument points. Calls to <a
4907href="#int_va_start"><tt>llvm.va_start</tt></a> and <a href="#int_va_copy">
4908<tt>llvm.va_copy</tt></a> must be matched exactly with calls to
4909<tt>llvm.va_end</tt>.</p>
4910
4911</div>
4912
4913<!-- _______________________________________________________________________ -->
4914<div class="doc_subsubsection">
4915 <a name="int_va_copy">'<tt>llvm.va_copy</tt>' Intrinsic</a>
4916</div>
4917
4918<div class="doc_text">
4919
4920<h5>Syntax:</h5>
4921
4922<pre>
4923 declare void @llvm.va_copy(i8* &lt;destarglist&gt;, i8* &lt;srcarglist&gt;)
4924</pre>
4925
4926<h5>Overview:</h5>
4927
4928<p>The '<tt>llvm.va_copy</tt>' intrinsic copies the current argument position
4929from the source argument list to the destination argument list.</p>
4930
4931<h5>Arguments:</h5>
4932
4933<p>The first argument is a pointer to a <tt>va_list</tt> element to initialize.
4934The second argument is a pointer to a <tt>va_list</tt> element to copy from.</p>
4935
4936
4937<h5>Semantics:</h5>
4938
4939<p>The '<tt>llvm.va_copy</tt>' intrinsic works just like the <tt>va_copy</tt>
4940macro available in C. In a target-dependent way, it copies the source
4941<tt>va_list</tt> element into the destination <tt>va_list</tt> element. This
4942intrinsic is necessary because the <tt><a href="#int_va_start">
4943llvm.va_start</a></tt> intrinsic may be arbitrarily complex and require, for
4944example, memory allocation.</p>
4945
4946</div>
4947
4948<!-- ======================================================================= -->
4949<div class="doc_subsection">
4950 <a name="int_gc">Accurate Garbage Collection Intrinsics</a>
4951</div>
4952
4953<div class="doc_text">
4954
4955<p>
4956LLVM support for <a href="GarbageCollection.html">Accurate Garbage
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]4957Collection</a> (GC) requires the implementation and generation of these
4958intrinsics.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4959These intrinsics allow identification of <a href="#int_gcroot">GC roots on the
4960stack</a>, as well as garbage collector implementations that require <a
4961href="#int_gcread">read</a> and <a href="#int_gcwrite">write</a> barriers.
4962Front-ends for type-safe garbage collected languages should generate these
4963intrinsics to make use of the LLVM garbage collectors. For more details, see <a
4964href="GarbageCollection.html">Accurate Garbage Collection with LLVM</a>.
4965</p>
Christopher Lambcfe00962007-12-17 01:00:21 +0000[diff] [blame]4966
4967<p>The garbage collection intrinsics only operate on objects in the generic
4968 address space (address space zero).</p>
4969
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4970</div>
4971
4972<!-- _______________________________________________________________________ -->
4973<div class="doc_subsubsection">
4974 <a name="int_gcroot">'<tt>llvm.gcroot</tt>' Intrinsic</a>
4975</div>
4976
4977<div class="doc_text">
4978
4979<h5>Syntax:</h5>
4980
4981<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]4982 declare void @llvm.gcroot(i8** %ptrloc, i8* %metadata)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4983</pre>
4984
4985<h5>Overview:</h5>
4986
4987<p>The '<tt>llvm.gcroot</tt>' intrinsic declares the existence of a GC root to
4988the code generator, and allows some metadata to be associated with it.</p>
4989
4990<h5>Arguments:</h5>
4991
4992<p>The first argument specifies the address of a stack object that contains the
4993root pointer. The second pointer (which must be either a constant or a global
4994value address) contains the meta-data to be associated with the root.</p>
4995
4996<h5>Semantics:</h5>
4997
Chris Lattnera7d94ba2008-04-24 05:59:56 +0000[diff] [blame]4998<p>At runtime, a call to this intrinsic stores a null pointer into the "ptrloc"
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]4999location. At compile-time, the code generator generates information to allow
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]5000the runtime to find the pointer at GC safe points. The '<tt>llvm.gcroot</tt>'
5001intrinsic may only be used in a function which <a href="#gc">specifies a GC
5002algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5003
5004</div>
5005
5006
5007<!-- _______________________________________________________________________ -->
5008<div class="doc_subsubsection">
5009 <a name="int_gcread">'<tt>llvm.gcread</tt>' Intrinsic</a>
5010</div>
5011
5012<div class="doc_text">
5013
5014<h5>Syntax:</h5>
5015
5016<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5017 declare i8* @llvm.gcread(i8* %ObjPtr, i8** %Ptr)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5018</pre>
5019
5020<h5>Overview:</h5>
5021
5022<p>The '<tt>llvm.gcread</tt>' intrinsic identifies reads of references from heap
5023locations, allowing garbage collector implementations that require read
5024barriers.</p>
5025
5026<h5>Arguments:</h5>
5027
5028<p>The second argument is the address to read from, which should be an address
5029allocated from the garbage collector. The first object is a pointer to the
5030start of the referenced object, if needed by the language runtime (otherwise
5031null).</p>
5032
5033<h5>Semantics:</h5>
5034
5035<p>The '<tt>llvm.gcread</tt>' intrinsic has the same semantics as a load
5036instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]5037garbage collector runtime, as needed. The '<tt>llvm.gcread</tt>' intrinsic
5038may only be used in a function which <a href="#gc">specifies a GC
5039algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5040
5041</div>
5042
5043
5044<!-- _______________________________________________________________________ -->
5045<div class="doc_subsubsection">
5046 <a name="int_gcwrite">'<tt>llvm.gcwrite</tt>' Intrinsic</a>
5047</div>
5048
5049<div class="doc_text">
5050
5051<h5>Syntax:</h5>
5052
5053<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5054 declare void @llvm.gcwrite(i8* %P1, i8* %Obj, i8** %P2)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5055</pre>
5056
5057<h5>Overview:</h5>
5058
5059<p>The '<tt>llvm.gcwrite</tt>' intrinsic identifies writes of references to heap
5060locations, allowing garbage collector implementations that require write
5061barriers (such as generational or reference counting collectors).</p>
5062
5063<h5>Arguments:</h5>
5064
5065<p>The first argument is the reference to store, the second is the start of the
5066object to store it to, and the third is the address of the field of Obj to
5067store to. If the runtime does not require a pointer to the object, Obj may be
5068null.</p>
5069
5070<h5>Semantics:</h5>
5071
5072<p>The '<tt>llvm.gcwrite</tt>' intrinsic has the same semantics as a store
5073instruction, but may be replaced with substantially more complex code by the
Gordon Henriksen40393542007-12-25 02:31:26 +0000[diff] [blame]5074garbage collector runtime, as needed. The '<tt>llvm.gcwrite</tt>' intrinsic
5075may only be used in a function which <a href="#gc">specifies a GC
5076algorithm</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5077
5078</div>
5079
5080
5081
5082<!-- ======================================================================= -->
5083<div class="doc_subsection">
5084 <a name="int_codegen">Code Generator Intrinsics</a>
5085</div>
5086
5087<div class="doc_text">
5088<p>
5089These intrinsics are provided by LLVM to expose special features that may only
5090be implemented with code generator support.
5091</p>
5092
5093</div>
5094
5095<!-- _______________________________________________________________________ -->
5096<div class="doc_subsubsection">
5097 <a name="int_returnaddress">'<tt>llvm.returnaddress</tt>' Intrinsic</a>
5098</div>
5099
5100<div class="doc_text">
5101
5102<h5>Syntax:</h5>
5103<pre>
5104 declare i8 *@llvm.returnaddress(i32 &lt;level&gt;)
5105</pre>
5106
5107<h5>Overview:</h5>
5108
5109<p>
5110The '<tt>llvm.returnaddress</tt>' intrinsic attempts to compute a
5111target-specific value indicating the return address of the current function
5112or one of its callers.
5113</p>
5114
5115<h5>Arguments:</h5>
5116
5117<p>
5118The argument to this intrinsic indicates which function to return the address
5119for. Zero indicates the calling function, one indicates its caller, etc. The
5120argument is <b>required</b> to be a constant integer value.
5121</p>
5122
5123<h5>Semantics:</h5>
5124
5125<p>
5126The '<tt>llvm.returnaddress</tt>' intrinsic either returns a pointer indicating
5127the return address of the specified call frame, or zero if it cannot be
5128identified. The value returned by this intrinsic is likely to be incorrect or 0
5129for arguments other than zero, so it should only be used for debugging purposes.
5130</p>
5131
5132<p>
5133Note that calling this intrinsic does not prevent function inlining or other
5134aggressive transformations, so the value returned may not be that of the obvious
5135source-language caller.
5136</p>
5137</div>
5138
5139
5140<!-- _______________________________________________________________________ -->
5141<div class="doc_subsubsection">
5142 <a name="int_frameaddress">'<tt>llvm.frameaddress</tt>' Intrinsic</a>
5143</div>
5144
5145<div class="doc_text">
5146
5147<h5>Syntax:</h5>
5148<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5149 declare i8 *@llvm.frameaddress(i32 &lt;level&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5150</pre>
5151
5152<h5>Overview:</h5>
5153
5154<p>
5155The '<tt>llvm.frameaddress</tt>' intrinsic attempts to return the
5156target-specific frame pointer value for the specified stack frame.
5157</p>
5158
5159<h5>Arguments:</h5>
5160
5161<p>
5162The argument to this intrinsic indicates which function to return the frame
5163pointer for. Zero indicates the calling function, one indicates its caller,
5164etc. The argument is <b>required</b> to be a constant integer value.
5165</p>
5166
5167<h5>Semantics:</h5>
5168
5169<p>
5170The '<tt>llvm.frameaddress</tt>' intrinsic either returns a pointer indicating
5171the frame address of the specified call frame, or zero if it cannot be
5172identified. The value returned by this intrinsic is likely to be incorrect or 0
5173for arguments other than zero, so it should only be used for debugging purposes.
5174</p>
5175
5176<p>
5177Note that calling this intrinsic does not prevent function inlining or other
5178aggressive transformations, so the value returned may not be that of the obvious
5179source-language caller.
5180</p>
5181</div>
5182
5183<!-- _______________________________________________________________________ -->
5184<div class="doc_subsubsection">
5185 <a name="int_stacksave">'<tt>llvm.stacksave</tt>' Intrinsic</a>
5186</div>
5187
5188<div class="doc_text">
5189
5190<h5>Syntax:</h5>
5191<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5192 declare i8 *@llvm.stacksave()
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5193</pre>
5194
5195<h5>Overview:</h5>
5196
5197<p>
5198The '<tt>llvm.stacksave</tt>' intrinsic is used to remember the current state of
5199the function stack, for use with <a href="#int_stackrestore">
5200<tt>llvm.stackrestore</tt></a>. This is useful for implementing language
5201features like scoped automatic variable sized arrays in C99.
5202</p>
5203
5204<h5>Semantics:</h5>
5205
5206<p>
5207This intrinsic returns a opaque pointer value that can be passed to <a
5208href="#int_stackrestore"><tt>llvm.stackrestore</tt></a>. When an
5209<tt>llvm.stackrestore</tt> intrinsic is executed with a value saved from
5210<tt>llvm.stacksave</tt>, it effectively restores the state of the stack to the
5211state it was in when the <tt>llvm.stacksave</tt> intrinsic executed. In
5212practice, this pops any <a href="#i_alloca">alloca</a> blocks from the stack
5213that were allocated after the <tt>llvm.stacksave</tt> was executed.
5214</p>
5215
5216</div>
5217
5218<!-- _______________________________________________________________________ -->
5219<div class="doc_subsubsection">
5220 <a name="int_stackrestore">'<tt>llvm.stackrestore</tt>' Intrinsic</a>
5221</div>
5222
5223<div class="doc_text">
5224
5225<h5>Syntax:</h5>
5226<pre>
5227 declare void @llvm.stackrestore(i8 * %ptr)
5228</pre>
5229
5230<h5>Overview:</h5>
5231
5232<p>
5233The '<tt>llvm.stackrestore</tt>' intrinsic is used to restore the state of
5234the function stack to the state it was in when the corresponding <a
5235href="#int_stacksave"><tt>llvm.stacksave</tt></a> intrinsic executed. This is
5236useful for implementing language features like scoped automatic variable sized
5237arrays in C99.
5238</p>
5239
5240<h5>Semantics:</h5>
5241
5242<p>
5243See the description for <a href="#int_stacksave"><tt>llvm.stacksave</tt></a>.
5244</p>
5245
5246</div>
5247
5248
5249<!-- _______________________________________________________________________ -->
5250<div class="doc_subsubsection">
5251 <a name="int_prefetch">'<tt>llvm.prefetch</tt>' Intrinsic</a>
5252</div>
5253
5254<div class="doc_text">
5255
5256<h5>Syntax:</h5>
5257<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5258 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]5259</pre>
5260
5261<h5>Overview:</h5>
5262
5263
5264<p>
5265The '<tt>llvm.prefetch</tt>' intrinsic is a hint to the code generator to insert
5266a prefetch instruction if supported; otherwise, it is a noop. Prefetches have
5267no
5268effect on the behavior of the program but can change its performance
5269characteristics.
5270</p>
5271
5272<h5>Arguments:</h5>
5273
5274<p>
5275<tt>address</tt> is the address to be prefetched, <tt>rw</tt> is the specifier
5276determining if the fetch should be for a read (0) or write (1), and
5277<tt>locality</tt> is a temporal locality specifier ranging from (0) - no
5278locality, to (3) - extremely local keep in cache. The <tt>rw</tt> and
5279<tt>locality</tt> arguments must be constant integers.
5280</p>
5281
5282<h5>Semantics:</h5>
5283
5284<p>
5285This intrinsic does not modify the behavior of the program. In particular,
5286prefetches cannot trap and do not produce a value. On targets that support this
5287intrinsic, the prefetch can provide hints to the processor cache for better
5288performance.
5289</p>
5290
5291</div>
5292
5293<!-- _______________________________________________________________________ -->
5294<div class="doc_subsubsection">
5295 <a name="int_pcmarker">'<tt>llvm.pcmarker</tt>' Intrinsic</a>
5296</div>
5297
5298<div class="doc_text">
5299
5300<h5>Syntax:</h5>
5301<pre>
Chris Lattner38bd5dd2007-09-21 17:30:40 +0000[diff] [blame]5302 declare void @llvm.pcmarker(i32 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5303</pre>
5304
5305<h5>Overview:</h5>
5306
5307
5308<p>
5309The '<tt>llvm.pcmarker</tt>' intrinsic is a method to export a Program Counter
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]5310(PC) in a region of
5311code to simulators and other tools. The method is target specific, but it is
5312expected that the marker will use exported symbols to transmit the PC of the
5313marker.
5314The marker makes no guarantees that it will remain with any specific instruction
5315after optimizations. It is possible that the presence of a marker will inhibit
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5316optimizations. The intended use is to be inserted after optimizations to allow
5317correlations of simulation runs.
5318</p>
5319
5320<h5>Arguments:</h5>
5321
5322<p>
5323<tt>id</tt> is a numerical id identifying the marker.
5324</p>
5325
5326<h5>Semantics:</h5>
5327
5328<p>
5329This intrinsic does not modify the behavior of the program. Backends that do not
5330support this intrinisic may ignore it.
5331</p>
5332
5333</div>
5334
5335<!-- _______________________________________________________________________ -->
5336<div class="doc_subsubsection">
5337 <a name="int_readcyclecounter">'<tt>llvm.readcyclecounter</tt>' Intrinsic</a>
5338</div>
5339
5340<div class="doc_text">
5341
5342<h5>Syntax:</h5>
5343<pre>
5344 declare i64 @llvm.readcyclecounter( )
5345</pre>
5346
5347<h5>Overview:</h5>
5348
5349
5350<p>
5351The '<tt>llvm.readcyclecounter</tt>' intrinsic provides access to the cycle
5352counter register (or similar low latency, high accuracy clocks) on those targets
5353that support it. On X86, it should map to RDTSC. On Alpha, it should map to RPCC.
5354As the backing counters overflow quickly (on the order of 9 seconds on alpha), this
5355should only be used for small timings.
5356</p>
5357
5358<h5>Semantics:</h5>
5359
5360<p>
5361When directly supported, reading the cycle counter should not modify any memory.
5362Implementations are allowed to either return a application specific value or a
5363system wide value. On backends without support, this is lowered to a constant 0.
5364</p>
5365
5366</div>
5367
5368<!-- ======================================================================= -->
5369<div class="doc_subsection">
5370 <a name="int_libc">Standard C Library Intrinsics</a>
5371</div>
5372
5373<div class="doc_text">
5374<p>
5375LLVM provides intrinsics for a few important standard C library functions.
5376These intrinsics allow source-language front-ends to pass information about the
5377alignment of the pointer arguments to the code generator, providing opportunity
5378for more efficient code generation.
5379</p>
5380
5381</div>
5382
5383<!-- _______________________________________________________________________ -->
5384<div class="doc_subsubsection">
5385 <a name="int_memcpy">'<tt>llvm.memcpy</tt>' Intrinsic</a>
5386</div>
5387
5388<div class="doc_text">
5389
5390<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5391<p>This is an overloaded intrinsic. You can use llvm.memcpy on any integer bit
5392width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5393<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5394 declare void @llvm.memcpy.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5395 i8 &lt;len&gt;, i32 &lt;align&gt;)
5396 declare void @llvm.memcpy.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5397 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5398 declare void @llvm.memcpy.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5399 i32 &lt;len&gt;, i32 &lt;align&gt;)
5400 declare void @llvm.memcpy.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5401 i64 &lt;len&gt;, i32 &lt;align&gt;)
5402</pre>
5403
5404<h5>Overview:</h5>
5405
5406<p>
5407The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5408location to the destination location.
5409</p>
5410
5411<p>
5412Note that, unlike the standard libc function, the <tt>llvm.memcpy.*</tt>
5413intrinsics do not return a value, and takes an extra alignment argument.
5414</p>
5415
5416<h5>Arguments:</h5>
5417
5418<p>
5419The first argument is a pointer to the destination, the second is a pointer to
5420the source. The third argument is an integer argument
5421specifying the number of bytes to copy, and the fourth argument is the alignment
5422of the source and destination locations.
5423</p>
5424
5425<p>
5426If the call to this intrinisic has an alignment value that is not 0 or 1, then
5427the caller guarantees that both the source and destination pointers are aligned
5428to that boundary.
5429</p>
5430
5431<h5>Semantics:</h5>
5432
5433<p>
5434The '<tt>llvm.memcpy.*</tt>' intrinsics copy a block of memory from the source
5435location to the destination location, which are not allowed to overlap. It
5436copies "len" bytes of memory over. If the argument is known to be aligned to
5437some boundary, this can be specified as the fourth argument, otherwise it should
5438be set to 0 or 1.
5439</p>
5440</div>
5441
5442
5443<!-- _______________________________________________________________________ -->
5444<div class="doc_subsubsection">
5445 <a name="int_memmove">'<tt>llvm.memmove</tt>' Intrinsic</a>
5446</div>
5447
5448<div class="doc_text">
5449
5450<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5451<p>This is an overloaded intrinsic. You can use llvm.memmove on any integer bit
5452width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5453<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5454 declare void @llvm.memmove.i8(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5455 i8 &lt;len&gt;, i32 &lt;align&gt;)
5456 declare void @llvm.memmove.i16(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5457 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5458 declare void @llvm.memmove.i32(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5459 i32 &lt;len&gt;, i32 &lt;align&gt;)
5460 declare void @llvm.memmove.i64(i8 * &lt;dest&gt;, i8 * &lt;src&gt;,
5461 i64 &lt;len&gt;, i32 &lt;align&gt;)
5462</pre>
5463
5464<h5>Overview:</h5>
5465
5466<p>
5467The '<tt>llvm.memmove.*</tt>' intrinsics move a block of memory from the source
5468location to the destination location. It is similar to the
Chris Lattnerdba16ea2008-01-06 19:51:52 +0000[diff] [blame]5469'<tt>llvm.memcpy</tt>' intrinsic but allows the two memory locations to overlap.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5470</p>
5471
5472<p>
5473Note that, unlike the standard libc function, the <tt>llvm.memmove.*</tt>
5474intrinsics do not return a value, and takes an extra alignment argument.
5475</p>
5476
5477<h5>Arguments:</h5>
5478
5479<p>
5480The first argument is a pointer to the destination, the second is a pointer to
5481the source. The third argument is an integer argument
5482specifying the number of bytes to copy, and the fourth argument is the alignment
5483of the source and destination locations.
5484</p>
5485
5486<p>
5487If the call to this intrinisic has an alignment value that is not 0 or 1, then
5488the caller guarantees that the source and destination pointers are aligned to
5489that boundary.
5490</p>
5491
5492<h5>Semantics:</h5>
5493
5494<p>
5495The '<tt>llvm.memmove.*</tt>' intrinsics copy a block of memory from the source
5496location to the destination location, which may overlap. It
5497copies "len" bytes of memory over. If the argument is known to be aligned to
5498some boundary, this can be specified as the fourth argument, otherwise it should
5499be set to 0 or 1.
5500</p>
5501</div>
5502
5503
5504<!-- _______________________________________________________________________ -->
5505<div class="doc_subsubsection">
5506 <a name="int_memset">'<tt>llvm.memset.*</tt>' Intrinsics</a>
5507</div>
5508
5509<div class="doc_text">
5510
5511<h5>Syntax:</h5>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5512<p>This is an overloaded intrinsic. You can use llvm.memset on any integer bit
5513width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5514<pre>
Chris Lattner82c2e432008-11-21 16:42:48 +0000[diff] [blame]5515 declare void @llvm.memset.i8(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5516 i8 &lt;len&gt;, i32 &lt;align&gt;)
5517 declare void @llvm.memset.i16(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5518 i16 &lt;len&gt;, i32 &lt;align&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5519 declare void @llvm.memset.i32(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5520 i32 &lt;len&gt;, i32 &lt;align&gt;)
5521 declare void @llvm.memset.i64(i8 * &lt;dest&gt;, i8 &lt;val&gt;,
5522 i64 &lt;len&gt;, i32 &lt;align&gt;)
5523</pre>
5524
5525<h5>Overview:</h5>
5526
5527<p>
5528The '<tt>llvm.memset.*</tt>' intrinsics fill a block of memory with a particular
5529byte value.
5530</p>
5531
5532<p>
5533Note that, unlike the standard libc function, the <tt>llvm.memset</tt> intrinsic
5534does not return a value, and takes an extra alignment argument.
5535</p>
5536
5537<h5>Arguments:</h5>
5538
5539<p>
5540The first argument is a pointer to the destination to fill, the second is the
5541byte value to fill it with, the third argument is an integer
5542argument specifying the number of bytes to fill, and the fourth argument is the
5543known alignment of destination location.
5544</p>
5545
5546<p>
5547If the call to this intrinisic has an alignment value that is not 0 or 1, then
5548the caller guarantees that the destination pointer is aligned to that boundary.
5549</p>
5550
5551<h5>Semantics:</h5>
5552
5553<p>
5554The '<tt>llvm.memset.*</tt>' intrinsics fill "len" bytes of memory starting at
5555the
5556destination location. If the argument is known to be aligned to some boundary,
5557this can be specified as the fourth argument, otherwise it should be set to 0 or
55581.
5559</p>
5560</div>
5561
5562
5563<!-- _______________________________________________________________________ -->
5564<div class="doc_subsubsection">
5565 <a name="int_sqrt">'<tt>llvm.sqrt.*</tt>' Intrinsic</a>
5566</div>
5567
5568<div class="doc_text">
5569
5570<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5571<p>This is an overloaded intrinsic. You can use <tt>llvm.sqrt</tt> on any
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5572floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5573types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5574<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5575 declare float @llvm.sqrt.f32(float %Val)
5576 declare double @llvm.sqrt.f64(double %Val)
5577 declare x86_fp80 @llvm.sqrt.f80(x86_fp80 %Val)
5578 declare fp128 @llvm.sqrt.f128(fp128 %Val)
5579 declare ppc_fp128 @llvm.sqrt.ppcf128(ppc_fp128 %Val)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5580</pre>
5581
5582<h5>Overview:</h5>
5583
5584<p>
5585The '<tt>llvm.sqrt</tt>' intrinsics return the sqrt of the specified operand,
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5586returning the same value as the libm '<tt>sqrt</tt>' functions would. Unlike
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5587<tt>sqrt</tt> in libm, however, <tt>llvm.sqrt</tt> has undefined behavior for
Chris Lattnerf914a002008-01-29 07:00:44 +0000[diff] [blame]5588negative numbers other than -0.0 (which allows for better optimization, because
5589there is no need to worry about errno being set). <tt>llvm.sqrt(-0.0)</tt> is
5590defined to return -0.0 like IEEE sqrt.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5591</p>
5592
5593<h5>Arguments:</h5>
5594
5595<p>
5596The argument and return value are floating point numbers of the same type.
5597</p>
5598
5599<h5>Semantics:</h5>
5600
5601<p>
5602This function returns the sqrt of the specified operand if it is a nonnegative
5603floating point number.
5604</p>
5605</div>
5606
5607<!-- _______________________________________________________________________ -->
5608<div class="doc_subsubsection">
5609 <a name="int_powi">'<tt>llvm.powi.*</tt>' Intrinsic</a>
5610</div>
5611
5612<div class="doc_text">
5613
5614<h5>Syntax:</h5>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5615<p>This is an overloaded intrinsic. You can use <tt>llvm.powi</tt> on any
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5616floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5617types however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5618<pre>
Dale Johannesenf9adbb62007-10-02 17:47:38 +0000[diff] [blame]5619 declare float @llvm.powi.f32(float %Val, i32 %power)
5620 declare double @llvm.powi.f64(double %Val, i32 %power)
5621 declare x86_fp80 @llvm.powi.f80(x86_fp80 %Val, i32 %power)
5622 declare fp128 @llvm.powi.f128(fp128 %Val, i32 %power)
5623 declare ppc_fp128 @llvm.powi.ppcf128(ppc_fp128 %Val, i32 %power)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5624</pre>
5625
5626<h5>Overview:</h5>
5627
5628<p>
5629The '<tt>llvm.powi.*</tt>' intrinsics return the first operand raised to the
5630specified (positive or negative) power. The order of evaluation of
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5631multiplications is not defined. When a vector of floating point type is
5632used, the second argument remains a scalar integer value.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5633</p>
5634
5635<h5>Arguments:</h5>
5636
5637<p>
5638The second argument is an integer power, and the first is a value to raise to
5639that power.
5640</p>
5641
5642<h5>Semantics:</h5>
5643
5644<p>
5645This function returns the first value raised to the second power with an
5646unspecified sequence of rounding operations.</p>
5647</div>
5648
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5649<!-- _______________________________________________________________________ -->
5650<div class="doc_subsubsection">
5651 <a name="int_sin">'<tt>llvm.sin.*</tt>' Intrinsic</a>
5652</div>
5653
5654<div class="doc_text">
5655
5656<h5>Syntax:</h5>
5657<p>This is an overloaded intrinsic. You can use <tt>llvm.sin</tt> on any
5658floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5659types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5660<pre>
5661 declare float @llvm.sin.f32(float %Val)
5662 declare double @llvm.sin.f64(double %Val)
5663 declare x86_fp80 @llvm.sin.f80(x86_fp80 %Val)
5664 declare fp128 @llvm.sin.f128(fp128 %Val)
5665 declare ppc_fp128 @llvm.sin.ppcf128(ppc_fp128 %Val)
5666</pre>
5667
5668<h5>Overview:</h5>
5669
5670<p>
5671The '<tt>llvm.sin.*</tt>' intrinsics return the sine of the operand.
5672</p>
5673
5674<h5>Arguments:</h5>
5675
5676<p>
5677The argument and return value are floating point numbers of the same type.
5678</p>
5679
5680<h5>Semantics:</h5>
5681
5682<p>
5683This function returns the sine of the specified operand, returning the
5684same values as the libm <tt>sin</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5685conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5686</div>
5687
5688<!-- _______________________________________________________________________ -->
5689<div class="doc_subsubsection">
5690 <a name="int_cos">'<tt>llvm.cos.*</tt>' Intrinsic</a>
5691</div>
5692
5693<div class="doc_text">
5694
5695<h5>Syntax:</h5>
5696<p>This is an overloaded intrinsic. You can use <tt>llvm.cos</tt> on any
5697floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5698types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5699<pre>
5700 declare float @llvm.cos.f32(float %Val)
5701 declare double @llvm.cos.f64(double %Val)
5702 declare x86_fp80 @llvm.cos.f80(x86_fp80 %Val)
5703 declare fp128 @llvm.cos.f128(fp128 %Val)
5704 declare ppc_fp128 @llvm.cos.ppcf128(ppc_fp128 %Val)
5705</pre>
5706
5707<h5>Overview:</h5>
5708
5709<p>
5710The '<tt>llvm.cos.*</tt>' intrinsics return the cosine of the operand.
5711</p>
5712
5713<h5>Arguments:</h5>
5714
5715<p>
5716The argument and return value are floating point numbers of the same type.
5717</p>
5718
5719<h5>Semantics:</h5>
5720
5721<p>
5722This function returns the cosine of the specified operand, returning the
5723same values as the libm <tt>cos</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5724conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5725</div>
5726
5727<!-- _______________________________________________________________________ -->
5728<div class="doc_subsubsection">
5729 <a name="int_pow">'<tt>llvm.pow.*</tt>' Intrinsic</a>
5730</div>
5731
5732<div class="doc_text">
5733
5734<h5>Syntax:</h5>
5735<p>This is an overloaded intrinsic. You can use <tt>llvm.pow</tt> on any
5736floating point or vector of floating point type. Not all targets support all
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5737types however.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5738<pre>
5739 declare float @llvm.pow.f32(float %Val, float %Power)
5740 declare double @llvm.pow.f64(double %Val, double %Power)
5741 declare x86_fp80 @llvm.pow.f80(x86_fp80 %Val, x86_fp80 %Power)
5742 declare fp128 @llvm.pow.f128(fp128 %Val, fp128 %Power)
5743 declare ppc_fp128 @llvm.pow.ppcf128(ppc_fp128 %Val, ppc_fp128 Power)
5744</pre>
5745
5746<h5>Overview:</h5>
5747
5748<p>
5749The '<tt>llvm.pow.*</tt>' intrinsics return the first operand raised to the
5750specified (positive or negative) power.
5751</p>
5752
5753<h5>Arguments:</h5>
5754
5755<p>
5756The second argument is a floating point power, and the first is a value to
5757raise to that power.
5758</p>
5759
5760<h5>Semantics:</h5>
5761
5762<p>
5763This function returns the first value raised to the second power,
5764returning the
5765same values as the libm <tt>pow</tt> functions would, and handles error
Dan Gohmaneaba92e2007-10-17 18:05:13 +0000[diff] [blame]5766conditions in the same way.</p>
Dan Gohman361079c2007-10-15 20:30:11 +0000[diff] [blame]5767</div>
5768
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5769
5770<!-- ======================================================================= -->
5771<div class="doc_subsection">
5772 <a name="int_manip">Bit Manipulation Intrinsics</a>
5773</div>
5774
5775<div class="doc_text">
5776<p>
5777LLVM provides intrinsics for a few important bit manipulation operations.
5778These allow efficient code generation for some algorithms.
5779</p>
5780
5781</div>
5782
5783<!-- _______________________________________________________________________ -->
5784<div class="doc_subsubsection">
5785 <a name="int_bswap">'<tt>llvm.bswap.*</tt>' Intrinsics</a>
5786</div>
5787
5788<div class="doc_text">
5789
5790<h5>Syntax:</h5>
5791<p>This is an overloaded intrinsic function. You can use bswap on any integer
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5792type that is an even number of bytes (i.e. BitWidth % 16 == 0).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5793<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5794 declare i16 @llvm.bswap.i16(i16 &lt;id&gt;)
5795 declare i32 @llvm.bswap.i32(i32 &lt;id&gt;)
5796 declare i64 @llvm.bswap.i64(i64 &lt;id&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5797</pre>
5798
5799<h5>Overview:</h5>
5800
5801<p>
5802The '<tt>llvm.bswap</tt>' family of intrinsics is used to byte swap integer
5803values with an even number of bytes (positive multiple of 16 bits). These are
5804useful for performing operations on data that is not in the target's native
5805byte order.
5806</p>
5807
5808<h5>Semantics:</h5>
5809
5810<p>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5811The <tt>llvm.bswap.i16</tt> intrinsic returns an i16 value that has the high
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5812and low byte of the input i16 swapped. Similarly, the <tt>llvm.bswap.i32</tt>
5813intrinsic returns an i32 value that has the four bytes of the input i32
5814swapped, so that if the input bytes are numbered 0, 1, 2, 3 then the returned
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5815i32 will have its bytes in 3, 2, 1, 0 order. The <tt>llvm.bswap.i48</tt>,
5816<tt>llvm.bswap.i64</tt> and other intrinsics extend this concept to
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5817additional even-byte lengths (6 bytes, 8 bytes and more, respectively).
5818</p>
5819
5820</div>
5821
5822<!-- _______________________________________________________________________ -->
5823<div class="doc_subsubsection">
5824 <a name="int_ctpop">'<tt>llvm.ctpop.*</tt>' Intrinsic</a>
5825</div>
5826
5827<div class="doc_text">
5828
5829<h5>Syntax:</h5>
5830<p>This is an overloaded intrinsic. You can use llvm.ctpop on any integer bit
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5831width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5832<pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]5833 declare i8 @llvm.ctpop.i8(i8 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5834 declare i16 @llvm.ctpop.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5835 declare i32 @llvm.ctpop.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5836 declare i64 @llvm.ctpop.i64(i64 &lt;src&gt;)
5837 declare i256 @llvm.ctpop.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5838</pre>
5839
5840<h5>Overview:</h5>
5841
5842<p>
5843The '<tt>llvm.ctpop</tt>' family of intrinsics counts the number of bits set in a
5844value.
5845</p>
5846
5847<h5>Arguments:</h5>
5848
5849<p>
5850The only argument is the value to be counted. The argument may be of any
5851integer type. The return type must match the argument type.
5852</p>
5853
5854<h5>Semantics:</h5>
5855
5856<p>
5857The '<tt>llvm.ctpop</tt>' intrinsic counts the 1's in a variable.
5858</p>
5859</div>
5860
5861<!-- _______________________________________________________________________ -->
5862<div class="doc_subsubsection">
5863 <a name="int_ctlz">'<tt>llvm.ctlz.*</tt>' Intrinsic</a>
5864</div>
5865
5866<div class="doc_text">
5867
5868<h5>Syntax:</h5>
5869<p>This is an overloaded intrinsic. You can use <tt>llvm.ctlz</tt> on any
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5870integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5871<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5872 declare i8 @llvm.ctlz.i8 (i8 &lt;src&gt;)
5873 declare i16 @llvm.ctlz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5874 declare i32 @llvm.ctlz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5875 declare i64 @llvm.ctlz.i64(i64 &lt;src&gt;)
5876 declare i256 @llvm.ctlz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5877</pre>
5878
5879<h5>Overview:</h5>
5880
5881<p>
5882The '<tt>llvm.ctlz</tt>' family of intrinsic functions counts the number of
5883leading zeros in a variable.
5884</p>
5885
5886<h5>Arguments:</h5>
5887
5888<p>
5889The only argument is the value to be counted. The argument may be of any
5890integer type. The return type must match the argument type.
5891</p>
5892
5893<h5>Semantics:</h5>
5894
5895<p>
5896The '<tt>llvm.ctlz</tt>' intrinsic counts the leading (most significant) zeros
5897in a variable. If the src == 0 then the result is the size in bits of the type
5898of src. For example, <tt>llvm.ctlz(i32 2) = 30</tt>.
5899</p>
5900</div>
5901
5902
5903
5904<!-- _______________________________________________________________________ -->
5905<div class="doc_subsubsection">
5906 <a name="int_cttz">'<tt>llvm.cttz.*</tt>' Intrinsic</a>
5907</div>
5908
5909<div class="doc_text">
5910
5911<h5>Syntax:</h5>
5912<p>This is an overloaded intrinsic. You can use <tt>llvm.cttz</tt> on any
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5913integer bit width. Not all targets support all bit widths however.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5914<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5915 declare i8 @llvm.cttz.i8 (i8 &lt;src&gt;)
5916 declare i16 @llvm.cttz.i16(i16 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5917 declare i32 @llvm.cttz.i32(i32 &lt;src&gt;)
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5918 declare i64 @llvm.cttz.i64(i64 &lt;src&gt;)
5919 declare i256 @llvm.cttz.i256(i256 &lt;src&gt;)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5920</pre>
5921
5922<h5>Overview:</h5>
5923
5924<p>
5925The '<tt>llvm.cttz</tt>' family of intrinsic functions counts the number of
5926trailing zeros.
5927</p>
5928
5929<h5>Arguments:</h5>
5930
5931<p>
5932The only argument is the value to be counted. The argument may be of any
5933integer type. The return type must match the argument type.
5934</p>
5935
5936<h5>Semantics:</h5>
5937
5938<p>
5939The '<tt>llvm.cttz</tt>' intrinsic counts the trailing (least significant) zeros
5940in a variable. If the src == 0 then the result is the size in bits of the type
5941of src. For example, <tt>llvm.cttz(2) = 1</tt>.
5942</p>
5943</div>
5944
5945<!-- _______________________________________________________________________ -->
5946<div class="doc_subsubsection">
5947 <a name="int_part_select">'<tt>llvm.part.select.*</tt>' Intrinsic</a>
5948</div>
5949
5950<div class="doc_text">
5951
5952<h5>Syntax:</h5>
5953<p>This is an overloaded intrinsic. You can use <tt>llvm.part.select</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5954on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5955<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]5956 declare i17 @llvm.part.select.i17 (i17 %val, i32 %loBit, i32 %hiBit)
5957 declare i29 @llvm.part.select.i29 (i29 %val, i32 %loBit, i32 %hiBit)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5958</pre>
5959
5960<h5>Overview:</h5>
5961<p>The '<tt>llvm.part.select</tt>' family of intrinsic functions selects a
5962range of bits from an integer value and returns them in the same bit width as
5963the original value.</p>
5964
5965<h5>Arguments:</h5>
5966<p>The first argument, <tt>%val</tt> and the result may be integer types of
5967any bit width but they must have the same bit width. The second and third
5968arguments must be <tt>i32</tt> type since they specify only a bit index.</p>
5969
5970<h5>Semantics:</h5>
5971<p>The operation of the '<tt>llvm.part.select</tt>' intrinsic has two modes
5972of operation: forwards and reverse. If <tt>%loBit</tt> is greater than
5973<tt>%hiBits</tt> then the intrinsic operates in reverse mode. Otherwise it
5974operates in forward mode.</p>
5975<p>In forward mode, this intrinsic is the equivalent of shifting <tt>%val</tt>
5976right by <tt>%loBit</tt> bits and then ANDing it with a mask with
5977only the <tt>%hiBit - %loBit</tt> bits set, as follows:</p>
5978<ol>
5979 <li>The <tt>%val</tt> is shifted right (LSHR) by the number of bits specified
5980 by <tt>%loBits</tt>. This normalizes the value to the low order bits.</li>
5981 <li>The <tt>%loBits</tt> value is subtracted from the <tt>%hiBits</tt> value
5982 to determine the number of bits to retain.</li>
5983 <li>A mask of the retained bits is created by shifting a -1 value.</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]5984 <li>The mask is ANDed with <tt>%val</tt> to produce the result.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]5985</ol>
5986<p>In reverse mode, a similar computation is made except that the bits are
5987returned in the reverse order. So, for example, if <tt>X</tt> has the value
5988<tt>i16 0x0ACF (101011001111)</tt> and we apply
5989<tt>part.select(i16 X, 8, 3)</tt> to it, we get back the value
5990<tt>i16 0x0026 (000000100110)</tt>.</p>
5991</div>
5992
5993<div class="doc_subsubsection">
5994 <a name="int_part_set">'<tt>llvm.part.set.*</tt>' Intrinsic</a>
5995</div>
5996
5997<div class="doc_text">
5998
5999<h5>Syntax:</h5>
6000<p>This is an overloaded intrinsic. You can use <tt>llvm.part.set</tt>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6001on any integer bit width.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6002<pre>
Chandler Carrutha228e392007-08-04 01:51:18 +0000[diff] [blame]6003 declare i17 @llvm.part.set.i17.i9 (i17 %val, i9 %repl, i32 %lo, i32 %hi)
6004 declare i29 @llvm.part.set.i29.i9 (i29 %val, i9 %repl, i32 %lo, i32 %hi)
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6005</pre>
6006
6007<h5>Overview:</h5>
6008<p>The '<tt>llvm.part.set</tt>' family of intrinsic functions replaces a range
6009of bits in an integer value with another integer value. It returns the integer
6010with the replaced bits.</p>
6011
6012<h5>Arguments:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6013<p>The first argument, <tt>%val</tt>, and the result may be integer types of
6014any bit width, but they must have the same bit width. <tt>%val</tt> is the value
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6015whose bits will be replaced. The second argument, <tt>%repl</tt> may be an
6016integer of any bit width. The third and fourth arguments must be <tt>i32</tt>
6017type since they specify only a bit index.</p>
6018
6019<h5>Semantics:</h5>
6020<p>The operation of the '<tt>llvm.part.set</tt>' intrinsic has two modes
6021of operation: forwards and reverse. If <tt>%lo</tt> is greater than
6022<tt>%hi</tt> then the intrinsic operates in reverse mode. Otherwise it
6023operates in forward mode.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6024
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6025<p>For both modes, the <tt>%repl</tt> value is prepared for use by either
6026truncating it down to the size of the replacement area or zero extending it
6027up to that size.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6028
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6029<p>In forward mode, the bits between <tt>%lo</tt> and <tt>%hi</tt> (inclusive)
6030are replaced with corresponding bits from <tt>%repl</tt>. That is the 0th bit
6031in <tt>%repl</tt> replaces the <tt>%lo</tt>th bit in <tt>%val</tt> and etc. up
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6032to the <tt>%hi</tt>th bit.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6033
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6034<p>In reverse mode, a similar computation is made except that the bits are
6035reversed. That is, the <tt>0</tt>th bit in <tt>%repl</tt> replaces the
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6036<tt>%hi</tt> bit in <tt>%val</tt> and etc. down to the <tt>%lo</tt>th bit.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6037
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6038<h5>Examples:</h5>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6039
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6040<pre>
6041 llvm.part.set(0xFFFF, 0, 4, 7) -&gt; 0xFF0F
6042 llvm.part.set(0xFFFF, 0, 7, 4) -&gt; 0xFF0F
6043 llvm.part.set(0xFFFF, 1, 7, 4) -&gt; 0xFF8F
6044 llvm.part.set(0xFFFF, F, 8, 3) -&gt; 0xFFE7
6045 llvm.part.set(0xFFFF, 0, 3, 8) -&gt; 0xFE07
6046</pre>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6047
6048</div>
6049
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6050<!-- ======================================================================= -->
6051<div class="doc_subsection">
6052 <a name="int_overflow">Arithmetic with Overflow Intrinsics</a>
6053</div>
6054
6055<div class="doc_text">
6056<p>
6057LLVM provides intrinsics for some arithmetic with overflow operations.
6058</p>
6059
6060</div>
6061
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6062<!-- _______________________________________________________________________ -->
6063<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6064 <a name="int_sadd_overflow">'<tt>llvm.sadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6065</div>
6066
6067<div class="doc_text">
6068
6069<h5>Syntax:</h5>
6070
6071<p>This is an overloaded intrinsic. You can use <tt>llvm.sadd.with.overflow</tt>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6072on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6073
6074<pre>
6075 declare {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b)
6076 declare {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6077 declare {i64, i1} @llvm.sadd.with.overflow.i64(i64 %a, i64 %b)
6078</pre>
6079
6080<h5>Overview:</h5>
6081
6082<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6083a signed addition of the two arguments, and indicate whether an overflow
6084occurred during the signed summation.</p>
6085
6086<h5>Arguments:</h5>
6087
6088<p>The arguments (%a and %b) and the first element of the result structure may
6089be of integer types of any bit width, but they must have the same bit width. The
6090second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6091and <tt>%b</tt> are the two values that will undergo signed addition.</p>
6092
6093<h5>Semantics:</h5>
6094
6095<p>The '<tt>llvm.sadd.with.overflow</tt>' family of intrinsic functions perform
6096a signed addition of the two variables. They return a structure &mdash; the
6097first element of which is the signed summation, and the second element of which
6098is a bit specifying if the signed summation resulted in an overflow.</p>
6099
6100<h5>Examples:</h5>
6101<pre>
6102 %res = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %a, i32 %b)
6103 %sum = extractvalue {i32, i1} %res, 0
6104 %obit = extractvalue {i32, i1} %res, 1
6105 br i1 %obit, label %overflow, label %normal
6106</pre>
6107
6108</div>
6109
6110<!-- _______________________________________________________________________ -->
6111<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6112 <a name="int_uadd_overflow">'<tt>llvm.uadd.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6113</div>
6114
6115<div class="doc_text">
6116
6117<h5>Syntax:</h5>
6118
6119<p>This is an overloaded intrinsic. You can use <tt>llvm.uadd.with.overflow</tt>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6120on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6121
6122<pre>
6123 declare {i16, i1} @llvm.uadd.with.overflow.i16(i16 %a, i16 %b)
6124 declare {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6125 declare {i64, i1} @llvm.uadd.with.overflow.i64(i64 %a, i64 %b)
6126</pre>
6127
6128<h5>Overview:</h5>
6129
6130<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6131an unsigned addition of the two arguments, and indicate whether a carry occurred
6132during the unsigned summation.</p>
6133
6134<h5>Arguments:</h5>
6135
6136<p>The arguments (%a and %b) and the first element of the result structure may
6137be of integer types of any bit width, but they must have the same bit width. The
6138second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6139and <tt>%b</tt> are the two values that will undergo unsigned addition.</p>
6140
6141<h5>Semantics:</h5>
6142
6143<p>The '<tt>llvm.uadd.with.overflow</tt>' family of intrinsic functions perform
6144an unsigned addition of the two arguments. They return a structure &mdash; the
6145first element of which is the sum, and the second element of which is a bit
6146specifying if the unsigned summation resulted in a carry.</p>
6147
6148<h5>Examples:</h5>
6149<pre>
6150 %res = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %a, i32 %b)
6151 %sum = extractvalue {i32, i1} %res, 0
6152 %obit = extractvalue {i32, i1} %res, 1
6153 br i1 %obit, label %carry, label %normal
6154</pre>
6155
6156</div>
6157
6158<!-- _______________________________________________________________________ -->
6159<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6160 <a name="int_ssub_overflow">'<tt>llvm.ssub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6161</div>
6162
6163<div class="doc_text">
6164
6165<h5>Syntax:</h5>
6166
6167<p>This is an overloaded intrinsic. You can use <tt>llvm.ssub.with.overflow</tt>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6168on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6169
6170<pre>
6171 declare {i16, i1} @llvm.ssub.with.overflow.i16(i16 %a, i16 %b)
6172 declare {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6173 declare {i64, i1} @llvm.ssub.with.overflow.i64(i64 %a, i64 %b)
6174</pre>
6175
6176<h5>Overview:</h5>
6177
6178<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6179a signed subtraction of the two arguments, and indicate whether an overflow
6180occurred during the signed subtraction.</p>
6181
6182<h5>Arguments:</h5>
6183
6184<p>The arguments (%a and %b) and the first element of the result structure may
6185be of integer types of any bit width, but they must have the same bit width. The
6186second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6187and <tt>%b</tt> are the two values that will undergo signed subtraction.</p>
6188
6189<h5>Semantics:</h5>
6190
6191<p>The '<tt>llvm.ssub.with.overflow</tt>' family of intrinsic functions perform
6192a signed subtraction of the two arguments. They return a structure &mdash; the
6193first element of which is the subtraction, and the second element of which is a bit
6194specifying if the signed subtraction resulted in an overflow.</p>
6195
6196<h5>Examples:</h5>
6197<pre>
6198 %res = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %a, i32 %b)
6199 %sum = extractvalue {i32, i1} %res, 0
6200 %obit = extractvalue {i32, i1} %res, 1
6201 br i1 %obit, label %overflow, label %normal
6202</pre>
6203
6204</div>
6205
6206<!-- _______________________________________________________________________ -->
6207<div class="doc_subsubsection">
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6208 <a name="int_usub_overflow">'<tt>llvm.usub.with.overflow.*</tt>' Intrinsics</a>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6209</div>
6210
6211<div class="doc_text">
6212
6213<h5>Syntax:</h5>
6214
6215<p>This is an overloaded intrinsic. You can use <tt>llvm.usub.with.overflow</tt>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6216on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6217
6218<pre>
6219 declare {i16, i1} @llvm.usub.with.overflow.i16(i16 %a, i16 %b)
6220 declare {i32, i1} @llvm.usub.with.overflow.i32(i32 %a, i32 %b)
6221 declare {i64, i1} @llvm.usub.with.overflow.i64(i64 %a, i64 %b)
6222</pre>
6223
6224<h5>Overview:</h5>
6225
6226<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6227an unsigned subtraction of the two arguments, and indicate whether an overflow
6228occurred during the unsigned subtraction.</p>
6229
6230<h5>Arguments:</h5>
6231
6232<p>The arguments (%a and %b) and the first element of the result structure may
6233be of integer types of any bit width, but they must have the same bit width. The
6234second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6235and <tt>%b</tt> are the two values that will undergo unsigned subtraction.</p>
6236
6237<h5>Semantics:</h5>
6238
6239<p>The '<tt>llvm.usub.with.overflow</tt>' family of intrinsic functions perform
6240an unsigned subtraction of the two arguments. They return a structure &mdash; the
6241first element of which is the subtraction, and the second element of which is a bit
6242specifying if the unsigned subtraction resulted in an overflow.</p>
6243
6244<h5>Examples:</h5>
6245<pre>
6246 %res = call {i32, i1} @llvm.usub.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 %overflow, 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_smul_overflow">'<tt>llvm.smul.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>
6262
6263<p>This is an overloaded intrinsic. You can use <tt>llvm.smul.with.overflow</tt>
Bill Wendling3e1258b2009-02-08 04:04:40 +0000[diff] [blame]6264on any integer bit width.</p>
Bill Wendling3f8cebe2009-02-08 01:40:31 +0000[diff] [blame]6265
6266<pre>
6267 declare {i16, i1} @llvm.smul.with.overflow.i16(i16 %a, i16 %b)
6268 declare {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6269 declare {i64, i1} @llvm.smul.with.overflow.i64(i64 %a, i64 %b)
6270</pre>
6271
6272<h5>Overview:</h5>
6273
6274<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6275a signed multiplication of the two arguments, and indicate whether an overflow
6276occurred during the signed multiplication.</p>
6277
6278<h5>Arguments:</h5>
6279
6280<p>The arguments (%a and %b) and the first element of the result structure may
6281be of integer types of any bit width, but they must have the same bit width. The
6282second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6283and <tt>%b</tt> are the two values that will undergo signed multiplication.</p>
6284
6285<h5>Semantics:</h5>
6286
6287<p>The '<tt>llvm.smul.with.overflow</tt>' family of intrinsic functions perform
6288a signed multiplication of the two arguments. They return a structure &mdash;
6289the first element of which is the multiplication, and the second element of
6290which is a bit specifying if the signed multiplication resulted in an
6291overflow.</p>
6292
6293<h5>Examples:</h5>
6294<pre>
6295 %res = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %a, i32 %b)
6296 %sum = extractvalue {i32, i1} %res, 0
6297 %obit = extractvalue {i32, i1} %res, 1
6298 br i1 %obit, label %overflow, label %normal
6299</pre>
6300
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6301</div>
6302
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6303<!-- _______________________________________________________________________ -->
6304<div class="doc_subsubsection">
6305 <a name="int_umul_overflow">'<tt>llvm.umul.with.overflow.*</tt>' Intrinsics</a>
6306</div>
6307
6308<div class="doc_text">
6309
6310<h5>Syntax:</h5>
6311
6312<p>This is an overloaded intrinsic. You can use <tt>llvm.umul.with.overflow</tt>
6313on any integer bit width.</p>
6314
6315<pre>
6316 declare {i16, i1} @llvm.umul.with.overflow.i16(i16 %a, i16 %b)
6317 declare {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6318 declare {i64, i1} @llvm.umul.with.overflow.i64(i64 %a, i64 %b)
6319</pre>
6320
6321<h5>Overview:</h5>
6322
Bill Wendlingbda98b62009-02-08 23:00:09 +0000[diff] [blame]6323<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6324a unsigned multiplication of the two arguments, and indicate whether an overflow
6325occurred during the unsigned multiplication.</p>
6326
6327<h5>Arguments:</h5>
6328
6329<p>The arguments (%a and %b) and the first element of the result structure may
6330be of integer types of any bit width, but they must have the same bit width. The
6331second element of the result structure must be of type <tt>i1</tt>. <tt>%a</tt>
6332and <tt>%b</tt> are the two values that will undergo unsigned
6333multiplication.</p>
6334
6335<h5>Semantics:</h5>
6336
6337<p>The '<tt>llvm.umul.with.overflow</tt>' family of intrinsic functions perform
6338an unsigned multiplication of the two arguments. They return a structure &mdash;
6339the first element of which is the multiplication, and the second element of
6340which is a bit specifying if the unsigned multiplication resulted in an
6341overflow.</p>
6342
6343<h5>Examples:</h5>
6344<pre>
6345 %res = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %a, i32 %b)
6346 %sum = extractvalue {i32, i1} %res, 0
6347 %obit = extractvalue {i32, i1} %res, 1
6348 br i1 %obit, label %overflow, label %normal
6349</pre>
6350
6351</div>
6352
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6353<!-- ======================================================================= -->
6354<div class="doc_subsection">
6355 <a name="int_debugger">Debugger Intrinsics</a>
6356</div>
6357
6358<div class="doc_text">
6359<p>
6360The LLVM debugger intrinsics (which all start with <tt>llvm.dbg.</tt> prefix),
6361are described in the <a
6362href="SourceLevelDebugging.html#format_common_intrinsics">LLVM Source Level
6363Debugging</a> document.
6364</p>
6365</div>
6366
6367
6368<!-- ======================================================================= -->
6369<div class="doc_subsection">
6370 <a name="int_eh">Exception Handling Intrinsics</a>
6371</div>
6372
6373<div class="doc_text">
6374<p> The LLVM exception handling intrinsics (which all start with
6375<tt>llvm.eh.</tt> prefix), are described in the <a
6376href="ExceptionHandling.html#format_common_intrinsics">LLVM Exception
6377Handling</a> document. </p>
6378</div>
6379
6380<!-- ======================================================================= -->
6381<div class="doc_subsection">
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6382 <a name="int_trampoline">Trampoline Intrinsic</a>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6383</div>
6384
6385<div class="doc_text">
6386<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6387 This intrinsic makes it possible to excise one parameter, marked with
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6388 the <tt>nest</tt> attribute, from a function. The result is a callable
6389 function pointer lacking the nest parameter - the caller does not need
6390 to provide a value for it. Instead, the value to use is stored in
6391 advance in a "trampoline", a block of memory usually allocated
6392 on the stack, which also contains code to splice the nest value into the
6393 argument list. This is used to implement the GCC nested function address
6394 extension.
6395</p>
6396<p>
6397 For example, if the function is
6398 <tt>i32 f(i8* nest %c, i32 %x, i32 %y)</tt> then the resulting function
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]6399 pointer has signature <tt>i32 (i32, i32)*</tt>. It can be created as follows:</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6400<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6401 %tramp = alloca [10 x i8], align 4 ; size and alignment only correct for X86
6402 %tramp1 = getelementptr [10 x i8]* %tramp, i32 0, i32 0
6403 %p = call i8* @llvm.init.trampoline( i8* %tramp1, i8* bitcast (i32 (i8* nest , i32, i32)* @f to i8*), i8* %nval )
6404 %fp = bitcast i8* %p to i32 (i32, i32)*
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6405</pre>
Bill Wendlinge262b4c2007-09-22 09:23:55 +0000[diff] [blame]6406 <p>The call <tt>%val = call i32 %fp( i32 %x, i32 %y )</tt> is then equivalent
6407 to <tt>%val = call i32 %f( i8* %nval, i32 %x, i32 %y )</tt>.</p>
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6408</div>
6409
6410<!-- _______________________________________________________________________ -->
6411<div class="doc_subsubsection">
6412 <a name="int_it">'<tt>llvm.init.trampoline</tt>' Intrinsic</a>
6413</div>
6414<div class="doc_text">
6415<h5>Syntax:</h5>
6416<pre>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6417declare 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]6418</pre>
6419<h5>Overview:</h5>
6420<p>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6421 This fills the memory pointed to by <tt>tramp</tt> with code
6422 and returns a function pointer suitable for executing it.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6423</p>
6424<h5>Arguments:</h5>
6425<p>
6426 The <tt>llvm.init.trampoline</tt> intrinsic takes three arguments, all
6427 pointers. The <tt>tramp</tt> argument must point to a sufficiently large
6428 and sufficiently aligned block of memory; this memory is written to by the
Duncan Sands35012212007-08-22 23:39:54 +0000[diff] [blame]6429 intrinsic. Note that the size and the alignment are target-specific - LLVM
6430 currently provides no portable way of determining them, so a front-end that
6431 generates this intrinsic needs to have some target-specific knowledge.
6432 The <tt>func</tt> argument must hold a function bitcast to an <tt>i8*</tt>.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6433</p>
6434<h5>Semantics:</h5>
6435<p>
6436 The block of memory pointed to by <tt>tramp</tt> is filled with target
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6437 dependent code, turning it into a function. A pointer to this function is
6438 returned, but needs to be bitcast to an
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6439 <a href="#int_trampoline">appropriate function pointer type</a>
Duncan Sands7407a9f2007-09-11 14:10:23 +0000[diff] [blame]6440 before being called. The new function's signature is the same as that of
6441 <tt>func</tt> with any arguments marked with the <tt>nest</tt> attribute
6442 removed. At most one such <tt>nest</tt> argument is allowed, and it must be
6443 of pointer type. Calling the new function is equivalent to calling
6444 <tt>func</tt> with the same argument list, but with <tt>nval</tt> used for the
6445 missing <tt>nest</tt> argument. If, after calling
6446 <tt>llvm.init.trampoline</tt>, the memory pointed to by <tt>tramp</tt> is
6447 modified, then the effect of any later call to the returned function pointer is
6448 undefined.
Duncan Sands38947cd2007-07-27 12:58:54 +0000[diff] [blame]6449</p>
6450</div>
6451
6452<!-- ======================================================================= -->
6453<div class="doc_subsection">
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6454 <a name="int_atomics">Atomic Operations and Synchronization Intrinsics</a>
6455</div>
6456
6457<div class="doc_text">
6458<p>
6459 These intrinsic functions expand the "universal IR" of LLVM to represent
6460 hardware constructs for atomic operations and memory synchronization. This
6461 provides an interface to the hardware, not an interface to the programmer. It
Chris Lattner96451482008-08-05 18:29:16 +0000[diff] [blame]6462 is aimed at a low enough level to allow any programming models or APIs
6463 (Application Programming Interfaces) which
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6464 need atomic behaviors to map cleanly onto it. It is also modeled primarily on
6465 hardware behavior. Just as hardware provides a "universal IR" for source
6466 languages, it also provides a starting point for developing a "universal"
6467 atomic operation and synchronization IR.
6468</p>
6469<p>
6470 These do <em>not</em> form an API such as high-level threading libraries,
6471 software transaction memory systems, atomic primitives, and intrinsic
6472 functions as found in BSD, GNU libc, atomic_ops, APR, and other system and
6473 application libraries. The hardware interface provided by LLVM should allow
6474 a clean implementation of all of these APIs and parallel programming models.
6475 No one model or paradigm should be selected above others unless the hardware
6476 itself ubiquitously does so.
6477
6478</p>
6479</div>
6480
6481<!-- _______________________________________________________________________ -->
6482<div class="doc_subsubsection">
6483 <a name="int_memory_barrier">'<tt>llvm.memory.barrier</tt>' Intrinsic</a>
6484</div>
6485<div class="doc_text">
6486<h5>Syntax:</h5>
6487<pre>
6488declare void @llvm.memory.barrier( i1 &lt;ll&gt;, i1 &lt;ls&gt;, i1 &lt;sl&gt;, i1 &lt;ss&gt;,
6489i1 &lt;device&gt; )
6490
6491</pre>
6492<h5>Overview:</h5>
6493<p>
6494 The <tt>llvm.memory.barrier</tt> intrinsic guarantees ordering between
6495 specific pairs of memory access types.
6496</p>
6497<h5>Arguments:</h5>
6498<p>
6499 The <tt>llvm.memory.barrier</tt> intrinsic requires five boolean arguments.
6500 The first four arguments enables a specific barrier as listed below. The fith
6501 argument specifies that the barrier applies to io or device or uncached memory.
6502
6503</p>
6504 <ul>
6505 <li><tt>ll</tt>: load-load barrier</li>
6506 <li><tt>ls</tt>: load-store barrier</li>
6507 <li><tt>sl</tt>: store-load barrier</li>
6508 <li><tt>ss</tt>: store-store barrier</li>
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]6509 <li><tt>device</tt>: barrier applies to device and uncached memory also.</li>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6510 </ul>
6511<h5>Semantics:</h5>
6512<p>
6513 This intrinsic causes the system to enforce some ordering constraints upon
6514 the loads and stores of the program. This barrier does not indicate
6515 <em>when</em> any events will occur, it only enforces an <em>order</em> in
6516 which they occur. For any of the specified pairs of load and store operations
6517 (f.ex. load-load, or store-load), all of the first operations preceding the
6518 barrier will complete before any of the second operations succeeding the
6519 barrier begin. Specifically the semantics for each pairing is as follows:
6520</p>
6521 <ul>
6522 <li><tt>ll</tt>: All loads before the barrier must complete before any load
6523 after the barrier begins.</li>
6524
6525 <li><tt>ls</tt>: All loads before the barrier must complete before any
6526 store after the barrier begins.</li>
6527 <li><tt>ss</tt>: All stores before the barrier must complete before any
6528 store after the barrier begins.</li>
6529 <li><tt>sl</tt>: All stores before the barrier must complete before any
6530 load after the barrier begins.</li>
6531 </ul>
6532<p>
6533 These semantics are applied with a logical "and" behavior when more than one
6534 is enabled in a single memory barrier intrinsic.
6535</p>
6536<p>
6537 Backends may implement stronger barriers than those requested when they do not
6538 support as fine grained a barrier as requested. Some architectures do not
6539 need all types of barriers and on such architectures, these become noops.
6540</p>
6541<h5>Example:</h5>
6542<pre>
6543%ptr = malloc i32
6544 store i32 4, %ptr
6545
6546%result1 = load i32* %ptr <i>; yields {i32}:result1 = 4</i>
6547 call void @llvm.memory.barrier( i1 false, i1 true, i1 false, i1 false )
6548 <i>; guarantee the above finishes</i>
6549 store i32 8, %ptr <i>; before this begins</i>
6550</pre>
6551</div>
6552
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6553<!-- _______________________________________________________________________ -->
6554<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6555 <a name="int_atomic_cmp_swap">'<tt>llvm.atomic.cmp.swap.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6556</div>
6557<div class="doc_text">
6558<h5>Syntax:</h5>
6559<p>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6560 This is an overloaded intrinsic. You can use <tt>llvm.atomic.cmp.swap</tt> on
6561 any integer bit width and for different address spaces. Not all targets
6562 support all bit widths however.</p>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6563
6564<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6565declare i8 @llvm.atomic.cmp.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;cmp&gt;, i8 &lt;val&gt; )
6566declare i16 @llvm.atomic.cmp.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;cmp&gt;, i16 &lt;val&gt; )
6567declare i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;cmp&gt;, i32 &lt;val&gt; )
6568declare 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]6569
6570</pre>
6571<h5>Overview:</h5>
6572<p>
6573 This loads a value in memory and compares it to a given value. If they are
6574 equal, it stores a new value into the memory.
6575</p>
6576<h5>Arguments:</h5>
6577<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6578 The <tt>llvm.atomic.cmp.swap</tt> intrinsic takes three arguments. The result as
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6579 well as both <tt>cmp</tt> and <tt>val</tt> must be integer values with the
6580 same bit width. The <tt>ptr</tt> argument must be a pointer to a value of
6581 this integer type. While any bit width integer may be used, targets may only
6582 lower representations they support in hardware.
6583
6584</p>
6585<h5>Semantics:</h5>
6586<p>
6587 This entire intrinsic must be executed atomically. It first loads the value
6588 in memory pointed to by <tt>ptr</tt> and compares it with the value
6589 <tt>cmp</tt>. If they are equal, <tt>val</tt> is stored into the memory. The
6590 loaded value is yielded in all cases. This provides the equivalent of an
6591 atomic compare-and-swap operation within the SSA framework.
6592</p>
6593<h5>Examples:</h5>
6594
6595<pre>
6596%ptr = malloc i32
6597 store i32 4, %ptr
6598
6599%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6600%result1 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 4, %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6601 <i>; yields {i32}:result1 = 4</i>
6602%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6603%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6604
6605%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6606%result2 = call i32 @llvm.atomic.cmp.swap.i32.p0i32( i32* %ptr, i32 5, %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6607 <i>; yields {i32}:result2 = 8</i>
6608%stored2 = icmp eq i32 %result2, 5 <i>; yields {i1}:stored2 = false</i>
6609
6610%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 8</i>
6611</pre>
6612</div>
6613
6614<!-- _______________________________________________________________________ -->
6615<div class="doc_subsubsection">
6616 <a name="int_atomic_swap">'<tt>llvm.atomic.swap.*</tt>' Intrinsic</a>
6617</div>
6618<div class="doc_text">
6619<h5>Syntax:</h5>
6620
6621<p>
6622 This is an overloaded intrinsic. You can use <tt>llvm.atomic.swap</tt> on any
6623 integer bit width. Not all targets support all bit widths however.</p>
6624<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6625declare i8 @llvm.atomic.swap.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;val&gt; )
6626declare i16 @llvm.atomic.swap.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;val&gt; )
6627declare i32 @llvm.atomic.swap.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;val&gt; )
6628declare i64 @llvm.atomic.swap.i64.p0i64( i64* &lt;ptr&gt;, i64 &lt;val&gt; )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6629
6630</pre>
6631<h5>Overview:</h5>
6632<p>
6633 This intrinsic loads the value stored in memory at <tt>ptr</tt> and yields
6634 the value from memory. It then stores the value in <tt>val</tt> in the memory
6635 at <tt>ptr</tt>.
6636</p>
6637<h5>Arguments:</h5>
6638
6639<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6640 The <tt>llvm.atomic.swap</tt> intrinsic takes two arguments. Both the
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6641 <tt>val</tt> argument and the result must be integers of the same bit width.
6642 The first argument, <tt>ptr</tt>, must be a pointer to a value of this
6643 integer type. The targets may only lower integer representations they
6644 support.
6645</p>
6646<h5>Semantics:</h5>
6647<p>
6648 This intrinsic loads the value pointed to by <tt>ptr</tt>, yields it, and
6649 stores <tt>val</tt> back into <tt>ptr</tt> atomically. This provides the
6650 equivalent of an atomic swap operation within the SSA framework.
6651
6652</p>
6653<h5>Examples:</h5>
6654<pre>
6655%ptr = malloc i32
6656 store i32 4, %ptr
6657
6658%val1 = add i32 4, 4
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6659%result1 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val1 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6660 <i>; yields {i32}:result1 = 4</i>
6661%stored1 = icmp eq i32 %result1, 4 <i>; yields {i1}:stored1 = true</i>
6662%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 8</i>
6663
6664%val2 = add i32 1, 1
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6665%result2 = call i32 @llvm.atomic.swap.i32.p0i32( i32* %ptr, i32 %val2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6666 <i>; yields {i32}:result2 = 8</i>
6667
6668%stored2 = icmp eq i32 %result2, 8 <i>; yields {i1}:stored2 = true</i>
6669%memval2 = load i32* %ptr <i>; yields {i32}:memval2 = 2</i>
6670</pre>
6671</div>
6672
6673<!-- _______________________________________________________________________ -->
6674<div class="doc_subsubsection">
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6675 <a name="int_atomic_load_add">'<tt>llvm.atomic.load.add.*</tt>' Intrinsic</a>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6676
6677</div>
6678<div class="doc_text">
6679<h5>Syntax:</h5>
6680<p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6681 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.add</tt> on any
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6682 integer bit width. Not all targets support all bit widths however.</p>
6683<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6684declare i8 @llvm.atomic.load.add.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6685declare i16 @llvm.atomic.load.add.i16..p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6686declare i32 @llvm.atomic.load.add.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6687declare 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]6688
6689</pre>
6690<h5>Overview:</h5>
6691<p>
6692 This intrinsic adds <tt>delta</tt> to the value stored in memory at
6693 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6694</p>
6695<h5>Arguments:</h5>
6696<p>
6697
6698 The intrinsic takes two arguments, the first a pointer to an integer value
6699 and the second an integer value. The result is also an integer value. These
6700 integer types can have any bit width, but they must all have the same bit
6701 width. The targets may only lower integer representations they support.
6702</p>
6703<h5>Semantics:</h5>
6704<p>
6705 This intrinsic does a series of operations atomically. It first loads the
6706 value stored at <tt>ptr</tt>. It then adds <tt>delta</tt>, stores the result
6707 to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6708</p>
6709
6710<h5>Examples:</h5>
6711<pre>
6712%ptr = malloc i32
6713 store i32 4, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6714%result1 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 4 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6715 <i>; yields {i32}:result1 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6716%result2 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 2 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6717 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6718%result3 = call i32 @llvm.atomic.load.add.i32.p0i32( i32* %ptr, i32 5 )
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6719 <i>; yields {i32}:result3 = 10</i>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6720%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 15</i>
Andrew Lenharthe44f3902008-02-21 06:45:13 +0000[diff] [blame]6721</pre>
6722</div>
6723
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6724<!-- _______________________________________________________________________ -->
6725<div class="doc_subsubsection">
6726 <a name="int_atomic_load_sub">'<tt>llvm.atomic.load.sub.*</tt>' Intrinsic</a>
6727
6728</div>
6729<div class="doc_text">
6730<h5>Syntax:</h5>
6731<p>
6732 This is an overloaded intrinsic. You can use <tt>llvm.atomic.load.sub</tt> on
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6733 any integer bit width and for different address spaces. Not all targets
6734 support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6735<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6736declare i8 @llvm.atomic.load.sub.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6737declare i16 @llvm.atomic.load.sub.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6738declare i32 @llvm.atomic.load.sub.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6739declare 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]6740
6741</pre>
6742<h5>Overview:</h5>
6743<p>
6744 This intrinsic subtracts <tt>delta</tt> to the value stored in memory at
6745 <tt>ptr</tt>. It yields the original value at <tt>ptr</tt>.
6746</p>
6747<h5>Arguments:</h5>
6748<p>
6749
6750 The intrinsic takes two arguments, the first a pointer to an integer value
6751 and the second an integer value. The result is also an integer value. These
6752 integer types can have any bit width, but they must all have the same bit
6753 width. The targets may only lower integer representations they support.
6754</p>
6755<h5>Semantics:</h5>
6756<p>
6757 This intrinsic does a series of operations atomically. It first loads the
6758 value stored at <tt>ptr</tt>. It then subtracts <tt>delta</tt>, stores the
6759 result to <tt>ptr</tt>. It yields the original value stored at <tt>ptr</tt>.
6760</p>
6761
6762<h5>Examples:</h5>
6763<pre>
6764%ptr = malloc i32
6765 store i32 8, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6766%result1 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 4 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6767 <i>; yields {i32}:result1 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6768%result2 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6769 <i>; yields {i32}:result2 = 4</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6770%result3 = call i32 @llvm.atomic.load.sub.i32.p0i32( i32* %ptr, i32 5 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6771 <i>; yields {i32}:result3 = 2</i>
6772%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = -3</i>
6773</pre>
6774</div>
6775
6776<!-- _______________________________________________________________________ -->
6777<div class="doc_subsubsection">
6778 <a name="int_atomic_load_and">'<tt>llvm.atomic.load.and.*</tt>' Intrinsic</a><br>
6779 <a name="int_atomic_load_nand">'<tt>llvm.atomic.load.nand.*</tt>' Intrinsic</a><br>
6780 <a name="int_atomic_load_or">'<tt>llvm.atomic.load.or.*</tt>' Intrinsic</a><br>
6781 <a name="int_atomic_load_xor">'<tt>llvm.atomic.load.xor.*</tt>' Intrinsic</a><br>
6782
6783</div>
6784<div class="doc_text">
6785<h5>Syntax:</h5>
6786<p>
6787 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_and</tt>,
6788 <tt>llvm.atomic.load_nand</tt>, <tt>llvm.atomic.load_or</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6789 <tt>llvm.atomic.load_xor</tt> on any integer bit width and for different
6790 address spaces. Not all targets support all bit widths however.</p>
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6791<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6792declare i8 @llvm.atomic.load.and.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6793declare i16 @llvm.atomic.load.and.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6794declare i32 @llvm.atomic.load.and.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6795declare 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]6796
6797</pre>
6798
6799<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6800declare i8 @llvm.atomic.load.or.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6801declare i16 @llvm.atomic.load.or.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6802declare i32 @llvm.atomic.load.or.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6803declare 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]6804
6805</pre>
6806
6807<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6808declare i8 @llvm.atomic.load.nand.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6809declare i16 @llvm.atomic.load.nand.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6810declare i32 @llvm.atomic.load.nand.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6811declare 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]6812
6813</pre>
6814
6815<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6816declare i8 @llvm.atomic.load.xor.i8.p0i32( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6817declare i16 @llvm.atomic.load.xor.i16.p0i32( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6818declare i32 @llvm.atomic.load.xor.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6819declare 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]6820
6821</pre>
6822<h5>Overview:</h5>
6823<p>
6824 These intrinsics bitwise the operation (and, nand, or, xor) <tt>delta</tt> to
6825 the value stored in memory at <tt>ptr</tt>. It yields the original value
6826 at <tt>ptr</tt>.
6827</p>
6828<h5>Arguments:</h5>
6829<p>
6830
6831 These intrinsics take two arguments, the first a pointer to an integer value
6832 and the second an integer value. The result is also an integer value. These
6833 integer types can have any bit width, but they must all have the same bit
6834 width. The targets may only lower integer representations they support.
6835</p>
6836<h5>Semantics:</h5>
6837<p>
6838 These intrinsics does a series of operations atomically. They first load the
6839 value stored at <tt>ptr</tt>. They then do the bitwise operation
6840 <tt>delta</tt>, store the result to <tt>ptr</tt>. They yield the original
6841 value stored at <tt>ptr</tt>.
6842</p>
6843
6844<h5>Examples:</h5>
6845<pre>
6846%ptr = malloc i32
6847 store i32 0x0F0F, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6848%result0 = call i32 @llvm.atomic.load.nand.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6849 <i>; yields {i32}:result0 = 0x0F0F</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6850%result1 = call i32 @llvm.atomic.load.and.i32.p0i32( i32* %ptr, i32 0xFF )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6851 <i>; yields {i32}:result1 = 0xFFFFFFF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6852%result2 = call i32 @llvm.atomic.load.or.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6853 <i>; yields {i32}:result2 = 0xF0</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6854%result3 = call i32 @llvm.atomic.load.xor.i32.p0i32( i32* %ptr, i32 0F )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6855 <i>; yields {i32}:result3 = FF</i>
6856%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = F0</i>
6857</pre>
6858</div>
6859
6860
6861<!-- _______________________________________________________________________ -->
6862<div class="doc_subsubsection">
6863 <a name="int_atomic_load_max">'<tt>llvm.atomic.load.max.*</tt>' Intrinsic</a><br>
6864 <a name="int_atomic_load_min">'<tt>llvm.atomic.load.min.*</tt>' Intrinsic</a><br>
6865 <a name="int_atomic_load_umax">'<tt>llvm.atomic.load.umax.*</tt>' Intrinsic</a><br>
6866 <a name="int_atomic_load_umin">'<tt>llvm.atomic.load.umin.*</tt>' Intrinsic</a><br>
6867
6868</div>
6869<div class="doc_text">
6870<h5>Syntax:</h5>
6871<p>
6872 These are overloaded intrinsics. You can use <tt>llvm.atomic.load_max</tt>,
6873 <tt>llvm.atomic.load_min</tt>, <tt>llvm.atomic.load_umax</tt>, and
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6874 <tt>llvm.atomic.load_umin</tt> on any integer bit width and for different
6875 address spaces. Not all targets
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6876 support all bit widths however.</p>
6877<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6878declare i8 @llvm.atomic.load.max.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6879declare i16 @llvm.atomic.load.max.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6880declare i32 @llvm.atomic.load.max.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6881declare 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]6882
6883</pre>
6884
6885<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6886declare i8 @llvm.atomic.load.min.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6887declare i16 @llvm.atomic.load.min.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6888declare i32 @llvm.atomic.load.min.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6889declare 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]6890
6891</pre>
6892
6893<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6894declare i8 @llvm.atomic.load.umax.i8.p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6895declare i16 @llvm.atomic.load.umax.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6896declare i32 @llvm.atomic.load.umax.i32.p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6897declare 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]6898
6899</pre>
6900
6901<pre>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6902declare i8 @llvm.atomic.load.umin.i8..p0i8( i8* &lt;ptr&gt;, i8 &lt;delta&gt; )
6903declare i16 @llvm.atomic.load.umin.i16.p0i16( i16* &lt;ptr&gt;, i16 &lt;delta&gt; )
6904declare i32 @llvm.atomic.load.umin.i32..p0i32( i32* &lt;ptr&gt;, i32 &lt;delta&gt; )
6905declare 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]6906
6907</pre>
6908<h5>Overview:</h5>
6909<p>
6910 These intrinsics takes the signed or unsigned minimum or maximum of
6911 <tt>delta</tt> and the value stored in memory at <tt>ptr</tt>. It yields the
6912 original value at <tt>ptr</tt>.
6913</p>
6914<h5>Arguments:</h5>
6915<p>
6916
6917 These intrinsics take two arguments, the first a pointer to an integer value
6918 and the second an integer value. The result is also an integer value. These
6919 integer types can have any bit width, but they must all have the same bit
6920 width. The targets may only lower integer representations they support.
6921</p>
6922<h5>Semantics:</h5>
6923<p>
6924 These intrinsics does a series of operations atomically. They first load the
6925 value stored at <tt>ptr</tt>. They then do the signed or unsigned min or max
6926 <tt>delta</tt> and the value, store the result to <tt>ptr</tt>. They yield
6927 the original value stored at <tt>ptr</tt>.
6928</p>
6929
6930<h5>Examples:</h5>
6931<pre>
6932%ptr = malloc i32
6933 store i32 7, %ptr
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6934%result0 = call i32 @llvm.atomic.load.min.i32.p0i32( i32* %ptr, i32 -2 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6935 <i>; yields {i32}:result0 = 7</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6936%result1 = call i32 @llvm.atomic.load.max.i32.p0i32( i32* %ptr, i32 8 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6937 <i>; yields {i32}:result1 = -2</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6938%result2 = call i32 @llvm.atomic.load.umin.i32.p0i32( i32* %ptr, i32 10 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6939 <i>; yields {i32}:result2 = 8</i>
Mon P Wangce3ac892008-07-30 04:36:53 +0000[diff] [blame]6940%result3 = call i32 @llvm.atomic.load.umax.i32.p0i32( i32* %ptr, i32 30 )
Mon P Wang6bde9ec2008-06-25 08:15:39 +0000[diff] [blame]6941 <i>; yields {i32}:result3 = 8</i>
6942%memval1 = load i32* %ptr <i>; yields {i32}:memval1 = 30</i>
6943</pre>
6944</div>
Andrew Lenharth785610d2008-02-16 01:24:58 +0000[diff] [blame]6945
6946<!-- ======================================================================= -->
6947<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6948 <a name="int_general">General Intrinsics</a>
6949</div>
6950
6951<div class="doc_text">
6952<p> This class of intrinsics is designed to be generic and has
6953no specific purpose. </p>
6954</div>
6955
6956<!-- _______________________________________________________________________ -->
6957<div class="doc_subsubsection">
6958 <a name="int_var_annotation">'<tt>llvm.var.annotation</tt>' Intrinsic</a>
6959</div>
6960
6961<div class="doc_text">
6962
6963<h5>Syntax:</h5>
6964<pre>
6965 declare void @llvm.var.annotation(i8* &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
6966</pre>
6967
6968<h5>Overview:</h5>
6969
6970<p>
6971The '<tt>llvm.var.annotation</tt>' intrinsic
6972</p>
6973
6974<h5>Arguments:</h5>
6975
6976<p>
6977The first argument is a pointer to a value, the second is a pointer to a
6978global string, the third is a pointer to a global string which is the source
6979file name, and the last argument is the line number.
6980</p>
6981
6982<h5>Semantics:</h5>
6983
6984<p>
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6985This intrinsic allows annotation of local variables with arbitrary strings.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6986This can be useful for special purpose optimizations that want to look for these
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]6987annotations. These have no other defined use, they are ignored by code
6988generation and optimization.
6989</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]6990</div>
6991
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6992<!-- _______________________________________________________________________ -->
6993<div class="doc_subsubsection">
Tanya Lattnerc9869b12007-09-21 23:57:59 +0000[diff] [blame]6994 <a name="int_annotation">'<tt>llvm.annotation.*</tt>' Intrinsic</a>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]6995</div>
6996
6997<div class="doc_text">
6998
6999<h5>Syntax:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]7000<p>This is an overloaded intrinsic. You can use '<tt>llvm.annotation</tt>' on
7001any integer bit width.
7002</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7003<pre>
Tanya Lattner09161fe2007-09-22 00:03:01 +0000[diff] [blame]7004 declare i8 @llvm.annotation.i8(i8 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7005 declare i16 @llvm.annotation.i16(i16 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7006 declare i32 @llvm.annotation.i32(i32 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7007 declare i64 @llvm.annotation.i64(i64 &lt;val&gt;, i8* &lt;str&gt;, i8* &lt;str&gt;, i32 &lt;int&gt; )
7008 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]7009</pre>
7010
7011<h5>Overview:</h5>
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]7012
7013<p>
7014The '<tt>llvm.annotation</tt>' intrinsic.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7015</p>
7016
7017<h5>Arguments:</h5>
7018
7019<p>
7020The first argument is an integer value (result of some expression),
7021the second is a pointer to a global string, the third is a pointer to a global
7022string which is the source file name, and the last argument is the line number.
Tanya Lattnere545be72007-09-21 23:56:27 +0000[diff] [blame]7023It returns the value of the first argument.
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7024</p>
7025
7026<h5>Semantics:</h5>
7027
7028<p>
7029This intrinsic allows annotations to be put on arbitrary expressions
7030with arbitrary strings. This can be useful for special purpose optimizations
7031that want to look for these annotations. These have no other defined use, they
7032are ignored by code generation and optimization.
Dan Gohman2672f3e2008-10-14 16:51:45 +0000[diff] [blame]7033</p>
Tanya Lattnerb306a9e2007-09-21 22:59:12 +0000[diff] [blame]7034</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7035
Anton Korobeynikove6e764f2008-01-15 22:31:34 +0000[diff] [blame]7036<!-- _______________________________________________________________________ -->
7037<div class="doc_subsubsection">
7038 <a name="int_trap">'<tt>llvm.trap</tt>' Intrinsic</a>
7039</div>
7040
7041<div class="doc_text">
7042
7043<h5>Syntax:</h5>
7044<pre>
7045 declare void @llvm.trap()
7046</pre>
7047
7048<h5>Overview:</h5>
7049
7050<p>
7051The '<tt>llvm.trap</tt>' intrinsic
7052</p>
7053
7054<h5>Arguments:</h5>
7055
7056<p>
7057None
7058</p>
7059
7060<h5>Semantics:</h5>
7061
7062<p>
7063This intrinsics is lowered to the target dependent trap instruction. If the
7064target does not have a trap instruction, this intrinsic will be lowered to the
7065call of the abort() function.
7066</p>
7067</div>
7068
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7069<!-- _______________________________________________________________________ -->
7070<div class="doc_subsubsection">
Misha Brukman5dd7f4d2008-11-22 23:55:29 +0000[diff] [blame]7071 <a name="int_stackprotector">'<tt>llvm.stackprotector</tt>' Intrinsic</a>
Bill Wendlinge4164592008-11-19 05:56:17 +0000[diff] [blame]7072</div>
7073<div class="doc_text">
7074<h5>Syntax:</h5>
7075<pre>
7076declare void @llvm.stackprotector( i8* &lt;guard&gt;, i8** &lt;slot&gt; )
7077
7078</pre>
7079<h5>Overview:</h5>
7080<p>
7081 The <tt>llvm.stackprotector</tt> intrinsic takes the <tt>guard</tt> and stores
7082 it onto the stack at <tt>slot</tt>. The stack slot is adjusted to ensure that
7083 it is placed on the stack before local variables.
7084</p>
7085<h5>Arguments:</h5>
7086<p>
7087 The <tt>llvm.stackprotector</tt> intrinsic requires two pointer arguments. The
7088 first argument is the value loaded from the stack guard
7089 <tt>@__stack_chk_guard</tt>. The second variable is an <tt>alloca</tt> that
7090 has enough space to hold the value of the guard.
7091</p>
7092<h5>Semantics:</h5>
7093<p>
7094 This intrinsic causes the prologue/epilogue inserter to force the position of
7095 the <tt>AllocaInst</tt> stack slot to be before local variables on the
7096 stack. This is to ensure that if a local variable on the stack is overwritten,
7097 it will destroy the value of the guard. When the function exits, the guard on
7098 the stack is checked against the original guard. If they're different, then
7099 the program aborts by calling the <tt>__stack_chk_fail()</tt> function.
7100</p>
7101</div>
7102
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7103<!-- *********************************************************************** -->
7104<hr>
7105<address>
7106 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7107 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7108 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]7109 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]7110
7111 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
7112 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
7113 Last modified: $Date$
7114</address>
Chris Lattner08497ce2008-01-04 04:33:49 +0000[diff] [blame]7115
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7116</body>
7117</html>